This PDF file contains the front matter associated with SPIE Proceedings Volume 9788, including the Title Page, Copyright information, Table of Contents, and Conference Committee listing.

There is a need for novel imaging techniques for the earlier detection of Alzheimer's disease (AD). Two hallmarks of AD are amyloid beta (Aβ) plaques and tau tangles that are formed in the brain. Well-characterized x-ray cross sections of Aβ and tau proteins in a variety of structural states could potentially be used as AD biomarkers for small-angle x-ray scattering (SAXS) imaging without the need for injectable probes or contrast agents. First, however, the protein structures must be controlled and measured to determine accurate biomarkers for SAXS imaging. Here we report SAXS measurements of Aβ₄₂ and tau₃₅₂ in a 50% dimethyl sulfoxide (DMSO) solution in which these proteins are believed to remain monomeric because of the stabilizing interaction of DMSO solution. Our SAXS analysis showed the aggregation of both proteins. In particular, we found that the aggregation of Aβ₄₂ slowly progresses with time in comparison to tau₃₅₂ that aggregates at a faster rate and reaches a steady-state. Furthermore, the measured signals were compared to the theoretical SAXS profiles of Aβ₄₂ monomer, Aβ₄₂ fibril, and tau₃₅₂ that were computed from their respective protein data bank structures. We have begun the work to systematically control the structural states of these proteins in vitro using various solvent conditions. Our future work is to utilize the distinct SAXS profiles of various structural states of Aβ and tau to build a library of signals of interest for SAXS imaging in brain tissue.

This project investigated the signal thresholding effectiveness at reducing the instrument noise of an electron multiplying charged coupled device (EMCCD) based micro-CT system at low x-ray exposure levels. Scans of a mouse spine and an iodine phantom were taken using an EMCCD detector coupled with a micro-CT system. An iodine filter of 4 mg/cm² area density was placed in the beam. The output signal was thresholded using some multiple of the inherent background noise. For each threshold, 100, 200, and 300 frames were summed for each projection to evaluate the effect on the reconstructed image. The projection images from the scans were compared using line profiles and their SNR. Our results indicate that, as the threshold was increased, the line profiles of the projection images showed less statistical variation, but also lower signal levels, so that the SNR of the projection images decreased as the threshold increased. When the line profile of a projection image obtained using a signal threshold is compared with one obtained using energy integrating mode, the profile obtained using thresholding had less variation than that obtained using energy integration, which indicates less instrument noise. The SNR at the edges of the scan object is higher in the thresholded images when compared with the energy integrated projection images. We conclude that thresholding the output signal from an EMCCD detector at low x-ray exposure levels is an effective method to reduce the instrument noise of an EMCCD detector.

Dynamic magnetic resonance imaging (DS-MRI) has been recognized as a promising method for visualizing articulatory motion of speech in scientific research and clinical applications. However, characterization of the gestural and acoustical properties of the vocal tract remains a challenging task for DS-MRI because it requires: 1) reconstructing high-quality spatiotemporal images by incorporating stronger prior knowledge; and 2) quantitatively interpreting the reconstructed images that contain great motion variability. This work presents a novel imaging method that simultaneously meets both requirements by integrating a spatiotemporal atlas into a Partial Separability (PS) model-based imaging framework. Through the use of an atlas-driven sparsity constraint, this method is capable of capturing high-quality articulatory dynamics at an imaging speed of 102 frames per second and a spatial resolution of 2.2 × 2.2 mm2. Moreover, the proposed method enables quantitative characterization of variability of speech motion, compared to the generic motion pattern across all subjects, through the spatial residual components.

This work describes a methodology for efficient removal of scatter radiation during digital breast tomosynthesis (DBT). The goal of this approach is to enable grid image obscuration without a large increase in radiation dose by minimizing misalignment of the grid focal point (GFP) and x-ray focal spot (XFS) during grid reciprocation. Hardware for the motion scheme was built and tested on the dual modality breast tomosynthesis (DMT) scanner, which combines DBT and molecular breast tomosynthesis (MBT) on a single gantry. The DMT scanner uses fully isocentric rotation of tube and x-ray detector for maintaining a fixed tube-detector alignment during DBT imaging. A cellular focused copper prototype grid with 80 cm focal length, 3.85 mm height, 0.1 mm thick lamellae, and 1.1 mm hole pitch was tested. Primary transmission of the grid at 28 kV tube voltage was on average 74% with the grid stationary and aligned for maximum transmission. It fell to 72% during grid reciprocation by the proposed method. Residual grid line artifacts (GLAs) in projection views and reconstructed DBT images are characterized and methods for reducing the visibility of GLAs in the reconstructed volume through projection image flat-field correction and spatial frequency-based filtering of the DBT slices are described and evaluated. The software correction methods reduce the visibility of these artifacts in the reconstructed volume, making them imperceptible both in the reconstructed DBT images and their Fourier transforms.

Purpose: To explore, by means of analytical and Monte Carlo modeling, performance of a novel lightweight and low-cost wearable helmet-shaped Brain PET (BET) camera based on thin-film digital Geiger Avalanche Photo Diode (dGAPD) with LSO and LaBr₃ scintillators for imaging in vivo human brain processes for freely moving and acting subjects responding to various stimuli in any environment.

Methods: We performed analytical and Monte Carlo modeling PET performance of a spherical cap BET device and cylindrical brain PET (CYL) device, both with 25 cm diameter and the same total mass of LSO scintillator. Total mass of LSO in both the BET and CYL systems is about 32 kg for a 25 mm thick scintillator, and 13 kg for 10 mm thick scintillator (assuming an LSO density of 7.3 g/ml). We also investigated a similar system using an LaBr3 scintillator corresponding to 22 kg and 9 kg for the 25 mm and 10 mm thick systems (assuming an LaBr₃ density of 5.08 g/ml). In addition, we considered a clinical whole body (WB) LSO PET/CT scanner with 82 cm ring diameter and 15.8 cm axial length to represent a reference system. BET consisted of distributed Autonomous Detector Arrays (ADAs) integrated into Intelligent Autonomous Detector Blocks (IADBs). The ADA comprised of an array of small LYSO scintillator volumes (voxels with base a×a: 1.0 ≤ a ≤ 2.0 mm and length c: 3.0 ≤ c ≤ 6.0 mm) with 5–65 μm thick reflective layers on its five sides and sixth side optically coupled to the matching array of dGAPDs and processing electronics with total thickness of 50 μm. Simulated energy resolution was 10.8% and 3.3% for LSO and LaBr₃ respectively and the coincidence window was set at 2 ns. The brain was simulated as a sphere of uniform F-18 activity with diameter of 10 cm embedded in a center of water sphere with diameter of 10 cm.

Results: Analytical and Monte Carlo models showed similar results for lower energy window values (458 keV versus 445 keV for LSO, and 492 keV versus 485 keV for LaBr₃), and for the relative performance of system sensitivity. Monte Carlo results further showed that the BET geometry had >50% better noise equivalent count (NEC) performance relative to the CYL geometry, and >1100% better performance than a WB geometry for 25 mm thick LSO and LaBr₃. For 10 mm thick LaBr₃ equivalent mass systems LSO (7 mm thick) performed ~40% higher NEC than LaBr₃. Analytic and Monte Carlo simulations also showed that 1×1×3 mm scintillator crystals can achieve ~1.2 mm FWHM spatial resolution.

Conclusions: This study shows that a spherical cap brain PET system can provide improved NEC while preserving spatial resolution when compared to an equivalent dedicated cylindrical PET brain camera and shows greatly improved PET performance relative to a conventional whole body PET/CT. In addition, our simulations show that LSO will generally outperform LaBr₃ for NEC unless the timing resolution for LaBr₃ is considerably smaller than presently used for LSO, i.e. well below 300 ps.

Choroid neovascularization (CNV) is a kind of pathology from the choroid and CNV-related disease is one important cause of vision loss. It is desirable to predict the CNV growth rate so that appropriate treatment can be planned. In this paper, we seek to find a method to predict the growth of CNV based on 3D longitudinal Optical Coherence Tomography (OCT) images. A reaction-diffusion model is proposed for prediction. The method consists of four phases: pre-processing, meshing, CNV growth modeling and prediction. We not only apply the reaction-diffusion model to the disease region, but also take the surrounding tissues into consideration including outer retinal layer, inner retinal layer and choroid layer. The diffusion in these tissues is considered as isotropic. The finite-element-method (FEM) is used to solve the partial differential equations (PDE) in the diffusion model. The curve of CNV growth with treatment are fitted and then we can predict the CNV status in a future time point. The preliminary results demonstrated that our proposed method is accurate and the validity and feasibility of our model is obvious.

Barrett's esophagus (BE) is a precancerous complication of gastroesophageal reflux disease in which normal stratified squamous epithelium lining the esophagus is replaced by intestinal metaplastic columnar epithelium. Repeated endoscopies and multiple biopsies are often necessary to establish the presence of intestinal metaplasia. Narrow Band Imaging (NBI) is an imaging technique commonly used with endoscopies that enhances the contrast of vascular pattern on the mucosa. We present a computer-based method for the automatic normal/metaplastic classification of endoscopic NBI images. Superpixel segmentation is used to identify and cluster pixels belonging to uniform regions. From each uniform clustered region of pixels, eight features maximizing differences among normal and metaplastic epithelium are extracted for the classification step. For each superpixel, the three mean intensities of each color channel are firstly selected as features. Three added features are the mean intensities for each superpixel after separately applying to the red-channel image three different morphological filters (top-hat filtering, entropy filtering and range filtering). The last two features require the computation of the Grey-Level Co-Occurrence Matrix (GLCM), and are reflective of the contrast and the homogeneity of each superpixel. The classification step is performed using an ensemble of 50 classification trees, with a 10-fold cross-validation scheme by training the classifier at each step on a random 70% of the images and testing on the remaining 30% of the dataset. Sensitivity and Specificity are respectively of 79.2% and 87.3%, with an overall accuracy of 83.9%.

Celiac disease (CD) is an immune-mediated enteropathy triggered by exposure to gluten and similar proteins, affecting genetically susceptible persons, increasing their risk of different complications. Small bowels mucosa damage due to CD involves various degrees of endoscopically relevant lesions, which are not easily recognized: their overall sensitivity and positive predictive values are poor even when zoom-endoscopy is used. Confocal Laser Endomicroscopy (CLE) allows skilled and trained experts to qualitative evaluate mucosa alteration such as a decrease in goblet cells density, presence of villous atrophy or crypt hypertrophy. We present a method for automatically classifying CLE images into three different classes: normal regions, villous atrophy and crypt hypertrophy. This classification is performed after a features selection process, in which four features are extracted from each image, through the application of homomorphic filtering and border identification through Canny and Sobel operators. Three different classifiers have been tested on a dataset of 67 different images labeled by experts in three classes (normal, VA and CH): linear approach, Naive-Bayes quadratic approach and a standard quadratic analysis, all validated with a ten-fold cross validation. Linear classification achieves 82.09% accuracy (class accuracies: 90.32% for normal villi, 82.35% for VA and 68.42% for CH, sensitivity: 0.68, specificity 1.00), Naive Bayes analysis returns 83.58% accuracy (90.32% for normal villi, 70.59% for VA and 84.21% for CH, sensitivity: 0.84 specificity: 0.92), while the quadratic analysis achieves a final accuracy of 94.03% (96.77% accuracy for normal villi, 94.12% for VA and 89.47% for CH, sensitivity: 0.89, specificity: 0.98).

Calcium imaging is a widely used technique in neuroscience permitting the simultaneous monitoring of electro- physiological activity of hundreds of neurons at single cell resolution. Identification of neuronal activity requires rapid and reliable image analysis techniques, especially when neurons fire and move simultaneously over time. Traditionally, image segmentation is performed to extract individual neurons in the first frame of a calcium sequence. Thereafter, the mean intensity is calculated from the same region of interest in each frame to infer calcium signals. However, when cells move, deform and fire, this segmentation on its own generates artefacts and therefore biased neuronal activity. Therefore, there is a pressing need to develop a more efficient cell tracking technique. We hereby present a novel vision-based cell tracking scheme using a thin-plate spline deformable model. The thin-plate spline warping is based on control points detected using the Fast from Accelerated Segment Test descriptor and tracked using the Lucas-Kanade optical flow. Our method is able to track neurons in calcium time-series, even when there are large changes in intensity, such as during a firing event. The robustness and efficiency of the proposed approach is validated on real calcium time-lapse images of a neuronal population.

Due to the digital X-ray imaging system's limited field of view, several individual sector images are required to capture the posture of an individual in standing position. These images are then “stitched together” to reconstruct the standing posture. We have created an image processing application that automates the stitching, therefore minimizing user input, optimizing workflow, and reducing human error. The application begins with pre-processing the input images by removing artifacts, filtering out isolated noisy regions, and amplifying a seamless bone edge. The resulting binary images are then registered together using a rigid-body intensity based registration algorithm. The identified registration transformations are then used to map the original sector images into the panorama image. Our method focuses primarily on the use of the anatomical content of the images to generate the panoramas as opposed to using external markers employed to aid with the alignment process. Currently, results show robust edge detection prior to registration and we have tested our approach by comparing the resulting automatically-stitched panoramas to the manually stitched panoramas in terms of registration parameters, target registration error of homologous markers, and the homogeneity of the digitally subtracted automatically- and manually-stitched images using 26 patient datasets.

