Three-dimensional lesion segmentation is required for analysis of radiomic features and lesion growth kinetics. In clinical trials, radiologists apply the Response Evaluation Criteria in Solid Tumors (RECIST), by manually annotating the long and short diameters of a lesion on a single 2D axial slice (RECIST slice), where the lesion looks largest. We developed a novel approach that leverages the RECIST annotations to segment lesions in 3D on CT scans. We start with bounding box and center point prompts derived from RECIST long and short diameters on RECIST slice. Iteratively, we perform prompted segmentation using Segment Anything Model (SAM) on off-RECIST slices towards the superior and inferior direction until all slices are segmented. To optimize the performance of SAM, we fine-tuned the mask decoder. In addition, it is crucial to detect where the lesions disappear at the superior and inferior direction to prevent over segmentation. We developed a multi-task framework for lesion existence classification and segmentation and further compared the parallel framework and cascaded framework. We used an internal dataset consisting of 2053 and 200 3D lesions for fine-tuning of SAM decoder and testing, respectively. Baseline SAM, SAM with fine tuning, SAM with parallel multi-task fine tuning, and SAM with cascaded multitask fine tuning have Dice scores of 0.4745±0.2138, 0.7136±0.1277, 0.6985±0.1312, and 0.7239±0.1321, respectively. Our experiments proves that multi-task learning is an effective way for 3D segmentation with SAM, and cascaded framework performs better than parallel framework.
The ability to evaluate tumor oxygenation in the clinic could indicate prognosis and enable treatment monitoring, since oxygen deficient cancer cells are often more resistant to chemotherapy and radiotherapy. MultiSpectral Optoacoustic Tomography (MSOT) is a hybrid technique combining the high contrast of optical imaging with spatial resolution and penetration depth similar to ultrasound. We hypothesized that MSOT could reveal both tumor vascular density and function based on modulation of blood oxygenation.
We performed MSOT on nude mice (n=8) bearing subcutaneous xenograft PC3 tumors using an inVision 256 (iThera Medical). The mice were maintained under inhalation anesthesia during imaging and respired oxygen content was modified from 21% to 100% and back. After imaging, Hoechst 33348 was injected to indicate vascular perfusion and permeability. Tumors were then extracted for histopathological analysis and fluorescence microscopy. The acquired data was analyzed to extract a bulk measurement of blood oxygenation (SO2MSOT) from the whole tumor using different approaches. The tumors were also automatically segmented into 5 regions to investigate the effect of depth on SO2MSOT.
Baseline SO2MSOT values at 21% and 100% oxygen breathing showed no relationship with ex vivo measures of vascular density or function, while the change in SO2MSOT showed a strong negative correlation to Hoechst intensity (r=- 0.92, p=0.0016). Tumor voxels responding to oxygen challenge were spatially heterogeneous. We observed a significant drop in SO2 MSOT value with tumor depth following a switch of respiratory gas from air to oxygen (0.323±0.017 vs. 0.11±0.05, p=0.009 between 0 and 1.5mm depth), but no such effect for air breathing (0.265±0.013 vs. 0.19±0.04, p=0.14 between 0 and 1.5mm depth).
Our results indicate that in subcutaneous prostate tumors, baseline SO2MSOT levels do not correlate to tumor vascular density or function while the magnitude of the response to oxygen challenge provides insight into these parameters. Future work will include validation using in vivo imaging and protocol optimization for clinical application.
Optoacoustic Tomography is a fast developing imaging modality, combining the high resolution and penetration depth of ultrasound detection with the high contrast available from optical absorption in tissue. The spectral profile of near infrared excitation light used in optoacoustic tomography instruments is modified by absorption and scattering as it propagates deep into biological tissue. The resulting images therefore provide only qualitative insight into the distribution of tissue chromophores. Knowledge of the spectral profile of excitation light across the mouse is needed for accurate determination of the absorption coefficient in vivo. Under the conditions of constant Grueneisen parameter and accurate knowledge of the light fluence, a linear relationship should exist between the initial optoacoustic pressure amplitude and the tissue absorption coefficient. Using data from a commercial optoacoustic tomography system, we implemented an iterative optimization based on the σ-Eddington approximation to the Radiative Transfer Equation to derive a light fluence map within a given object. We segmented the images based on the positions of phantom inclusions, or mouse organs, and used known scattering coefficients for initialization. Performing the fluence correction in simple phantoms allowed the expected linear relationship between recorded and independently measured absorption coefficients to be retrieved and spectral coloring to be compensated. For in vivo data, the correction resulted in an enhancement of signal intensities in deep tissues. This improved our ability to visualize organs at depth (> 5mm). Future work will aim to perform the optimization without data normalization and explore the need for methodology that enables routine implementation for in vivo imaging.
