Although multi-modal imaging tends to improve the segmentation and classification performance in the field of medical image processing, lacking certain modalities at test time limits its clinical applicability. In this paper, we explored the ability of cross-modal distillation for increasing the performance of T1w MRI-based brain tumor segmentation. More specifically, we considered having high resolution T1w and T2w MRI sequences available for training while having only a high resolution T1w MRI sequence available at test time. We investigated the efficacy of the proposed method to improve the whole tumor segmentation using the BRATS 2018 dataset. Both cross-modal knowledge distillation and cross-modal feature distillation approaches were confirmed to enrich the representation of the T1w MRI sequence by learning from the representation of the more informative T2w MRI sequence during training, thereby improving the mean Dice scores by 6.14 % and 7.02 %, respectively.
In this paper, we will describe a theoretical model of the spatial uncertainty for a line of response, due to the imperfect localization of events on the detector heads of the Positron Emission Tomography (PET) camera. We assume a Gaussian distribution of the position of interaction on a detector head, centered at the measured position. The probability that an event originates from a certain point in the FOV is calculated by integrating all the possible LORs through this point, weighted with the Gaussian probability of detection at the LORs end points. We have calculated these probabilities both for perpendicular and oblique coincidences. For the oblique coincidence case it was necessary to incorporate the effect of the crystal thickness in the calculations. We found that the probability function can not be analytically expressed in a closed form, and it was thus calculated by means of numerical integration. A Gaussian was fitted to the probability profiles for a given distance to the detectors. From these fits, we can conclude that the profiles can be accurately approximated by a Gaussian, both for perpendicular as for oblique coincidences. The FWHM reaches a maximum at the detector heads, and decreases towards the center of the FOV, as was expected.
The accurate quantification of brain perfusion for emission computed tomography data (PET-SPECT) is limited by partial volume effects (PVE). This study presents a new approach to estimate accurately the true tissue tracer activity within the grey matter tissue compartment. The methodology is based on the availability of additional anatomical side information and on the assumption that activity concentration within the white matter tissue compartment is constant. Starting from an initial estimate for the white matter grey matter activity, the true tracer activity within the grey matter tissue compartment is estimated by an alternating ML-EM-algorithm. During the updating step the constant activity concentration within the white matter compartment is modelled in the forward projection in order to reconstruct the true activity distribution within the grey matter tissue compartment, hence reducing partial volume averaging. Consequently the estimate for the constant activity in the white matter tissue compartment is updated based on the new estimated activity distribution in the grey matter tissue compartment. We have tested this methodology by means of computer simulations. A T1-weighted MR brainscan of a patient was segmented into white matter, grey matter and cerebrospinal fluid, using the segmentation package of the SPM-software (Statistical Parametric Mapping). The segmented grey and white matter were used to simulate a SPECT acquisition, modelling the noise and the distance dependant detector response. Scatter and attenuation were ignored. Following the above described strategy, simulations have shown it is possible to reconstruct the true activity distribution for the grey matter tissue compartment (activity/tissue volume), assuming constant activity in the white matter tissue compartment.
KEYWORDS: Brain, Photons, Data acquisition, Single photon emission computed tomography, Cameras, Imaging systems, Neuroimaging, Monte Carlo methods, Data corrections, Signal attenuation
A practical method for scatter compensation in SPECT imaging is the triple energy window technique (TEW) which estimates the fraction of scattered photons in the projection data pixel by pixel. This technique requires an acquisition of counts in three windows of the energy spectrum for each projection bin, which is not possible on every gamma camera. The aim of this study is to set up a scatter template for brain perfusion SPECT imaging by means of the scatter data acquired with the triple energy window technique. This scatter template can be used for scatter correction as follows: the scatter template is realigned with the acquired, by scatter degraded and reconstructed image by means of the corresponding emission template, which also includes scatter counts. The ratios between the voxel values of this emission template and the acquired and reconstructed image are used to locally adjust the scatter template. Finally the acquired and reconstructed image is corrected for scatter by subtracting the thus obtained scatter estimates. We compared the template based approach with the TEW scatter correction technique for data acquired with same gamma camera system and found a similar performance for both correction methods.