The paper presents an automatic segmentation methodology for the patellar bone, based on 3D gradient recalled echo and gradient recalled echo with fat suppression magnetic resonance images. Constricted search space outlines are incorporated into recursive ray-tracing to segment the outer cortical bone. A statistical analysis based on the dependence of information in adjacent slices is used to limit the search in each image to between an outer and inner search region. A section based recursive ray-tracing mechanism is used to skip inner noise regions and detect the edge boundary. The proposed method achieves higher segmentation accuracy (0.23mm) than the current state-of-the-art methods with the average dice similarity coefficient of 96.0% (SD 1.3%) agreement between the auto-segmentation and ground truth surfaces.

Nowadays many people are affected by arthritis, a condition of the joints with limited prevention measures, but with various options of treatment the most radical of which is surgical. In order for surgery to be successful, it can make use of careful analysis of patient-based models generated from medical images, usually by manual segmentation. In this work we show how to automate the segmentation of a crucial and complex joint — the knee. To achieve this goal we rely on our novel way of representing a 3D voxel volume as a hierarchical structure of partitions which we have named Image Partition Forest (IPF). The IPF contains several partition layers of increasing coarseness, with partitions nested across layers in the form of adjacency graphs. On the basis of a set of properties (size, mean intensity, coordinates) of each node in the IPF we classify nodes into different features. Values indicating whether or not any particular node belongs to the femur or tibia are assigned through node filtering and node-based region growing. So far we have evaluated our method on 15 MRI knee images. Our unsupervised segmentation compared against a hand-segmented gold standard has achieved an average Dice similarity coefficient of 0.95 for femur and 0.93 for tibia, and an average symmetric surface distance of 0.98 mm for femur and 0.73 mm for tibia. The paper also discusses ways to introduce stricter morphological and spatial conditioning in the bone labelling process.

In this paper we propose to use the Wavelet Leader (WL) transformation for studying trabecular bone patterns. Given an input image, its WL transformation is defined as the cross-channel-layer maximum pooling of an underlying wavelet transformation. WL inherits the advantage of the original wavelet transformation in capturing spatial-frequency statistics of texture images, while being more robust against scale and orientation thanks to the maximum pooling strategy. These properties make WL an attractive alternative to replace wavelet transformations which are used for trabecular analysis in previous studies. In particular, in this paper, after extracting wavelet leader descriptors from a trabecular texture patch, we feed them into two existing statistic texture characterization methods, namely the Gray Level Co-occurrence Matrix (GLCM) and the Gray Level Run Length Matrix (GLRLM). The most discriminative features, Energy of GLCM and Gray Level Non-Uniformity of GLRLM, are retained to distinguish two different populations between osteoporotic patients and control subjects. Receiver Operating Characteristics (ROC) curves are used to measure performance of classification. Experimental results on a recently released benchmark dataset show that WL significantly boosts the performance of baseline wavelet transformations by 5% in average.

The reconstruction of a patient-specific 3D anatomy is the crucial step in the computer-aided preoperative planning based on plain X-ray images. In this paper, we propose a robust and fast reconstruction methods based on fitting the statistical shape and intensity model of a femoral bone onto a pair of calibrated X-ray images. We formulate the registration as a non-linear least squares problem, allowing for the involvement of Levenberg-Marquardt optimisation. The proposed methods have been tested on a set of 96 virtual X-ray images. The reconstruction accuracy was evaluated using the symmetric Hausdorff distance between reconstructed and ground-truth bones. The accuracy of the intensity-based method reached 1.18 ± 1.57mm on average, the registration took 8.76 seconds on average.

Many studies using T1 magnetic resonance imaging (MRI) data have found associations between changes in global metrics (e.g. volume) of brain structures and preterm birth. In this work, we use the surface displacement feature extracted from the deformations of the surface models of the third ventricle, fourth ventricle and brainstem to capture the variation in shape in these structures at 8 years of age that may be due to differences in the trajectory of brain development as a result of very preterm birth (24-32 weeks gestation). Understanding the spatial patterns of shape alterations in these structures in children who were born very preterm as compared to those who were born at full term may lead to better insights into mechanisms of differing brain development between these two groups. The T1 MRI data for the brain was acquired from children born full term (FT, n=14, 8 males) and preterm (PT, n=51, 22 males) at age 8-years. Accurate segmentation labels for these structures were obtained via a multi-template fusion based segmentation method. A high dimensional non-rigid registration algorithm was utilized to register the target segmentation labels to a set of segmentation labels defined on an average-template. The surface displacement data for the brainstem and the third ventricle were found to be significantly different (p < 0.05) between the PT and FT groups. Further, spatially localized clusters with inward and outward deformation were found to be associated with lower gestational age. The results from this study present a shape analysis method for pediatric MRI data and reveal shape changes that may be due to preterm birth.

Both normal aging and neurodegenerative diseases such as Alzheimer’s disease cause morphological changes of the brain. To better distinguish between normal and abnormal cases, it is necessary to model changes in brain morphology owing to normal aging. To this end, we developed a method for analyzing and visualizing these changes for the entire brain morphology distribution in the general aging population. The method is applied to 1000 subjects from a large population imaging study in the elderly, from which 900 were used to train the model and 100 were used for testing. The results of the 100 test subjects show that the model generalizes to subjects outside the model population. Smooth percentile curves showing the brain morphology changes as a function of age and spatiotemporal atlases derived from the model population are publicly available via an interactive web application at agingbrain.bigr.nl.

We investigate the applicability of a computational framework, called mutual connectivity analysis (MCA), for directed functional connectivity analysis in both synthetic and resting-state functional MRI data. This framework comprises of first evaluating non-linear cross-predictability between every pair of time series prior to recovering the underlying network structure using community detection algorithms. We obtain the non-linear cross-prediction score between time series using Generalized Radial Basis Functions (GRBF) neural networks. These cross-prediction scores characterize the underlying functionally connected networks within the resting brain, which can be extracted using non-metric clustering approaches, such as the Louvain method. We first test our approach on synthetic models with known directional influence and network structure. Our method is able to capture the directional relationships between time series (with an area under the ROC curve = 0.92 ± 0.037) as well as the underlying network structure (Rand index = 0.87 ± 0.063) with high accuracy. Furthermore, we test this method for network recovery on resting-state fMRI data, where results are compared to the motor cortex network recovered from a motor stimulation sequence, resulting in a strong agreement between the two (Dice coefficient = 0.45). We conclude that our MCA approach is effective in analyzing non-linear directed functional connectivity and in revealing underlying functional network structure in complex systems.

We demonstrate an approach to measure the information flow between each pair of time series in resting-state functional MRI (fMRI) data of the human brain and subsequently recover its underlying network structure. By integrating dimensionality reduction into predictive time series modeling, large-scale Granger Causality (lsGC) analysis method can reveal directed information flow suggestive of causal influence at an individual voxel level, unlike other multivariate approaches. This method quantifies the influence each voxel time series has on every other voxel time series in a multivariate sense and hence contains information about the underlying dynamics of the whole system, which can be used to reveal functionally connected networks within the brain. To identify such networks, we perform non-metric network clustering, such as accomplished by the Louvain method. We demonstrate the effectiveness of our approach to recover the motor and visual cortex from resting state human brain fMRI data and compare it with the network recovered from a visuomotor stimulation experiment, where the similarity is measured by the Dice Coefficient (DC). The best DC obtained was 0.59 implying a strong agreement between the two networks. In addition, we thoroughly study the effect of dimensionality reduction in lsGC analysis on network recovery. We conclude that our approach is capable of detecting causal influence between time series in a multivariate sense, which can be used to segment functionally connected networks in the resting-state fMRI.

Depression is prevalent among patients with Parkinson's disease (PD); however the pathophysiology of depression in PD is not well understood. In order to investigate how depression and motor impairments differentially and interactively affect specific brain regions in Parkinson's disease, we introduced a new data driven approach, namely Frequency Component Analysis (FCA), to decompose the resting-state functional magnetic resonance imaging data of 59 subjects with Parkinson's disease into different frequency bands. We then evaluated the main effects of motor severity and depression, and their interactive effects on the BOLD-fMRI signal oscillation energy in these specific frequency components. Our results show that the severity of motor symptoms is more negatively correlated with energy in the frequency band of 0.10-0.25Hz in the bilateral thalamus (THA), but more positively correlated with energy in the frequency band of 0.01-0.027Hz in the bilateral postcentral gyrus (PoCG). In contrast, the severity of depressive symptoms is more associated with the higher energy of the high frequency oscillations (>0.1Hz) but lower energy of 0.01-0.027Hz in the bilateral subgenual gyrus (SGC). Importantly, the interaction between motor and depressive symptoms is negatively correlated with the energy of high frequency oscillations (>0.1Hz) in the substantia nigra/ventral tegmental area (SN/VTA), left hippocampus (HIPP), left inferior orbital frontal cortex (OFC), and left temporoparietal junction (TPJ), but positively correlated with the energy of 0.02-0.05Hz in the left inferior OFC, left TPJ, left inferior temporal gyrus (ITG), and bilateral cerebellum. These results demonstrated that FCA was a promising method in interrogating the neurophysiological implications of different brain rhythms. Our findings further revealed the neural bases underlying the interactions as well the dissociations between motor and depressive symptoms in Parkinson's disease.

The use of functional Magnetic Resonance Imaging (fMRI) has provided interesting insights into our understanding of the brain. In clinical setups these scans have been used to detect and study changes in the brain network properties in various neurological disorders. A large percentage of subjects infected with HIV present cognitive deficits, which are known as HIV associated neurocognitive disorder (HAND). In this study we propose to use our novel technique named Mutual Connectivity Analysis (MCA) to detect differences in brain networks in subjects with and without HIV infection. Resting state functional MRI scans acquired from 10 subjects (5 HIV+ and 5 HIV-) were subject to standard preprocessing routines. Subsequently, the average time-series for each brain region of the Automated Anatomic Labeling (AAL) atlas are extracted and used with the MCA framework to obtain a graph characterizing the interactions between them. The network graphs obtained for different subjects are then compared using Network-Based Statistics (NBS), which is an approach to detect differences between graphs edges while controlling for the family-wise error rate when mass univariate testing is performed. Applying this approach on the graphs obtained yields a single network encompassing 42 nodes and 65 edges, which is significantly different between the two subject groups. Specifically connections to the regions in and around the basal ganglia are significantly decreased. Also some nodes corresponding to the posterior cingulate cortex are affected. These results are inline with our current understanding of pathophysiological mechanisms of HIV associated neurocognitive disease (HAND) and other HIV based fMRI connectivity studies. Hence, we illustrate the applicability of our novel approach with network-based statistics in a clinical case-control study to detect differences connectivity patterns.

The objective of this study is to investigate effects of methylphenidate on brain activity in individuals with cocaine use disorder (CUD) using functional MRI (fMRI). Methylphenidate hydrochloride (MPH) is an indirect dopamine agonist commonly used for treating attention deficit/hyperactivity disorders; it was also shown to have some positive effects on CUD subjects, such as improved stop signal reaction times associated with better control/inhibition,¹ as well as normalized task-related brain activity² and resting-state functional connectivity in specific areas.³ While prior fMRI studies of MPH in CUDs have focused on mass-univariate statistical hypothesis testing, this paper evaluates multivariate, whole-brain effects of MPH as captured by the generalization (prediction) accuracy of different classification techniques applied to features extracted from resting-state functional networks (e.g., node degrees). Our multivariate predictive results based on resting-state data from3 suggest that MPH tends to normalize network properties such as voxel degrees in CUD subjects, thus providing additional evidence for potential benefits of MPH in treating cocaine addiction.

Optical coherence tomography (OCT) of the human retina is now becoming established as an important modality for the detection and tracking of various ocular diseases. Voxel based morphometry (VBM) is a long standing neuroimaging analysis technique that allows for the exploration of the regional differences in the brain. There has been limited work done in developing registration based methods for OCT, which has hampered the advancement of VBM analyses in OCT based population studies. Following on from our recent development of an OCT registration method, we explore the potential benefits of VBM analysis in cohorts of healthy controls (HCs) and multiple sclerosis (MS) patients. Specifically, we validate the stability of VBM analysis in two pools of HCs showing no significant difference between the two populations. Additionally, we also present a retrospective study of age and sex matched HCs and relapsing remitting MS patients, demonstrating results consistent with the reported literature while providing insight into the retinal changes associated with this MS subtype.