The ability to evaluate tumor oxygenation in the clinic could indicate prognosis and enable treatment monitoring, since oxygen deficient cancer cells are more resistant to chemotherapy and radiotherapy. MultiSpectral Optoacoustic Tomography (MSOT) is a hybrid technique combining the high contrast of optical imaging with the spatial resolution and penetration depth similar to ultrasound. We aim to demonstrate that MSOT can be used to monitor the development of tumor vasculature.
To establish the relationship between MSOT derived imaging biomarkers and biological changes during tumor development, we performed MSOT on nude mice (n=10) bearing subcutaneous xenograft U87 glioblastoma tumors using a small animal optoacoustic tomography system. The mice were maintained under inhalation anesthesia during imaging and respired oxygen content was modified between 21% and 100%. The measurements from early (week 4) and late (week 7) stages of tumor development were compared. To further explore the functionality of the blood vessels, we examined the evolution of changes in the abundance of oxy- and deoxyhemoglobin in the tumors in response to a gas challenge. We found that the kinetics of the change in oxygen saturation (SO2) were significantly different between small tumors and the healthy blood vessels in nearby normal tissue (p=0.0054). Furthermore, we showed that there was a significant difference in the kinetics of the gas challenge between small and large tumors (p=0.0015). We also found that the tumor SO2 was significantly correlated (p=0.0057) with the tumor necrotic fraction as assessed by H&E staining in histology. In the future, this approach may be of use in the clinic as a method for tumor staging and assessment of treatment response.
MultiSpectral Optoacoustic Tomography (MSOT) is a fast developing imaging modality, combining the high resolution and penetration depth of ultrasound with the excellent contrast from optical imaging of tissue. Absorption and scattering of the near infrared excitation light modulates the spectral profile of light as it propagates deep into biological tissue, meaning the images obtained provide only qualitative insight into the distribution of tissue chromophores. The goal of this work is to accurately recover the spectral profile of excitation light by modelling light fluence in the data reconstruction, to enable quantitative imaging. We worked with a commercial small animal MSOT scanner and developed our light fluence correction for its' cylindrical geometry. Optoacoustic image reconstruction pinpoints the sources of acoustic waves detected by the transducers and returns the initial pressure amplitude at these points. This pressure is the product of the dimensionless Grüneisen parameter, the absorption coefficient and the light fluence. Under the condition of constant Grüneisen parameter and well modelled light fluence, there is a linear relationship between the initial pressure amplitude measured in the optoacoustic image and the absorption coefficient. We were able to reproduce this linear relationship in different physical regions of an agarose gel phantom containing targets of known optical absorption coefficient, demonstrating that our light fluence model was working. We also demonstrate promising results of light fluence correction effects on in vivo data.
MultiSpectral optoacoustic tomography (MSOT) is an emerging modality that combines the high contrast of optical imaging with the spatial resolution and penetration depth of ultrasound, to provide detailed images of hemoglobin concentration and oxygenation. To facilitate accurate determination of changes in the vascularity and oxygenation of a biological tissue over time, a tumor in response to cancer therapy for example, an extensive study of stability and reproducibility of a small animal MSOT system has been performed. Investigations were first made with a stable phantom imaged repeatedly over time scales of hours, days and months to evaluate the reproducibility of the system over time. We found that the small animal MSOT system exhibited excellent reproducibility with a coefficient of variation (COV) in the measured MSOT signals of less than 8% over the course of 30 days and within 1.5% over a single day. Experiments performed in vivo demonstrated the potential for measurement of oxyhemoglobin over time in a realistic experimental setting. The effect of breathing medical air or oxygen under conditions of fixed respiration rate and body temperature within normal organs, including the spleen and kidneys, were investigated. The COV for oxyhemoglobin signals retrieved from spectral unmixing was assessed within both biological (different mouse) and imaging (different scan) replicates. As expected, biological replicates produced a large COV (up to 40% within the spleen) compared to imaging replicates within a single mouse (up to 10% within the spleen). Furthermore, no significant difference was found between data acquired by different operators. The data presented here suggest that MSOT is highly reproducible for both phantom and in vivo imaging, hence could reliably detect changes in oxygenation occurring in living subjects.
KEYWORDS: Cameras, Calibration, Digital cameras, 3D modeling, Californium, Information technology, Reconstruction algorithms, Electronics, Photonics, Astronomy
This paper presents an algorithm for camera calibration, applying digital images to calculate camera parameters, position and orientation. A linear decomposition technique is proposed to solve nonlinear pixel equations in which camera parameters are involved.
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.