Simulations and measurements of triple head PET acquisitions of a hot sphere phantom were performed to evaluate the performance of two different reconstruction algorithms (projection based ML-EM and listmode ML-EM)for triple head gamma camera coincidence systems. A geometric simulator assuming a detector with 100 percent detection efficiency and only detection of trues was used. The resolution was equal to the camera system. The measurements were performed with a triple headed gamma camera. Simulated and measured data were stored in listmode format, which allowed the flexibility for different reconstruction algorithms. As a measure for the performance the hot spot detectability was taken because tumor imaging is the most important clinical application for gamma camera coincidence systems. The detectability was evaluated by calculating the recovered contrast and the contrast-to-noise ratio. Results show a slightly improved contrast but a clearly higher contrast-to-noise ratio for list mode reconstruction.
Gamma camera PET (Positron Emission Tomography) offers a low-cost alternative for dedicated PET scanners. However, sensitivity and count rate capabilities of dual-headed gamma cameras with PET capabilities are still limited compared to full-ring dedicated PET scanners. To improve the geometric sensitivity of these systems, triple-headed gamma camera PET has been proposed. As is the case for dual-headed PET, the sensitivity of these devices varies with the position within the field of view (FOV) of the camera. This variation should be corrected for when reconstructing the images. In earlier work, we calculated the two-dimensional sensitivity variation for any triple-headed configuration. This can be used to correct the data if the acquisition is done using axial filters, which effectively limit the axial angle of incidence of the photons, comparable to 2D dedicated PET. More recently, these results were extended to a fully 3D calculation of the geometric sensitivity variation. In this work, the results of these calculations are compared to the standard approach to correct for 3D geometric sensitivity variation. Current implementations of triple-headed gamma camera PET use two independent corrections to account for three-dimensional sensitivity variations: one in the transaxial direction and one in the axial direction. This approach implicitly assumes that the actual variation is separable in two independent components. We recently derived a theoretical expression for the 3D sensitivity variation, and in this work we investigate the separability of our result. To investigate the separability of the sensitivity variations, an axial and transaxial profile through the calculated variation was taken, and these two were multiplied, thus creating a separable function. If the variation were perfectly separable, this function would be identical to the calculated variation. As a measure of separability, we calculated the percentual deviation of the separable function to the original variation. We investigated the separability for several camera configurations and rotation radii. We found that, for all configurations, the variation is not separable , and becomes less separable as the rotation radius tends to smaller values. This indicates that in this case, our sensitivity correction will give better results than the separable correction now applied.
The 3D acquisition data from positron coincidence detection on a gamma camera, can be stored in list-mode or histogram format. The standard processing of the list mode-data is Single Slice Rebinning (with a maximum acceptance angle) to 2D histogrammed projections followed by Ordered Subsets Expectation Maximization reconstruction. This method has several disadvantages: sampling accuracy is lost by histogramming events, axial resolution degrades with increasing distance from the center of rotation and useful events, with angle bigger than the acceptance angle, are not included in the reconstruction. Therefore an iterative reconstruction algorithm, operating directly on list-mode data, has been implemented. The 2D and 3D version of this iterative list-mode algorithm have been compared with the aforementioned standard reconstruction method. A higher number of events is used in the reconstruction, which results in a lower standard deviation. Resolution is fairly constant over the Field of View. The use of a fast projector and backprojector reduces the reconstruction time to clinical acceptable times.
In the near future, it will be possible to perform coincidence detection on a gamma camera with three heads, which increases the geometric sensitivity of the system. Different geometric configurations are possible, and each configuration yields a different geometric sensitivity. The purpose of this work was to calculate the sensitivities for different three-headed configurations as a function of the position in the field of view, the dimensions of the detector heads and the distance of the heads from the center of the field of view. The configurations that were compared are: a regular two headed configuration (180 deg. opposed), a triple-headed configuration with the three heads in an equilateral triangle (120 deg.), and a triple-headed configuration with two heads in a regular two headed configuration, and the third perpendicular between the first two, which makes a U-shaped configuration. An expression was derived for any planar detector configuration to calculate the geometric sensitivity for each Line Of Response (LOR). This sensitivity was integrated to get the sensitivity profile, which gives the geometric sensitivity at a certain distance from the center of rotation. We found that the triangular configuration gave the best sensitivities when placed very near to each other (nearly full ring configuration), but for larger fields of view, the U-shaped configuration performed better.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.