X-ray luminescence computed tomography (XLCT) is a new hybrid imaging modality, which has the capability to improve optical spatial resolution to hundreds of micrometers for deep targets. In this paper, we report a multiple pinhole collimator based microscopic X-ray luminescence computed tomography (microXLCT) system for small animal imaging. Superfine collimated X-ray pencil beams are used to excite deeply embedded phosphor particles, allowing us to obtain sub-millimeter optical spatial resolution in deep tissues. Multiple collimated X-ray beams are generated by mounting an array of pinholes in the front of a powerful X-ray tube. With multiple X-ray beams scanning, the phosphor particles in the region of the multiple beams are excited simultaneously, which requires less scanning time compared with a single beam scanning. The emitted optical photons on the top surface of the phantom are measured with an electron multiplying charge-coupled device (EMCCD) camera. Meanwhile, an X-ray detector is used to determine the X-ray beam size and position, which are used as structural guidance in the microXLCT image reconstruction. To validate the performance of our proposed multiple pinhole based microXLCT imaging system, we have performed numerical simulations and a phantom experiment. In the numerical simulations, we simulated a cylindrical phantom with two and six embedded targets, respectively. In the simulations, we used four parallel X-ray beams with the beam diameter of 0.1 mm and the beam interval of 3.2 mm. We can reconstruct deeply embedded multiple targets with a target diameter of 0.2 mm using measurements in six projections, which indicated that four parallel X-ray beam scan could reduce scanning time without comprising the reconstructed image quality. In the phantom experiment, we generated two parallel X-ray beams with the beam diameter of 0.5 mm and the beam interval of 4.2 mm. We scanned a phantom of one target with the two parallel X-ray beams. The target was reconstructed successfully, which indicated that multiple collimated X-ray beam scan approach is feasible in small animal imaging.

We performed numerical simulations and phantom experiments with a conical mirror based fluorescence molecular tomography (FMT) imaging system to optimize its performance. With phantom experiments, we have compared three measurement modes in FMT: the whole surface measurement mode, the transmission mode, and the reflection mode. Our results indicated that the whole surface measurement mode performed the best. Then, we applied two different neutral density (ND) filters to improve the measurement's dynamic range. The benefits from ND filters are not as much as predicted. Finally, with numerical simulations, we have compared two laser excitation patterns: line and point. With the same excitation position number, we found that the line laser excitation had slightly better FMT reconstruction results than the point laser excitation. In the future, we will implement Monte Carlo ray tracing simulations to calculate multiple reflection photons, and create a look-up table accordingly for calibration.

Bioluminescence tomography (BLT) is becoming a promising tool because it can resolve the biodistribution of bioluminescent reporters associated with cellular and subcellular function through several millimeters with to centimeters of tissues in vivo. However, BLT reconstruction is an ill-posed problem. By incorporating sparse a priori information about bioluminescent source, enhanced image quality is obtained for sparsity based reconstruction algorithm. Therefore, sparsity based BLT reconstruction algorithm has a great potential. Here, we proposed a novel reconstruction method based on Bayesian compressive sensing and investigated its feasibility and effectiveness with a heterogeneous phantom. The results demonstrate the potential and merits of the proposed algorithm.

Purpose: The goal of this study was to achieve a real time 3D visualisation of the murine cardiovascular system by intravenously injected superparamagnetic nanoparticles using Magnetic particle imaging (MPI).

Material and Methods: MPI scans of FVB mice were performed using a 3D imaging sequence (1T/m gradient strength, 10mT drive-field strength). A dynamic scan with a temporal resolution of 21.5ms per 3D volume acquisition was performed. 50μl ferucarbotran (Resovist®, Bayer Healthcare AG) were injected into the tail vein after baseline MPI measurements. As MPI delivers no anatomic information, MRI scans at a 7T ClinScan (Bruker) were performed using a T2-weighted 2D TSE sequence. The reconstruction of the MPI data was performed on the MPI console (ParaVision 6.0/MPI, Bruker). Image fusion was done using additional image processing software (Imalytics, Philips). The dynamic information was extracted using custom software developed in the Julia programming environment.

Results: The combined MRI-MPI measurements were carried out successfully. MPI data clearly demonstrated the passage of the SPIO tracer through the inferior vena cava, the heart and finally the liver. By co-registration with MRI the anatomical regions were identified. Due to the volume frame rate of about 46 volumes per second a signal modulation with the frequency of the heart beat was detectable and a heart beat of 520 beats per minute (bpm) has been assumed. Moreover, the blood flow velocity of approximately 5cm/s in the vena cava has been estimated.

Conclusions: The high temporal resolution of MPI allows real-time imaging and bolus tracking of intravenous injected nanoparticles and offers a real time tool to assess blood flow velocity.

Gold nanoparticles (AuNPs) are emerging as promising agents for both cancer therapy and CT imaging. AuNPs are delivered to tumors via the enhanced permeability and retention effect and they preferentially accumulate in close proximity to the tumor blood vessels. AuNPs produce low-energy, short-range photoelectrons during external beam radiation therapy (RT), boosting dose. This work is focused on understanding how tumor vascular permeability is influenced by AuNP-augmented radiation therapy (RT), and how this knowledge can potentially improve the delivery of additional nanoparticle-based chemotherapeutics. We use dual energy (DE) CT to detect accumulation of AuNPs and increased vascular permeability to liposomal iodine (i.e. a surrogate for chemotherapeutics with liposome encapsulation) following RT. We used sarcoma tumors generated in LSL-Kras^G12D; p53^FL/FL conditional mutant mice. A total of n=37 mice were used in this study. The treated mice were injected with 20 mg AuNP (0.1 ml/25 g mouse) 24 hours before delivery of 5 Gy RT (n=5), 10 Gy RT (n=3) or 20 Gy RT (n=6). The control mice received no AuNP injection and either no RT (n=6), 5 Gy RT (n=3), 10 Gy RT (n=3), 20 Gy RT (n=11). Twenty four hours post-RT, the mice were injected with liposomal iodine (0.3 ml/25 mouse) and imaged with DE-CT three days later. The results suggest that independent of any AuNP usage, RT levels of 10 Gy and 20 Gy increase the permeability of tumor vasculature to liposomal iodine and that the increase in permeability is dose-dependent. We found that the effect of RT on vasculature may already be at its maximum response i.e. saturated at 20 Gy, and therefore the addition of AuNPs had almost no added benefit. Similarly, at 5 Gy RT, our data suggests that there was no effect of AuNP augmentation on tumor vascular permeability. However, by using AuNPs with 10 Gy RT, we observed an increase in the vascular permeability, however this is not yet statistically significant due to the small number of mice in these groups. Such an approach can be used together with a liposomal drug delivery system to increase specific tumor delivery of chemotherapeutics. Our method has the potential to significantly improve tumor therapy and reduce the side effects from both RT and chemotherapy.

The interaction between the left and right heart ventricles (LV and RV) depends on load and pressure conditions that are affected by cardiac contraction and respiration cycles. A novel MRI sequence, XD-GRASP, allows the acquisition of multi-dimensional, respiration-sorted and cardiac-synchronized free-breathing image data. In these data, effects of the cardiac and respiratory cycles on the LV/RV interaction can be observed independently. To enable the analysis of such data, we developed a semi-automatic exploration workflow. After tracking a cross-sectional line positioned over the heart, over all motion states, the septum and heart wall border locations are detected by analyzing the grey-value profile under the lines. These data are used to quantify septum motion, both in absolute units and as a fraction of the heart size, to compare values for different subjects. In addition to conventional visualization techniques, we used color maps for intuitive exploration of the variable values for this multi-dimensional data set. We acquired short-axis image data of nine healthy volunteers, to analyze the position and the motion of the interventricular septum in different breathing states and different cardiac cycle phases. The results indicate a consistent range of normal septum motion values, and also suggest that respiratory phase-dependent septum motion is greatest near end-diastolic phases. These new methods are a promising tool to assess LV/RV ventricle interaction and the effects of respiration on this interaction.

We present a semi-automated approach to comprehensively examine coronary remodeling over the entire length of intravascular ultrasound (IVUS) imaged vessels. Serial measurements at baseline and 12-month follow-up are analyzed rather than static data obtained at a single time point. Every IVUS pullback is segmented automatically, and then reviewed and algorithmically refined by an expert using a computer-aided just-enough-interaction approach. Subsequently, pairs of serial IVUS pullbacks are registered automatically using 3D graph optimization approach. Based on plaque volume increases or decreases over time, pullback frames are divided into two groups — progression and regression. It is shown that plaque progression rates are positively correlated with percent stenosis (PS) indices (p≪0.01) while plaque regression rates are negatively correlated with percent stenosis indices (p≪0.01). Moreover, for the progression group, adventitia area increases in direct relation with the baseline percent stenosis (p=0.007) when PS is less than 50%. Significance of such a correlation is not observed when percent stenosis exceeds 50%. Conversely, for the regression group, change of adventitia area is relatively constant for percent stenosis <50%; but decreases in direct relation with baseline stenosis (p≪0.01) when stenosis > 50%. This strongly suggests that lipid lowering treatment may effectively suppress plaque progression and accelerate plaque regression, especially for larger values of percent stenosis, and further accelerate the corresponding adventitia-remodeling process.

Coronary Artery Disease (CAD) is the leading cause of death globally [1]. Modern cardiac computed tomography angiography (CCTA) is highly effective at identifying and assessing coronary blockages associated with CAD. The diagnostic value of this anatomical information can be substantially increased in combination with a non-invasive, low-dose, correlative, quantitative measure of blood supply to the myocardium. While CT perfusion has shown promise of providing such indications of ischemia, artifacts due to motion, beam hardening, and other factors confound clinical findings and can limit quantitative accuracy. In this paper, we investigate the impact of applying a novel motion correction algorithm to correct for motion in the myocardium. This motion compensation algorithm (originally designed to correct for the motion of the coronary arteries in order to improve CCTA images) has been shown to provide substantial improvements in both overall image quality and diagnostic accuracy of CCTA. We have adapted this technique for application beyond the coronary arteries and present an assessment of its impact on image quality and quantitative accuracy within the context of dual-energy CT perfusion imaging. We conclude that motion correction is a promising technique that can help foster the routine clinical use of dual-energy CT perfusion. When combined, the anatomical information of CCTA and the hemodynamic information from dual-energy CT perfusion should facilitate better clinical decisions about which patients would benefit from treatments such as stent placement, drug therapy, or surgery and help other patients avoid the risks and costs associated with unnecessary, invasive, diagnostic coronary angiography procedures.

The pharmacokinetic parameters derived from dynamic contrast enhanced (DCE) MRI have shown promise as biomarkers for tumor response to therapy. However, standard methods of analyzing DCE MRI data (Tofts model) require high temporal resolution, high signal-to-noise ratio (SNR), and the Arterial Input Function (AIF). Such models produce reliable biomarkers of response only when a therapy has a large effect on the parameters. We recently reported a method that solves the limitations, the Linear Reference Region Model (LRRM). Similar to other reference region models, the LRRM needs no AIF. Additionally, the LRRM is more accurate and precise than standard methods at low SNR and slow temporal resolution, suggesting LRRM-derived biomarkers could be better predictors. Here, the LRRM, Non-linear Reference Region Model (NRRM), Linear Tofts model (LTM), and Non-linear Tofts Model (NLTM) were used to estimate the R^Ktrans between muscle and tumor (or the K^trans for Tofts) and the tumor k_ep,TOI for 39 breast cancer patients who received neoadjuvant chemotherapy (NAC). These parameters and the receptor statuses of each patient were used to construct cross-validated predictive models to classify patients as complete pathological responders (pCR) or non-complete pathological responders (non-pCR) to NAC. Model performance was evaluated using area under the ROC curve (AUC). The AUC for receptor status alone was 0.62, while the best performance using predictors from the LRRM, NRRM, LTM, and NLTM were AUCs of 0.79, 0.55, 0.60, and 0.59 respectively. This suggests that the LRRM can be used to predict response to NAC in breast cancer.

Hyperspectral imaging (HSI) is an emerging modality for medical applications and holds great potential for noninvasive early detection of cancer. It has been reported that early cancer detection can improve the survival and quality of life of head and neck cancer patients. In this paper, we explored the possibility of differentiating between premalignant lesions and healthy tongue tissue using hyperspectral imaging in a chemical induced oral cancer animal model. We proposed a novel classification algorithm for cancer detection using hyperspectral images. The method detected the dysplastic tissue with an average area under the curve (AUC) of 0.89. The hyperspectral imaging and classification technique may provide a new tool for oral cancer detection.

Hyperspectral imaging (HSI) is an emerging imaging modality for medical applications. HSI acquires two dimensional images at various wavelengths. The combination of both spectral and spatial information provides quantitative information for cancer detection and diagnosis. This paper proposes using superpixels, principal component analysis (PCA), and support vector machine (SVM) to distinguish regions of tumor from healthy tissue. The classification method uses 2 principal components decomposed from hyperspectral images and obtains an average sensitivity of 93% and an average specificity of 85% for 11 mice. The hyperspectral imaging technology and classification method can have various applications in cancer research and management.

Alteration in genetic expression is as important as gene mutation in cancer development and proliferation. Epigenetic changes affect gene expression without altering the DNA sequence. Histone deacetylase (HDAC), an enzyme facilitating histone remodelling, can lead to silencing of tumor suppressor genes making HDAC inhibitors viable anticancer drugs against tumors with increased activity of the enzyme. In this study we evaluated ¹⁸F-fluroacetamido-1-hexanoicanilide (¹⁸F-FAHA), an artificial HDAC substrate, as imaging probe of HDAC activity of human tumor xenografts in immunocompromised host mice. Human breast and melanoma cell lines, MDA-MB-468 and MDA-MB-435 respectively, known to overexpress HDAC activity were xenografted into immunocompromised mice and HDAC activity was imaged using ¹⁸F-FAHA. The melanoma group was treated with saline, SAHA (suberoylanilide hydroxamic acid, an approved anticancer HDAC inhibitor) in DMSO, or DMSO as positive control. Tracer kinetic modelling and SUV were used to estimate HDAC activity from dynamic PET data. Both breast tumor and melanoma group showed great variability in binding rate constant (BRC) of ¹⁸F-FAHA suggesting highly variable inter- and intra-tumoral HDAC activity. For the SAHA treated melanoma group, HDAC activity, as monitored by BRC of ¹⁸F-FAHA, decreased more than the two (positive and negative) control groups but not tumor growth. Our preliminary study showed that noninvasive PET imaging with ¹⁸F-FAHA has the potential to identify patients for whom treatment with HDAC inhibitors are appropriate, to assess the effectiveness of that treatment as an early marker of target reduction, and also eliminate the need for invasive tissue biopsy to individualize treatment.

We have developed a MRI method that can measure extracellular pH in tumor tissues, known as acidoCEST MRI. This method relies on the detection of Chemical Exchange Saturation Transfer (CEST) of iopamidol, an FDA-approved CT contrast agent that has two CEST signals. A log₁₀ ratio of the two CEST signals is linearly correlated with pH, but independent of agent concentration, endogenous T₁ relaxation time, and B₁ inhomogeneity. Therefore, detecting both CEST effects of iopamidol during in vivo studies can be used to accurately measure the extracellular pH in tumor tissues. Past in vivo studies using acidoCEST MRI have suffered from respiration artifacts in orthotopic and lung tumor models that have corrupted pH measurements. In addition, the non-linear fitting method used to analyze results is unreliable as it is subject to over-fitting especially with noisy CEST spectra. To improve the technique, we have recently developed a respiration gated CEST MRI pulse sequence that has greatly reduced motion artifacts, and we have included both a prescan and post scan to remove endogenous CEST effects. In addition, we fit the results by parameterizing the contrast of the exogenous agent with respect to pH via the Bloch equations modified for chemical exchange, which is less subject to over-fitting than the non-linear method. These advances in the acidoCEST MRI technique and analysis methods have made pH measurements more reliable, especially in areas of the body subject to respiratory motion.

Pulmonary vascular dysfunction has been implicated in smoking-related susceptibility to emphysema. With the growing interest in characterizing arterial morphology for early evaluation of the vascular role in pulmonary diseases, there is an increasing need for the standardization of a framework for arterial morphological assessment at airway segmental levels. In this paper, we present an effective and robust semi-automatic framework to segment pulmonary arteries at different anatomic airway branches and measure their cross-sectional area (CSA). The method starts with user-specified endpoints of a target arterial segment through a custom-built graphical user interface. It then automatically detect the centerline joining the endpoints, determines the local structure orientation and computes the CSA along the centerline after filtering out the adjacent pulmonary structures, such as veins or airway walls. Several new techniques are presented, including collision-impact based cost function for centerline detection, radial sample-line based CSA computation, and outlier analysis of radial distance to subtract adjacent neighboring structures in the CSA measurement. The method was applied to repeat-scan pulmonary multirow detector CT (MDCT) images from ten healthy subjects (age: 21-48 Yrs, mean: 28.5 Yrs; 7 female) at functional residual capacity (FRC). The reproducibility of computed arterial CSA from four airway segmental regions in middle and lower lobes was analyzed. The overall repeat-scan intra-class correlation (ICC) of the computed CSA from all four airway regions in ten subjects was 96% with maximum ICC found at LB10 and RB4 regions.

Chest fat estimation is important for identifying high-risk lung transplant candidates. In this paper, an approach to chest fat quantification based on a recently formulated concept of standardized anatomic space (SAS) is presented. The goal of this paper is to seek answers to the following questions related to chest fat quantification on single slice versus whole volume CT, which have not been addressed in the literature. What level of correlation exists between total chest fat volume and fat areas measured on single abdominal and thigh slices? What is the anatomic location in the chest where maximal correlation of fat area with fat volume can be expected? Do the components of subcutaneous adipose tissue (SAT) and visceral adipose tissue (VAT) have the same area-to-volume correlative behavior or do they differ? The SAS approach includes two steps: calibration followed by transformation which will map the patient slice locations non-linearly to SAS. The optimal slice locations found for SAT and VAT based on SAS are different and at the mid-level of the T8 vertebral body for SAT and mid-level of the T7 vertebral body for VAT. Fat volume and area on optimal slices for SAT and VAT are correlated with Pearson correlation coefficients of 0.97 and 0.86, respectively. The correlation of chest fat volume with abdominal and thigh fat areas is weak to modest.

Patient-specific simulation of air flows in lungs is now straightforward using segmented airways trees from CT scans as the basis for Computational Fluid Dynamics (CFD) simulations. These models generally use static geometries, which do not account for the motion of the lungs and its influence on important clinical indicators, such as airway resistance. This paper is concerned with the simulation of tidal breathing, including the dynamic motion of the lungs, and the required analysis workflow. Geometries are based on CT scans obtained at the extremes of the breathing cycle, Total Lung Capacity (TLC) and Functional Residual Capacity (FRC). It describes how topologically consistent geometries are obtained at TLC and FRC, using a ‘skeleton’ of the network of airway branches. From this a 3D computational mesh which morphs between TLC and FRC is generated. CFD results for a number of patient-specific cases, healthy and asthmatic, are presented. Finally their potential use in evaluation of the progress of the disease is discussed, focusing on an important clinical indicator, the airway resistance.

Research results from the National Lung Screening Trial revealed that screening for lung cancer with low-dose CT (LDCT) reduces lung cancer mortality in heavy smokers by 20% compared to radiography. While this study does show the efficacy of CT-based screening, radiologists often face the problem of estimating the malignant likelihoods of pulmonary nodules detected on LDCT screening for maximizing patient survival and for preserving lung function. Spiculation is considered as one of the indicators of nodule malignancy and an important feature to assess requirements on a patient-tailored follow-up procedure. However, the spiculation is also observed in some benign nodules, particularly in tuberculoma. The elucidation of the spliculation morphology in 3D thoracic CT images is an important preliminary step towards developing the malignant discrimination strategies from benign nodules. In this study, we present a visualization method to reveal a spatial configuration of spiculation of pulmonary nodules in three-dimensional thoracic CT images. Applying the method to an example of malignant nodule with the spiculated margins, the visualizing preliminary result of the spatial configuration reveals the presence of membrane structures of spiculation.

Correct segmentation and labeling of lungs in thorax MSCT is a requirement in pulmonary/respiratory disease analysis as a basis for further processing or direct quantitative measures: lung texture classification, respiratory functional simulations, intrapulmonary vascular remodeling evaluation, detection of pleural effusion or subpleural opacities, are only few clinical applications related to this requirement. Whereas lung segmentation appears trivial for normal anatomo-pathological conditions, the presence of disease may complicate this task for fully-automated algorithms. The challenges come either from regional changes of lung texture opacity or from complex anatomic configurations (e.g., thin septum between lungs making difficult proper lung separation). They make difficult or even impossible the use of classic algorithms based on adaptive thresholding, 3-D connected component analysis and shape regularization. The objective of this work is to provide a robust segmentation approach of the pulmonary field, with individualized labeling of the lungs, able to overcome the mentioned limitations. The proposed approach relies on 3-D mathematical morphology and exploits the concept of controlled relief flooding (to identify contrasted lung areas) together with patient-specific shape properties for peripheral dense tissue detection. Tested on a database of 40 MSCT of pathological lungs, the proposed approach showed correct identification of lung areas with high sensitivity and specificity in locating peripheral dense opacities.

From early childhood to adulthood, synaptogenesis and synaptic pruning continuously reshape the structural architecture and neural connection in developmental human brains. Disturbance of the precisely balanced strengthening of certain axons and pruning of others may cause mental disorders such as autism and schizophrenia. To characterize this balance, we proposed a novel measurement based on cortical parcellation and diffusion MRI (dMRI) tractography, a cortical connectivity maturation index (CCMI). To evaluate the spatiotemporal sensitivity of CCMI as a potential biomarker, dMRI and T₁ weighted datasets of 21 healthy subjects 2-25 years were acquired. Brain cortex was parcellated into 68 gyral labels using T₁ weighted images, then transformed into dMRI space to serve as the seed region of interest for dMRI-based tractography. Cortico-cortical association fibers initiated from each gyrus were categorized into long- and short-range ones, based on the other end of fiber terminating in non-adjacent or adjacent gyri of the seed gyrus, respectively. The regional CCMI was defined as the ratio between number of short-range association tracts and that of all association tracts traced from one of 68 parcellated gyri. The developmental trajectory of the whole brain CCMI follows a quadratic model with initial decreases from 2 to 16 years followed by later increases after 16 years. Regional CCMI is heterogeneous among different cortical gyri with CCMI dropping to the lowest value earlier in primary somatosensory cortex and visual cortex while later in the prefrontal cortex. The proposed CCMI may serve as sensitive biomarker for brain development under normal or pathological conditions.

To explore the alteration in cerebral blood flow (CBF) and functional connectivity between survivors with recent onset post-traumatic stress disorder (PTSD) and without PTSD, survived from the same coal mine flood disaster. In this study, a processing pipeline using arterial spin labeling (ASL) sequence was proposed. Considering low spatial resolution of ASL sequence, a linear regression method was firstly used to correct the partial volume (PV) effect for better CBF estimation. Then the alterations of CBF between two groups were analyzed using both uncorrected and PV-corrected CBF maps. Based on altered CBF regions detected from the CBF analysis as seed regions, the functional connectivity abnormities in PTSD patients was investigated. The CBF analysis using PV-corrected maps indicates CBF deficits in the bilateral frontal lobe, right superior frontal gyrus and right corpus callosum of PTSD patients, while only right corpus callosum was identified in uncorrected CBF analysis. Furthermore, the regional CBF of the right superior frontal gyrus exhibits significantly negative correlation with the symptom severity in PTSD patients. The resting-state functional connectivity indicates increased connectivity between left frontal lobe and right parietal lobe. These results indicate that PV-corrected CBF exhibits more subtle perfusion changes and may benefit further perfusion and connectivity analysis. The symptom-specific perfusion deficits and aberrant connectivity in above memory-related regions may be putative biomarkers for recent onset PTSD induced by a single prolonged trauma exposure and help predict the severity of PTSD.

The aim of this study was to investigate the factional anisotropy (FA) in various regions of developing rabbit brain using magnetic resonance diffusion tensor imaging (MR DTI) at 3 T. A whole-body clinical MR imaging (MRI) scanner with a 15-channel high resolution knee coil was used. An echo-planar-imaging (EPI)-DTI pulse sequence was performed. Five 5 week-old New Zealand white (NZW) rabbits underwent MRI once per week for 24 weeks. After scanning, FA maps were obtained. ROIs (regions of interests) in the frontal lobe, parietal & temporal lobe, and occipital lobe were measured. FA changes with time were evaluated with a linear regression analysis. The results show that the FA values in all lobes of the brain increased linearly with age. The ranking of FA values was FA(frontal lobe) < FA(parietal & temporal lobe) > FA(occipital lobe). There was significant difference (p < 0.05) among these lobes. FA values are associated with the nerve development and brain functions. The FA change rate could be a biomarker to monitor the brain development. Understanding the FA values of various lobes during development could provide helpful information to diagnosis the abnormal syndrome earlier and have a better treatment and prognosis. This study established a brain MR-DTI protocol for rabbits to investigate the brain anatomy during development using clinical MRI. This technique can be further applied to the pre-clinical diagnosis, treatment, prognosis and follow-up of brain lesions.

To explore the alteration in white matter between survivors with recent onset post-traumatic stress disorder (PTSD) and without PTSD, who survived from the same coal mine flood disaster, the diffusion tensor imaging (DTI) sequences were analyzed using DTI studio and statistical parametric mapping (SPM) packages in this paper. From DTI sequence, the fractional anisotropy (FA) value describes the degree of anisotropy of a diffusion process, while the apparent diffusion coefficient (ADC) value reflects the magnitude of water diffusion. The DTI analyses between PTSD and non-PTSD indicate lower FA values in the right caudate nucleus, right middle temporal gyrus, right fusiform gyrus, and right superior temporal gyrus, and higher ADC values in the right superior temporal gyrus and right corpus callosum of the subjects with PTSD. These results are partly in line with our previous volume and cortical thickness analyses, indicating the importance of multi-modality analysis for PTSD.

Diffusion Tensor Imaging (DTI) uses single shell gradient encoding scheme for studying brain tissue diffusion. NODDI (Neurite Orientation Dispersion and Density Imaging) incorporates a gradient scheme with multiple b-values which is used to characterize neurite density and coherence of neuron fiber orientations. Similarly, the diffusion kurtosis imaging also uses a multiple shell scheme to quantify non-Gaussian diffusion but does not assume a tissue model like NODDI. In this study we investigate the connection between metrics derived by NODDI and DKI in children with ages from 46 weeks to 6 years. We correlate the NODDI metrics and Kurtosis measures from the same ROIs in multiple brain regions. We compare the range of these metrics between neonates (46 - 47 weeks), infants (2 -10 months) and young children (2 – 6 years). We find that there exists strong correlation between neurite density vs. mean kurtosis, orientation dispersion vs. kurtosis fractional anisotropy (FA) in pediatric brain imaging.

MRI-based multi-site trials now routinely include some form of diffusion-weighted imaging (DWI) in their protocol. These studies can include data originating from scanners built by different vendors, each with their own set of unique protocol restrictions, including restrictions on the number of available gradient directions, whether an externally generated list of gradient directions can be used, and restrictions on the echo time (TE). One challenge of multi-site studies is to create a common imaging protocol that will result in a reliable and accurate set of diffusion metrics. The present study describes the effect of site, scanner vendor, field strength, and TE on two common metrics: the first moment of the diffusion tensor field (mean diffusivity, MD), and the fractional anisotropy (FA). We have shown in earlier work that ROI metrics and the mean of MD and FA histograms are not sufficiently sensitive for use in site characterization. Here we use the distance between whole brain histograms of FA and MD to investigate within- and between-site effects. We concluded that the variability of DTI metrics due to site, vendor, field strength, and echo time could influence the results in multi-center trials and that histogram distance is sensitive metrics for each of these variables.

The aim of this study was to examine the use of R2* mapping in maternal and fetal sub-regions of the placenta with the aim of providing a reference for blood oxygenation levels during normative development. There have been a number of MR relaxation studies of placental tissues in-utero, but none have reported R2* value changes with age, or examined differences in sub-regions of the placenta. Here specialized long-duration Multi-frame R2* imaging was used to create a stable estimate for R2* values in different placental regions in healthy pregnant volunteers not imaged for clinical reasons. 27 subjects were recruited and scanned up to 3 times during their pregnancy. A multi-slice dual echo EPI based BOLD acquisition was employed and repeated between 90 and 150 times over 3 to 5 minutes to provide a high accuracy estimate of the R2* signal level. Acquisitions were also repeated in 13 cases within a visit to evaluate reproducibility of the method in a given subject. Experimental results showed R2* measurements were highly repeatable within a visit with standard deviation of (0.76). Plots of all visits against gestational age indicated clear correlations showing decreases in R2* with age. This increase was consistent was also consistent over time in multiple visits of the same volunteer during their pregnancy. Maternal and fetal regional changes with gestational age followed the same trend with increase in R2* over the gestational age.

Two-photon fluorescence microscopy (TPFM) is a perfect optical imaging equipment to monitor the interaction between fast moving viruses and hosts. However, due to strong unavoidable background noises from the culture, videos obtained by this technique are too noisy to elaborate this fast infection process without video image processing. In this study, we developed a novel scheme to eliminate background noises, recover background bacteria images and improve video qualities. In our scheme, we modified and implemented the following methods for both host and virus videos: correlation method, round identification method, tree-structured nonlinear filters, Kalman filters, and cell tracking method. After these procedures, most of noises were eliminated and host images were recovered with their moving directions and speed highlighted in the videos. From the analysis of the processed videos, 93% bacteria and 98% viruses were correctly detected in each frame on average.

Currently, unlike IALSC-defined thoracic lymph node zones, no explicitly provided definitions for lymph nodes in other body regions are available. Yet, definitions are critical for standardizing the recognition, delineation, quantification, and reporting of lymphadenopathy in other body regions. Continuing from our previous work in the thorax, this paper proposes a standardized definition of the grouping of pelvic lymph nodes into 10 zones. We subsequently employ our earlier Automatic Anatomy Recognition (AAR) framework designed for body-wide organ modeling, recognition, and delineation to actually implement these zonal definitions where the zones are treated as anatomic objects. First, all 10 zones and key anatomic organs used as anchors are manually delineated under expert supervision for constructing fuzzy anatomy models of the assembly of organs together with the zones. Then, optimal hierarchical arrangement of these objects is constructed for the purpose of achieving the best zonal recognition. For actual localization of the objects, two strategies are used — optimal thresholded search for organs and one-shot method for the zones where the known relationship of the zones to key organs is exploited. Based on 50 computed tomography (CT) image data sets for the pelvic body region and an equal division into training and test subsets, automatic zonal localization within 1–3 voxels is achieved.

In this contribution we study different methods of automatic volume estimation for pancreatic islets which can be used in the quality control step prior to the islet transplantation. The total islet volume is an important criterion in the quality control. Also, the individual islet volume distribution is interesting — it has been indicated that smaller islets can be more effective. A 2D image of a microscopy slice containing the islets is acquired. The input of the volume estimation methods are segmented images of individual islets. The segmentation step is not discussed here. We consider simple methods of volume estimation assuming that the islets have spherical or ellipsoidal shape. We also consider a local stereological method, namely the nucleator. The nucleator does not rely on any shape assumptions and provides unbiased estimates if isotropic sections through the islets are observed. We present a simulation study comparing the performance of the volume estimation methods in different scenarios and an experimental study comparing the methods on a real dataset.

Natural image classification has been a hot topic in computer vision and pattern recognition research field. Since the performance of an image classification system can be improved by feature selection, many image feature selection methods have been developed. However, the existing supervised feature selection methods are typically driven by the class label information that are identical for different samples from the same class, ignoring with-in class image variability and therefore degrading the feature selection performance. In this study, we propose a novel feature selection method, driven by human brain activity signals collected using fMRI technique when human subjects were viewing natural images of different categories. The fMRI signals associated with subjects viewing different images encode the human perception of natural images, and therefore may capture image variability within- and cross- categories. We then select image features with the guidance of fMRI signals from brain regions with active response to image viewing. Particularly, bag of words features based on GIST descriptor are extracted from natural images for classification, and a sparse regression base feature selection method is adapted to select image features that can best predict fMRI signals. Finally, a classification model is built on the select image features to classify images without fMRI signals. The validation experiments for classifying images from 4 categories of two subjects have demonstrated that our method could achieve much better classification performance than the classifiers built on image feature selected by traditional feature selection methods.

The active contour model has good performance in boundary extraction for medical images; particularly, Gradient Vector Flow (GVF) active contour model shows good performance at concavity convergence and insensitivity to initialization, yet it is susceptible to edge leaking, deep and narrow concavities, and has some issues handling noisy images. This paper proposes a novel external force, called Iterative Weighted Average Diffusion (IWAD), which used in tandem with parametric active contours, provides superior performance in images with high values of concavity. The image gradient is first turned into an edge image, smoothed, and modified with enhanced corner detection, then the IWAD algorithm diffuses the force at a given pixel based on its 3x3 pixel neighborhood. A forgetting factor, φ, is employed to ensure that forces being spread away from the boundary of the image will attenuate. The experimental results show better behavior in high curvature regions, faster convergence, and less edge leaking than GVF when both are compared to expert manual segmentation of the images.

Even with recent hardware improvements, superpixel segmentation of large 3D medical images at interactive speed (<500 ms) remains a challenge. We will describe methods to achieve such performances using a GPU based hybrid framework implementing wavefront propagation and cellular automata resolution.

Tasks will be scheduled in blocks (work units) using a wavefront propagation strategy, therefore allowing sparse scheduling. Because work units has been designed as spatially cohesive, the fast Thread Group Shared Memory can be used and reused through a Gauss-Seidel like acceleration. The work unit partitioning scheme will however vary on odd- and even-numbered iterations to reduce convergence barriers. Synchronization will be ensured by an 8-step 3D variant of the traditional Red Black Ordering scheme. An attack model and early termination will also be described and implemented as additional acceleration techniques.

Using our hybrid framework and typical operating parameters, we were able to compute the superpixels of a high-resolution 512x512x512 aortic angioCT scan in 283 ms using a AMD R9 290X GPU. We achieved a 22.3X speed-up factor compared to the published reference GPU implementation.

In this paper, an automatic method is proposed to recognize the liver on clinical 3D CT images. The proposed method effectively use statistical shape model of the liver. Our approach consist of three main parts: (1) model training, in which shape variability is detected using principal component analysis from the manual annotation; (2) model localization, in which a fast Euclidean distance transformation based method is able to localize the liver in CT images; (3) liver recognition, the initial mesh is locally and iteratively adapted to the liver boundary, which is constrained with the trained shape model. We validate our algorithm on a dataset which consists of 20 3D CT images obtained from different patients. The average ARVD was 8.99%, the average ASSD was 2.69mm, the average RMSD was 4.92mm, the average MSD was 28.841mm, and the average MSD was 13.31%.

Dynamic contrast enhanced (DCE) MRI has emerged as a reliable and diagnostically useful functional imaging technique. DCE protocol typically lasts 3-15 minutes and results in a time series of N volumes. For automated analysis, it is important that volumes acquired at different times be spatially coregistered. We have recently introduced a novel 4D, or volume time series, coregistration tool based on a user-specified target volume of interest (VOI). However, the relationship between coregistration accuracy and target VOI size has not been investigated. In this study, coregistration accuracy was quantitatively measured using various sized target VOIs. Coregistration of 10 DCE-MRI mouse head image sets were performed with various sized VOIs targeting the mouse brain. Accuracy was quantified by measures based on the union and standard deviation of the coregistered volume time series. Coregistration accuracy was determined to improve rapidly as the size of the VOI increased and approached the approximate volume of the target (mouse brain). Further inflation of the VOI beyond the volume of the target (mouse brain) only marginally improved coregistration accuracy. The CPU time needed to accomplish coregistration is a linear function of N that varied gradually with VOI size. From the results of this study, we recommend the optimal size of the VOI to be slightly overinclusive, approximately by 5 voxels, of the target for computationally efficient and accurate coregistration.

In recent years, several studies have shown that the canine retina model offers important insight for our understanding of human retinal diseases. Several therapies developed to treat blindness in such models have already moved onto human clinical trials, with more currently under development [1]. Optical coherence tomography (OCT) offers a high resolution imaging modality for performing in-vivo analysis of the retinal layers. However, existing algorithms for automatically segmenting and analyzing such data have been mostly focused on the human retina. As a result, canine retinal images are often still being analyzed using manual segmentations, which is a slow and laborious task. In this work, we propose a method for automatically segmenting 5 boundaries in canine retinal OCT. The algorithm employs the position relationships between different boundaries to adaptively enhance the gradient map. A region growing algorithm is then used on the enhanced gradient maps to find the five boundaries separately. The automatic segmentation was compared against manual segmentations showing an average absolute error of 5.82 ± 4.02 microns.

The trabecular bone microstructure is an important factor in the development of osteoporosis. It is well known that its deterioration is one effect when osteoporosis occurs. Previous research showed that the analysis of trabecular bone microstructure enables more precise diagnoses of osteoporosis compared to a sole measurement of the mineral density. Microstructure parameters are assessed on volumetric images of the bone acquired either with high-resolution magnetic resonance imaging, high-resolution peripheral quantitative computed tomography or high-resolution computed tomography (CT), with only CT being applicable to the spine, which is one of clinically most relevant fracture sites. However, due to the high radiation exposure for imaging the whole spine these measurements are not applicable in current clinical routine. In this work, twelve vertebrae from three different donors were scanned with standard and low radiation dose. Trabecular bone microstructure parameters were assessed for CT images reconstructed with statistical iterative reconstruction (SIR) and analytical filtered backprojection (FBP). The resulting structure parameters were correlated to the biomechanically determined fracture load of each vertebra. Microstructure parameters assessed for low-dose data reconstructed with SIR significantly correlated with fracture loads as well as parameters assessed for standard-dose data reconstructed with FBP. Ideal results were achieved with low to zero regularization strength yielding microstructure parameters not significantly different from those assessed for standard-dose FPB data. Moreover, in comparison to other approaches, superior noise-resolution trade-offs can be found with the proposed methods.

Trabecular bone and its microarchitecture are of prime importance for health. This paper focuses onto the relationship between bone microarchitecture and the texture found on micro CT projections. From a small animal study, 5 mice legs were studied by microCT at a resolution of 6μm. The 3D reconstructions are only used as ground truth for their microarchitecture parameters. The study uses 2 different sets of tomographic data : 3 volumes acquired at ANU in Canberra and 2 volumes acquired in Nantes. For each projection set, we determine the texture orientation onto a ROI region of both medial epiphysis and diaphisys using a local variance computed onto Mojette projections from the ROI.

Ribs and clavicles in posterior-anterior (PA) digital chest radiographs often overlap with lung abnormalities such as nodules, and cause missing of these abnormalities, it is therefore necessary to remove or reduce the ribs in chest radiographs. The purpose of this study was to develop a fully automated algorithm to segment ribs within lung area in digital radiography (DR) for removal of the ribs. The rib segmentation algorithm consists of three steps. Firstly, a radiograph was pre-processed for contrast adjustment and noise removal; second, generalized Hough transform was employed to localize the lower boundary of the ribs. In the third step, a novel bilateral dynamic programming algorithm was used to accurately segment the upper and lower boundaries of ribs simultaneously. The width of the ribs and the smoothness of the rib boundaries were incorporated in the cost function of the bilateral dynamic programming for obtaining consistent results for the upper and lower boundaries. Our database consisted of 93 DR images, including, respectively, 23 and 70 images acquired with a DR system from Shanghai United-Imaging Healthcare Co. and from GE Healthcare Co. The rib localization algorithm achieved a sensitivity of 98.2% with 0.1 false positives per image. The accuracy of the detected ribs was further evaluated subjectively in 3 levels: "1", good; "2", acceptable; "3", poor. The percentages of good, acceptable, and poor segmentation results were 91.1%, 7.2%, and 1.7%, respectively. Our algorithm can obtain good segmentation results for ribs in chest radiography and would be useful for rib reduction in our future study.

Computational medicine aims at developing patient-specific models to help physicians in the diagnosis and treatment selection for patients. The spine, and other skeletal structures, is an articulated object, composed of rigid bones (vertebrae) and non-rigid parts (intervertebral discs (IVD), ligaments and muscles). These components are usually extracted from different image modalities, involving patient repositioning. In the case of the spine, these models require the segmentation of IVDs from MR and vertebrae from CT. In the literature, there exists a vast selection of segmentations methods, but there is a lack of approaches to align the vertebrae and IVDs. This paper presents a method to create patient-specific finite element meshes for biomechanical simulations, integrating rigid and non-rigid parts of articulated objects. First, the different parts are aligned in a complete surface model. Vertebrae extracted from CT are rigidly repositioned in between the IVDs, initially using the IVDs location and then refining the alignment using the MR image with a rigid active shape model algorithm. Finally, a mesh morphing algorithm, based on B-splines, is employed to map a template finite-element (volumetric) mesh to the patient-specific surface mesh. This morphing reduces possible misalignments and guarantees the convexity of the model elements. Results show that the accuracy of the method to align vertebrae into MR, together with IVDs, is similar to that of the human observers. Thus, this method is a step forward towards the automation of patient-specific finite element models for biomechanical simulations.

Important features in Parkinson's disease (PD) are degenerations and losses of dopamine neurons in corpus striatum. 123I-FP-CIT can visualize activities of the dopamine neurons. The activity radio of background to corpus striatum is used for diagnosis of PD and Dementia with Lewy Bodies (DLB). The specific activity can be observed in the corpus striatum on SPECT images, but the location and the shape of the corpus striatum on SPECT images only are often lost because of the low uptake. In contrast, MR images can visualize the locations of the corpus striatum. The purpose of this study was to realize a quantitative image analysis for the SPECT images by using image registration technique with brain MR images that can determine the region of corpus striatum. In this study, the image fusion technique was used to fuse SPECT and MR images by intervening CT image taken by SPECT/CT. The mutual information (MI) for image registration between CT and MR images was used for the registration. Six SPECT/CT and four MR scans of phantom materials are taken by changing the direction. As the results of the image registrations, 16 of 24 combinations were registered within 1.3mm. By applying the approach to 32 clinical SPECT/CT and MR cases, all of the cases were registered within 0.86mm. In conclusions, our registration method has a potential in superimposing MR images on SPECT images.

About 50% of subjects infected with HIV present deficits in cognitive domains, which are known collectively as HIV associated neurocognitive disorder (HAND). The underlying synaptodendritic damage can be captured using resting state functional MRI, as has been demonstrated by a few earlier studies. Such damage may induce topological changes of brain connectivity networks. We test this hypothesis by capturing the functional interdependence of 90 brain network nodes using a Mutual Connectivity Analysis (MCA) framework with non-linear time series modeling based on Generalized Radial Basis function (GRBF) neural networks. The network nodes are selected based on the regions defined in the Automated Anatomic Labeling (AAL) atlas. Each node is represented by the average time series of the voxels of that region. The resulting networks are then characterized using graph-theoretic measures that quantify various network topology properties at a global as well as at a local level. We tested for differences in these properties in network graphs obtained for 10 subjects (6 male and 4 female, 5 HIV+ and 5 HIV-). Global network properties captured some differences between these subject cohorts, though significant differences were seen only with the clustering coefficient measure. Local network properties, such as local efficiency and the degree of connections, captured significant differences in regions of the frontal lobe, precentral and cingulate cortex amongst a few others. These results suggest that our method can be used to effectively capture differences occurring in brain network connectivity properties revealed by resting-state functional MRI in neurological disease states, such as HAND.

Transcranial Direct Current Stimulation (tDCS) is a non-invasive electrical stimulation technique investigated for a broad range of medical and performance indications. Whereas prior studies have focused exclusively on direct neuron polarization, our hypothesis is that tDCS directly modulates endothelial cells leading to transient changes in blood-brain-barrier (BBB) permeability (P) that are highly meaningful for neuronal activity. For this, we developed state-of-the-art imaging and animal models to quantify P to various sized solutes after tDCS treatment. tDCS was administered using a constant current stimulator to deliver a 1mA current to the right frontal cortex of rat (approximately 2 mm posterior to bregma and 2 mm right to sagittal suture) to obtain similar physiological outcome as that in the human tDCS application studies. Sodium fluorescein (MW=376), or FITC-dextrans (20K and 70K), in 1% BSA mammalian Ringer was injected into the rat (SD, 250-300g) cerebral circulation via the ipsilateral carotid artery by a syringe pump at a constant rate of ~3 ml/min. To determine P, multiphoton microscopy with 800-850 nm wavelength laser was applied to take the images from the region of interest (ROI) with proper microvessels, which are 100-200 micron below the pia mater. It shows that the relative increase in P is about 8-fold for small solute, sodium fluorescein, ~35-fold for both intermediate sized (Dex-20k) and large (Dex-70k) solutes, 10 min after 20 min tDCS pretreatment. All of the increased permeability returns to the control after 20 min post treatment. The results confirmed our hypothesis.

Current connectivity diagrams of human brain image data are either overly complex or overly simplistic. In this work we introduce simple yet accurate interactive visual representations of multiple brain image structures and the connectivity among them. We map cortical surfaces extracted from human brain magnetic resonance imaging (MRI) data onto 2D surfaces that preserve shape (angle), extent (area), and spatial (neighborhood) information for 2D (circular disk) and 3D (spherical) mapping, split these surfaces into separate patches, and cluster functional and diffusion tractography MRI connections between pairs of these patches. The resulting visualizations are easier to compute on and more visually intuitive to interact with than the original data, and facilitate simultaneous exploration of multiple data sets, modalities, and statistical maps.

This work reports on a comparative study between five manual and automated methods for intra-subject pair-wise registration of images from different modalities. The study includes a variety of inter-modal image registrations (MR-CT, PET-CT, PET-MR) utilizing different methods including two manual point-based techniques using rigid and similarity transformations, one automated point-based approach based on Iterative Closest Point (ICP) algorithm, and two automated intensity-based methods using mutual information (MI) and normalized mutual information (NMI). These techniques were employed for inter-modal registration of brain images of 9 subjects from a publicly available dataset, and the results were evaluated qualitatively via checkerboard images and quantitatively using root mean square error and MI criteria. In addition, for each inter-modal registration, a paired t-test was performed on the quantitative results in order to find any significant difference between the results of the studied registration techniques.

The goal of this project was to develop two age appropriate atlases (neonatal and one year old) that account for the rapid growth and maturational changes that occur during early development. Tissue maps from this age group were initially created by manually correcting the resulting tissue maps after applying an expectation maximization (EM) algorithm and an adult atlas to pediatric subjects. The EM algorithm classified each voxel into one of ten possible tissue types including several subcortical structures. This was followed by a novel level set segmentation designed to improve differentiation between distal cortical gray matter and white matter. To minimize the req uired manual corrections, the adult atlas was registered to the pediatric scans using high -dimensional, symmetric image normalization (SyN) registration. The subject images were then mapped to an age specific atlas space, again using SyN registration, and the resulting transformation applied to the manually corrected tissue maps. The individual maps were averaged in the age specific atlas space and blurred to generate the age appropriate anatomical priors. The resulting anatomical priors were then used by the EM algorithm to re-segment the initial training set as well as an independent testing set. The results from the adult and age-specific anatomical priors were compared to the manually corrected results. The age appropriate atlas provided superior results as compared to the adult atlas. The image analysis pipeline used in this work was built using the open source software package BRAINSTools.

The goal of this study is predicting human age using neuro-metrics derived from structural MRI, as well as investigating the relationships between age and predictive neuro-metrics. To this end, a cohort of healthy subjects were recruited from 1000 Functional Connectomes Project. The ages of the participations were ranging from 7 to 83 (36.17±20.46). The structural MRI for each subject was preprocessed using FreeSurfer, resulting in regional cortical thickness, mean curvature, regional volume and regional surface area for 148 anatomical parcellations. The individual age was predicted from the combination of regional and inter-regional neuro-metrics. The prediction accuracy is r = 0.835, p < 0.00001, evaluated by Pearson correlation coefficient between predicted ages and actual ages. Moreover, the LASSO linear regression also found certain predictive features, most of which were inter-regional features. The turning-point of the developmental trajectories in human brain was around 40 years old based on regional cortical thickness. In conclusion, structural MRI could be potential biomarkers for the aging in human brain. The human age could be successfully predicted from the combination of regional morphometry and inter-regional morphological similarity. The inter-regional measures could be beneficial to investigating human brain connectome.

Neuroimaging consortia such as ENIGMA can significantly improve power to discover factors that affect the human brain by pooling statistical inferences across cohorts to draw generalized conclusions from populations around the world. Voxelwise analyses such as tensor-based morphometry also allow an unbiased search for effects throughout the brain. Even so, such consortium-based analyses are limited by a lack of high-powered methods to harmonize voxelwise information across study populations and scanners. While the simplest approach may be to map all images to a single standard space, the benefits of cohort-specific templates have long been established. Here we studied methods to pool voxel-wise data across sites using templates customized for each cohort but providing a meaningful common space across all studies for voxelwise comparisons. As non-linear 3D MRI registrations represent mappings between images at millimeter resolution, we need to consider the reliability of these mappings. To evaluate these mappings, we calculated test-retest statistics on the volumetric maps of expansion and contraction. Further, we created study-specific brain templates for ten T1-weighted MRI datasets, and a common space from four study-specific templates. We evaluated the efficacy of using a two-step registration framework versus a single standard space. We found that the two-step framework more reliably mapped subjects to a common space.

Parkinson's Disease (PD) is a movement disorder characterized by the loss of dopamine neurons in the substantia nigra pars compacta (SNpc) and norepinephrine neurons in the locus coeruleus (LC). To further understand the pathophysiology of PD, the input neurons of the SNpc and LC will be transsynapticly traced in mice using a fluorescent recombinant rabies virus (RbV) and imaged using serial two-photon tomography (STP). A mapping between these images and a brain atlas must be found to accurately determine the locations of input neurons in the brain. Therefore a registration pipeline to align the Allen Reference Atlas (ARA) to these types of images was developed. In the preprocessing step, a brain mask was generated from the transsynaptic tracing images using simple morphological operators. The masks were then registered to the ARA using Large Deformation Diffeomorphic Metric Mapping (LDDMM), an algorithm specialized for calculating anatomically realistic transforms between images. The pipeline was then tested on an STP scan of a mouse brain labeled by an adeno-associated virus (AAV). Based on qualitative evaluation of the registration results, the pipeline was found to be sufficient for use with transsynaptic RbV tracing.

In this paper we present preliminary results of comparison of automatic segmentations of the infarct core, between that obtained from CT perfusion (based on time to peak parameter) and diffusion weighted imaging (DWI). For each patient, the two imaging volumes were automatically co-registered to a common frame of reference based on an acquired CT angiography image. The accuracy of image registration is measured by the overlap of the segmented brain from both images (CT perfusion and DWI), measured within their common field of view. Due to the limitations of the study, DWI was acquired as a follow up scan up to a week after initial CT based imaging. However, we found significant overlap of the segmented brain (Jaccard indices of approximately 0.8) and the percentage of infarcted brain tissue from the two modalities were still fairly highly correlated (correlation coefficient of approximately 0.9). The results are promising with more data needed in future for clinical inference.

Image based atlases for rats brain have a significant impact on pre-clinical research. In this project we acquired T1-weighted images from Wistar rodent brains with fine 59μm isotropical resolution for generation of the atlas template image. By applying post-process procedures using a semi-automatic brain extraction method, we delineated the brain tissues from source data. Furthermore, we applied a symmetric group-wise normalization method to generate an optimized template of T1 image of rodent brain, then aligned our template to the Waxholm Space. In addition, we defined several simple and explicit landmarks to corresponding our template with the well known Paxinos stereotaxic reference system. Anchoring at the origin of the Waxholm Space, we applied piece-wise linear transformation method to map the voxels of the template into the coordinates system in Paxinos' stereotoxic coordinates to facilitate the labelling task. We also cross-referenced our data with both published rodent brain atlas and image atlases available online, methodologically labelling the template to produce a Wistar brain atlas identifying more than 130 structures. Particular attention was paid to the cortex and cerebellum, as these areas encompass the most researched aspects of brain functions. Moreover, we adopted the structure hierarchy and naming nomenclature common to various atlases, so that the names and hierarchy structure presented in the atlas are readily recognised for easy use. It is believed the atlas will present a useful tool in rodent brain functional and pharmaceutical studies.

Resting state functional connectivity (rsFC) studies using fMRI provides a great deal of knowledge on the spatiotemporal organization of the brain. The relationships between and within a number of resting state functional networks, namely the default mode network (DMN), salience network (SN) and executive control network (ECN) have been intensely studied in basic and clinical cognitive neuroscience [1]. However, the presumption of spatial and temporal stationarity has mostly restricted the assessment of rsFC [1]. In this study, sliding window correlation analysis and k-means clustering were exploited to examine the temporal dynamics of rsFC of these three networks in 24 abstinent methamphetamine dependents. Afterwards, using canonical correlation analysis (CCA) the possible relationship between the level of self-reported craving and the temporal dynamics was examined. Results indicate that the rsFC transits between 6 discrete "FC states" in the meth dependents. CCA results show that higher levels of craving are associated with higher probability of transiting from state 4 to 6 (positive FC of DMN-ECN getting weak and negative FC of DMN-SN appearing) and staying in state 4 (positive FC of DMN-ECN), lower probability of staying in state 2 (negative FC of DMN-ECN), transiting from state 4 to 2 (change of positive FC of DMN-ECN to negative FC), and transiting from state 3 to 5 (appearance of negative FC of DMN-SN and positive FC of DMN-ECN with the presence of negative FC of SN-ECN). Quantitative measures of temporal dynamics in large-scale brain networks could bring new added values to increase potentials for applications of rsfMRI in addiction medicine.

Brain imaging studies identified brain networks that play a key role in nicotine dependence-related behavior. Functional connectivity of the brain is dynamic; it changes over time due to different causes such as learning, or quitting a habit. Functional connectivity analysis is useful in discovering and comparing patterns between functional magnetic resonance imaging (fMRI) scans of patients’ brains. In the resting state, the patient is asked to remain calm and not do any task to minimize the contribution of external stimuli. The study of resting-state fMRI networks have shown functionally connected brain regions that have a high level of activity during this state. In this project, we are interested in the relationship between these functionally connected brain regions to identify nicotine dependent patients, who underwent a smoking cessation treatment. Our approach is on the comparison of the set of connections between the fMRI scans before and after treatment. We applied support vector machines, a machine learning technique, to classify patients based on receiving the treatment or the placebo. Using the functional connectivity (CONN) toolbox, we were able to form a correlation matrix based on the functional connectivity between different regions of the brain. The experimental results show that there is inadequate predictive information to classify nicotine dependent patients using the SVM classifier. We propose other classification methods be explored to better classify the nicotine dependent patients.

A hybrid system combining fluorescence molecular tomography (FMT) and magnetic resonance imaging (MRI) is attractive for preclinical imaging as it allows fusion of molecular information derived from FMT and anatomical reference data derived from MRI. We have previously developed such a system and demonstrated its performance in biological applications. For reconstruction slab geometry with homogeneous optical parameters was assumed, which led to undesirable artifacts. In order to exploit the power of the hybrid system, the use of MRI derived anatomical information, as a constraint for FMT reconstruction, appears logical. Heterogeneity of tissues and irregular surface derived from MRI can be accounted for by generating a mesh using the finite element method (FEM), and attributing optical parameters to individual mesh points. We have established a forward simulation tool based on TOAST++ to mimic an FMT experiment. MRI images were recorded on a 9.4T MR scanner using a T1-weighted pulse sequence. The voxelized dataset was processed by iso2mesh to yield a 3D-mesh. Four steps of FMT simulation were included: 1) Assignment of optical properties, 2) Specification of boundary conditions and generation of 3) excitation and 4) emission maps. FEM-derived results were compared with those obtained using the analytical solution of Green's function and with experimental data with a single fluorescent inclusion in a silicon phantom. Once, the forward modeling method is properly validated it will be used as a central element of a reconstruction algorithm for analyzing data derived from a hybrid FMT/MRI setup.

Confocal Raman microscopy has become an advancing technique in the characterization of drug transport into the skin. In this study the skin penetration of a local anesthetic from a semisolid preparation was investigated. Furthermore, the effect of the chemical enhancers propylene glycol and POE-23-lauryl ether on its penetration was investigated. The results show that confocal Raman microscopy may provide detailed information on the penetration of APIs into the skin and may elucidate their distribution within the skin with high resolution. The results of the CRM analysis are fully in line with those of conventional permeation and penetration experiments.

High resolution, 100 frames/sec intravascular optical coherence tomography (IVOCT) can distinguish plaque types, but further validation is needed, especially for automated plaque characterization. We developed experimental and 3D registration methods, to provide validation of IVOCT pullback volumes using microscopic, brightfield and fluorescent cryoimage volumes, with optional, exactly registered cryo-histology. The innovation was a method to match an IVOCT pullback images, acquired in the catheter reference frame, to a true 3D cryo-image volume. Briefly, an 11-parameter, polynomial virtual catheter was initialized within the cryo-image volume, and perpendicular images were extracted, mimicking IVOCT image acquisition. Virtual catheter parameters were optimized to maximize cryo and IVOCT lumen overlap. Local minima were possible, but when we started within reasonable ranges, every one of 24 digital phantom cases converged to a good solution with a registration error of only +1.34±2.65μm (signed distance). Registration was applied to 10 ex-vivo cadaver coronary arteries (LADs), resulting in 10 registered cryo and IVOCT volumes yielding a total of 421 registered 2D-image pairs. Image overlays demonstrated high continuity between vascular and plaque features. Bland- Altman analysis comparing cryo and IVOCT lumen area, showed mean and standard deviation of differences as 0.01±0.43 mm2. DICE coefficients were 0.91±0.04. Finally, visual assessment on 20 representative cases with easily identifiable features suggested registration accuracy within one frame of IVOCT (±200μm), eliminating significant misinterpretations introduced by 1mm errors in the literature. The method will provide 3D data for training of IVOCT plaque algorithms and can be used for validation of other intravascular imaging modalities.

Telemonitoring is becoming increasingly important as the proportion of the population living with cardiovascular disease (CVD) increases. Currently used health parameters in the suite of telemonitoring tools lack the sensitivity and specificity to accurately predict heart failure events, forcing physicians to play a reactive versus proactive role in patient care. A novel cardiac output (CO) monitoring device is proposed that leverages a custom smart phone application and a wearable electrical impedance tomography (EIT) system. The purpose of this work is to explore the potential of using respiratory-gated EIT to quantify stroke volume (SV) and assess its feasibility using real data. Simulations were carried out using the 4D XCAT model to create anatomically realistic meshes and electrical conductivity profiles representing the human thorax and the intrathoracic tissue. A single 5-second period respiration cycle with chest/lung expansion was modeled with end-diastole (ED) and end-systole (ES) heart volumes to evaluate how effective EIT-based conductivity changes represent clinically significant differences in SV. After establishing a correlation between conductivity changes and SV, the applicability of the respiratory-gated EIT was refined using data from the PhysioNet database to estimate the number of useful end-diastole (ED) and end-systole (ES) heart events attained over a 3.3 minute period. The area associated with conductivity changes was found to correlate to SV with a correlation coefficient of 0.92. A window of 12.5% around peak exhalation was found to be the optimal phase of the respiratory cycle from which to record EIT data. Within this window, ~47 useable ED and ES were found with a standard deviation of 28 using 3.3 minutes of data for 20 patients.

Segmentation of left ventricle (LV) in contrast-enhanced cardiac MR images is a challenging task because of high variability in the image intensity. This is due to a) wash-in and wash-out of the contrast agent over time and b) poor contrast around the epicardium (outer wall) region. Current approaches for segmentation of the endocardium (inner wall) usually involve application of a threshold within the region of interest, followed by refinement techniques like active contours. A limitation of this method is under-segmentation of the inner wall because of gradual loss of contrast at the wall boundary. On the other hand, the challenge in outer wall segmentation is the lack of reliable boundaries because of poor contrast. There are four main contributions in this paper to address the aforementioned issues. First, a seed image is selected using variance based approach on 4D time-frame images over which initial endocardium and epicardium is segmented. Secondly, we propose a patch based feature which overcomes the problem of gradual contrast loss for LV endocardium segmentation. Third, we propose a novel Iterative-Edge-Refinement (IER) technique for epicardium segmentation. Fourth, we propose a greedy search algorithm for propagating the initial contour segmented on seed-image across other time frame images. We have experimented our technique on five contrast-enhanced cardiac MR Datasets (4D) having a total of 1097 images. The segmentation results for all 1097 images have been visually inspected by a clinical expert and have shown good accuracy.

Myocardial perfusion imaging using CT (MPI-CT) has the potential to provide quantitative measures of myocardial blood flow (MBF) which can aid the diagnosis of coronary artery disease. We evaluated the quantitative accuracy of MPI-CT in a porcine model of balloon-induced LAD coronary artery ischemia guided by fractional flow reserve (FFR). We quantified MBF at baseline (FFR=1.0) and under moderate ischemia (FFR=0.7) using MPI-CT and compared to fluorescent microsphere-based MBF from high-resolution cryo-images. Dynamic, contrast-enhanced CT images were obtained using a spectral detector CT (Philips Healthcare). Projection-based mono-energetic images were reconstructed and processed to obtain MBF. Three MBF quantification approaches were evaluated: singular value decomposition (SVD) with fixed Tikhonov regularization (ThSVD), SVD with regularization determined by the L-Curve criterion (LSVD), and Johnson-Wilson parameter estimation (JW). The three approaches over-estimated MBF compared to cryo-images. JW produced the most accurate MBF, with average error 33.3±19.2mL/min/100g, whereas LSVD and ThSVD had greater over-estimation, 59.5±28.3mL/min/100g and 78.3±25.6 mL/min/100g, respectively. Relative blood flow as assessed by a flow ratio of LAD-to-remote myocardium was strongly correlated between JW and cryo-imaging, with R²=0.97, compared to R²=0.88 and 0.78 for LSVD and ThSVD, respectively. We assessed tissue impulse response functions (IRFs) from each approach for sources of error. While JW was constrained to physiologic solutions, both LSVD and ThSVD produced IRFs with non-physiologic properties due to noise. The L-curve provided noise-adaptive regularization but did not eliminate non-physiologic IRF properties or optimize for MBF accuracy. These findings suggest that model-based MPI-CT approaches may be more appropriate for quantitative MBF estimation and that cryo-imaging can support the development of MPI-CT by providing spatial distributions of MBF.

We present a new method for nonrigid registration of coronary artery models with layered motion information. 2D nonrigid registration method is proposed that brings layered motion information into correspondence with fluoroscopic angiograms. The registered model is overlaid on top of interventional angiograms to provide surgical assistance during image-guided chronic total occlusion procedures. The proposed methodology is divided into two parts: layered structures alignments and local nonrigid registration. In the first part, inpainting method is used to estimate a layered rigid transformation that aligns layered motion information. In the second part, a nonrigid registration method is implemented and used to compensate for any local shape discrepancy. Experimental evaluation conducted on a set of 7 fluoroscopic angiograms results in a reduced target registration error, which showed the effectiveness of the proposed method over single layered approach.

An evaluation of the relation between parametric imaging results obtained from Digital Subtraction Angiography (DSA) images and blood-flow velocity measured using Doppler ultrasound in patient-specific neurovascular phantoms is provided. A silicone neurovascular phantom containing internal carotid artery, middle cerebral artery and anterior communicating artery was embedded in a tissue equivalent gel. The gel prevented movement of the vessels when blood mimicking fluid was pumped through it to obtain Colour Doppler images. The phantom was connected to a peristaltic pump, simulating physiological flow conditions. To obtain the parametric images, water was pumped through the phantom at various flow rates (100, 120 and 160 ml/min) and 10 ml contrast boluses were injected. DSA images were obtained at 10 frames/sec from the Toshiba C-arm and DSA image sequences were input into LabVIEW software to get parametric maps from time-density curves. The parametric maps were compared with velocities determined by Doppler ultrasound at the internal carotid artery. The velocities measured by the Doppler ultrasound were 38, 48 and 65 cm/s for flow rates of 100, 120 and 160 ml/min, respectively. For the 20% increase in flow rate, the percentage change of blood velocity measured by Doppler ultrasound was 26.3%. Correspondingly, there was a 20% decrease of Bolus Arrival Time (BAT) and 14.3% decrease of Mean Transit Time (MTT), showing strong inverse correlation with Doppler measured velocity. The parametric imaging parameters are quite sensitive to velocity changes and are well correlated to the velocities measured by Doppler ultrasound.

The detection of subendocardial ischemia exhibiting an abnormal transmural perfusion gradient (TPG) may help identify ischemic conditions due to micro-vascular dysfunction. We evaluated the effect of beam hardening (BH) artifacts on TPG quantification using myocardial CT perfusion (CTP). We used a prototype spectral detector CT scanner (Philips Healthcare) to acquire dynamic myocardial CTP scans in a porcine ischemia model with partial occlusion of the left anterior descending (LAD) coronary artery guided by pressure wire-derived fractional flow reserve (FFR) measurements. Conventional 120 kVp and 70 keV projection-based mono-energetic images were reconstructed from the same projection data and used to compute myocardial blood flow (MBF) using the Johnson-Wilson model. Under moderate LAD occlusion (FFR~0.7), we used three 5 mm short axis slices and divided the myocardium into three LAD segments and three remote segments. For each slice and each segment, we characterized TPG as the mean "endo-to-epi" transmural flow ratio (TFR). BH-induced hypoenhancement on the ischemic anterior wall at 120 kVp resulted in significantly lower mean TFR value as compared to the 70 keV TFR value (0.29±0.01 vs. 0.55±0.01; p<1e-05). No significant difference was measured between 120 kVp and 70 keV mean TFR values on segments moderately affected or unaffected by BH. In the entire ischemic LAD territory, 120 kVp mean endocardial flow was significantly reduced as compared to mean epicardial flow (15.80±10.98 vs. 40.85±23.44 ml/min/100g; p<1e-04). At 70 keV, BH was effectively minimized resulting in mean endocardial MBF of 40.85±15.3407 ml/min/100g vs. 74.09±5.07 ml/min/100g (p=0.0054) in the epicardium. We also found that BH artifact in the conventional 120 kVp images resulted in falsely reduced MBF measurements even under non-ischemic conditions.

Catheter ablation is a safe and effective therapy for drug-refractory patients symptomatic of atrial fibrillation (AF), with up to 80% of patients experiencing long-term arrhythmia-free survival. However, up to 20–40% of patients require more than one procedure in order to become arrhythmia-free. Therefore, appropriate patient selection is paramount to the effective implementation and long-term success of ablation therapy for patients with atrial fibrillation (AF). In this study, as a precursor to evaluating clinical significance of specific LA shape metrics as pre-procedural predictors of AF recurrence following ablative pulmonary vein isolation therapy, we report on a computational geometric analysis in a pilot cohort evaluating relationships between various patient-specific metrics of LA shape which might have such predictive value. This study specifically is focused on establishing the relationship between LA volume and sphericity, using a novel methodology for computing atrial sphericity based on regional shape.

Diffuse optical tomography (DOT) has attracted attentions in the last two decades due to its intrinsic sensitivity in imaging chromophores of tissues such as blood, water, and lipid. However, DOT has not been clinically accepted yet due to its low spatial resolution caused by strong optical scattering in tissues. Structural guidance provided by an anatomical imaging modality enhances the DOT imaging substantially. Here, we propose a computed tomography (CT) guided multispectral DOT imaging system for breast cancer detection. To validate its feasibility, we have built a prototype DOT imaging system which consists of a laser at wavelengths of 650 and an electron multiplying charge coupled device (EMCCD) camera. We have validated the CT guided DOT reconstruction algorithms with numerical simulations and phantom experiments, in which different imaging setup parameters, such as projection number of measurements, the width of measurement patch, have been investigated. Our results indicate that an EMCCD camera with air cooling is good enough for the transmission mode DOT imaging. We have also found that measurements at six projections are sufficient for DOT to reconstruct the optical targets with 4 times absorption contrast when the CT guidance is applied. Finally, we report our effort and progress on the integration of the multispectral DOT imaging system into a breast CT scanner.

We created a metastasis imaging, analysis platform consisting of software and multi-spectral cryo-imaging system suitable for evaluating emerging imaging agents targeting micro-metastatic tumor. We analyzed CREKA-Gd in MRI, followed by cryo-imaging which repeatedly sectioned and tiled microscope images of the tissue block face, providing anatomical bright field and molecular fluorescence, enabling 3D microscopic imaging of the entire mouse with single metastatic cell sensitivity. To register MRI volumes to the cryo bright field reference, we used our standard mutual information, non-rigid registration which proceeded: preprocess → affine → B-spline non-rigid 3D registration. In this report, we created two modified approaches: mask where we registered locally over a smaller rectangular solid, and sliding organ. Briefly, in sliding organ, we segmented the organ, registered the organ and body volumes separately and combined results. Though sliding organ required manual annotation, it provided the best result as a standard to measure other registration methods. Regularization parameters for standard and mask methods were optimized in a grid search. Evaluations consisted of DICE, and visual scoring of a checkerboard display. Standard had accuracy of 2 voxels in all regions except near the kidney, where there were 5 voxels sliding. After mask and sliding organ correction, kidneys sliding were within 2 voxels, and Dice overlap increased 4%–10% in mask compared to standard. Mask generated comparable results with sliding organ and allowed a semi-automatic process.

Quantities of MRI data, much larger than can be objectively and efficiently analyzed manually, are routinely generated in preclinical research. We aim to develop an automated image segmentation and registration pipeline to aid in analysis of image data from our high-throughput 9.4 Tesla small animal MRI imaging center. T2-weighted, fat-suppressed MRIs were acquired over 4 life-cycle time-points [up to 12 to 18 weeks] of twelve C3(1) SV40 Large T-antigen mice for a total of 46 T2-weighted MRI volumes; each with a matrix size of 192 x 256, 62 slices, in plane resolution 0.1 mm, and slice thickness 0.5 mm. These image sets were acquired with the goal of tracking and quantifying progression of mammary intraepithelial neoplasia (MIN) to invasive cancer in mice, believed to be similar to ductal carcinoma in situ (DCIS) in humans. Our segmentation algorithm takes 2D seed-points drawn by the user at the center of the 4 co-registered volumes associated with each mouse. The level set then evolves in 3D from these 2D seeds. The contour evolution incorporates texture information, edge information, and a statistical shape model in a two-step process. Volumetric DICE coefficients comparing the automatic with manual segmentations were computed and ranged between 0.75 and 0.58 for averages over the 4 life-cycle time points of the mice. Incorporation of these personalized atlases with intra and inter mouse registration is expected to enable locally and globally tracking of the morphological and textural changes in the mammary tissue and associated lesions of these mice.

The characterization of tumors after being imaged is currently a qualitative process performed by skilled professionals. If we can aid their diagnosis by identifying quantifiable features associated with tumor classification, we may avoid invasive procedures such as biopsies and enhance efficiency. The aim of this paper is to describe the 3D EdgeRunner Pipeline which characterizes the shape of a tumor. Shape analysis is relevant as malignant tumors tend to be more lobular and benign ones tare generally more symmetrical. The method described considers the distance from each point on the edge of the tumor to the centre of a synthetically created field of view. The method then determines coordinates where the measured distances are rapidly changing (peaks) using a second derivative found by five point differentiation. The list of coordinates considered to be peaks can then be used as statistical data to compare tumors quantitatively. We have found this process effectively captures the peaks on a selection of kidney tumors.

Lung boundary image segmentation is important for many tasks including for example in development of radiation treatment plans for subjects with thoracic malignancies. In this paper, we describe a method and parameter settings for accurate 3D lung boundary segmentation based on graph-cuts from X-ray CT data¹. Even though previously several researchers have used graph-cuts for image segmentation, to date, no systematic studies have been performed regarding the range of parameter that give accurate results. The energy function in the graph-cuts algorithm requires 3 suitable parameter settings: K, a large constant for assigning seed points, c, the similarity coefficient for n-links, and λ, the terminal coefficient for t-links. We analyzed the parameter sensitivity with four lung data sets from subjects with lung cancer using error metrics. Large values of K created artifacts on segmented images, and relatively much larger value of c than the value of λ influenced the balance between the boundary term and the data term in the energy function, leading to unacceptable segmentation results. For a range of parameter settings, we performed 3D image segmentation, and in each case compared the results with the expert-delineated lung boundaries. We used simple 6-neighborhood systems for n-link in 3D. The 3D image segmentation took 10 minutes for a 512x512x118 ~ 512x512x190 lung CT image volume. Our results indicate that the graph-cuts algorithm was more sensitive to the K and λ parameter settings than to the C parameter and furthermore that amongst the range of parameters tested, K=5 and λ=0.5 yielded good results.

Traumatic meningeal injury is a novel imaging marker of traumatic brain injury, which appears as enhancement of the dura on post-contrast T2-weighted FLAIR images, and is likely associated with inflammation of the meninges. Dynamic Contrast Enhanced MRI provides a better discrimination of abnormally perfused regions. A method to properly identify those regions is presented. Images of seventeen patients scanned within 96 hours of head injury with positive traumatic meningeal injury were normalized and aligned. The difference between the pre- and last post-contrast acquisitions was segmented and voxels in the higher class were spatially clustered. Spatial and morphological descriptors were used to identify the regions of enhancement: a) centroid; b) distance to the brain mask from external voxels; c) distance from internal voxels; d) size; e) shape. The method properly identified thirteen regions among all patients. The method failed in one case due to the presence of a large brain lesion that altered the mask boundaries. Most false detections were correctly rejected resulting in a sensitivity and specificity of 92.9% and 93.6%, respectively.

We applied a modified probabilistic fiber-tracking method for the extraction of fiber pathways to quantify decreased white matter integrity as a surrogate of structural loss in connectivity due to cranial radiation therapy (CRT) as treatment for pediatric medulloblastoma. Thirty subjects were examined (n=8 average-risk, n=22 high-risk) and the groups did not differ significantly in age at examination. The pathway analysis created a structural connectome focused on sub-networks within the central executive network (CEN) for comparison between baseline and post-CRT scans and for comparison between standard and high dose CRT. A paired-wise comparison of the connectivity between baseline and post-CRT scans showed the irradiation did have a significant detrimental impact on white matter integrity (decreased fractional anisotropy (FA) and decreased axial diffusivity (AX)) in most of the CEN sub-networks. Group comparisons of the change in the connectivity revealed that patients receiving high dose CRT experienced significant AX decreases in all sub-networks while the patients receiving standard dose CRT had relatively stable AX measures across time. This study on pediatric patients with medulloblastoma demonstrated the utility of this method to identify specific sub-networks within the developing brain affected by CRT.

In this paper, a principal axes registration technique is presented, with the relevant application to segmented volumes. The purpose of the proposed registration is to compare multi-temporal volumes of carpal bones from Magnetic Resonance Imaging (MRI) acquisitions. Starting from the study of the second-order moment matrix, the eigenvectors are calculated to allow the rotation of volumes with respect to reference axes. Then the volumes are spatially translated to become perfectly overlapped.

A quantitative evaluation of the results obtained is carried out by computing classical indices from the confusion matrix, which depict similarity measures between the volumes of the same organ as extracted from MRI acquisitions executed at different moments. Within the medical field, the way a registration can be used to compare multi-temporal images is of great interest, since it provides the physician with a tool which allows a visual monitoring of a disease evolution.

The segmentation method used herein is based on the graph theory and is a robust, unsupervised and parameters independent method. Patients affected by rheumatic diseases have been considered.

Tract-based spatial statistics (TBSS)⁶ is a software pipeline widely employed in comparative analysis of the white matter integrity from diffusion tensor imaging (DTI) datasets. In this study, we seek to evaluate the relationship between different methods of atlas registration for use with TBSS and different measurements of DTI (fractional anisotropy, FA, axial diffusivity, AD, radial diffusivity, RD, and medial diffusivity, MD). To do so, we have developed a novel tool that builds on existing diffusion atlas building software, integrating it into an adapted version of TBSS called DAB-TBSS (DTI Atlas Builder-Tract-Based Spatial Statistics) by using the advanced registration offered in DTI Atlas Builder⁷. To compare the effectiveness of these two versions of TBSS, we also propose a framework for simulating population differences for diffusion tensor imaging data, providing a more substantive means of empirically comparing DTI group analysis programs such as TBSS. In this study, we used 33 diffusion tensor imaging datasets and simulated group-wise changes in this data by increasing, in three different simulations, the principal eigenvalue (directly altering AD), the second and third eigenvalues (RD), and all three eigenvalues (MD) in the genu, the right uncinate fasciculus, and the left IFO. Additionally, we assessed the benefits of comparing the tensors directly using a functional analysis of diffusion tensor tract statistics (FADTTS¹⁰). Our results indicate comparable levels of FA-based detection between DAB-TBSS and TBSS, with standard TBSS registration reporting a higher rate of false positives in other measurements of DTI. Within the simulated changes investigated here, this study suggests that the use of DTI Atlas Builder’s registration enhances TBSS group-based studies.

There are evidences in the literature that Temporal Lobe Epilepsy (TLE) causes some lateralized atrophy and deformation on hippocampus and other substructures of the brain. Magnetic Resonance Imaging (MRI), due to high-contrast soft tissue imaging, is one of the most popular imaging modalities being used in TLE diagnosis and treatment procedures. Using an algorithm to help clinicians for better and more effective shape deformations analysis could improve the diagnosis and treatment of the disease. In this project our purpose is to design, implement and test a classification algorithm for MRIs based on hippocampal asymmetry detection using shape and size-based features. Our method consisted of two main parts; (1) shape feature extraction, and (2) image classification. We tested 11 different shape and size features and selected four of them that detect the asymmetry in hippocampus significantly in a randomly selected subset of the dataset. Then, we employed a support vector machine (SVM) classifier to classify the remaining images of the dataset to normal and epileptic images using our selected features. The dataset contains 25 patient images in which 12 cases were used as a training set and the rest 13 cases for testing the performance of classifier. We measured accuracy, specificity and sensitivity of, respectively, 76%, 100%, and 70% for our algorithm. The preliminary results show that using shape and size features for detecting hippocampal asymmetry could be helpful in TLE diagnosis in MRI.

It is now common for magnetic-resonance-imaging (MRI) based multi-site trials to include diffusion-weighted imaging (DWI) as part of the protocol. It is also common for these sites to possess MR scanners of different manufacturers, different software and hardware, and different software licenses. These differences mean that scanners may not be able to acquire data with the same number of gradient amplitude values and number of available gradient directions. Variability can also occur in achievable b-values and minimum echo times. The challenge of a multi-site study then, is to create a common protocol by understanding and then minimizing the effects of scanner variability and identifying reliable and accurate diffusion metrics. This study describes the effect of site, scanner vendor, field strength, and TE on two diffusion metrics: the first moment of the diffusion tensor field (mean diffusivity, MD), and the fractional anisotropy (FA) using two common analyses (region-of-interest and mean-bin value of whole brain histograms). The goal of the study was to identify sources of variability in diffusion-sensitized imaging and their influence on commonly reported metrics. The results demonstrate that the site, vendor, field strength, and echo time all contribute to variability in FA and MD, though to different extent. We conclude that characterization of the variability of DTI metrics due to site, vendor, field strength, and echo time is a worthwhile step in the construction of multi-center trials.