2024 Program with Abstracts

Session 3B: Imaging - Campania A

le 7 juin 2024 from 13h00 CDT to 14h30 CDT

Scientific Session 3B – Imaging
Friday, June 7, 2024, 13:00-14:30

Scientific Session 3B:Imaging– Presentation 1

Measuring radiation survival of prostate tumor spheroids with dynamic optical coherence tomography (dOCT).

Stephanie Swanson, Keyu Chen, Elahe Cheraghi, Ernest Osei, Kostadinka Bizheva
University of Waterloo, Grand River Regional Cancer Centre

Purpose: Clonogenic assay is the gold standard method for measuring in vitro cell survival fraction following radiation treatment; however, it requires 3D tumor spheroids to be disaggregated, plated in 2D, and observed for multiple weeks. This study used dynamic optical coherence tomography (dOCT) to non-invasively estimate the spatial and fractional radiobiological survival of prostate tumor spheroids after radiation treatment.

Methods: A Varian TrueBeam linear accelerator was used to deliver various doses of 6 MV radiation to spheroids 24 hours after seeding 1,000 PC3 prostate cancer cells per spheroid. Survival fraction was estimated by imaging the spheroids with dOCT and performing clonogenic assay 24 hours after irradiation. Spatial and fractional survival measured by dOCT was compared with 3D agent-based computer simulation of tumor spheroid growth and irradiation that probabilistically employed survival fraction measured by clonogenic assay.

Results: With increasing radiation dose, dOCT images of spheroids demonstrated decreasing survival and agreed well with simulation. Clonogenic assay estimated much lower survival fraction than dOCT and simulation, likely by including cell death that occurs after dOCT imaging, which estimates survival with instantaneous cellular activity


Scientific Session 3B:Imaging – Presentation 2

Enhancing heart contrast in cine EPID images acquired during DIBH Breast Hybrid IMRT treatments.

Jonathan Redekopp, Jorge, Alpuche Aviles
University of Manitoba

Purpose: To develop a method of enhancing the contrast between the heart edge and lung in EPID images from Deep Inspiration Breath Hold (DIBH) breast radiotherapy treatments.

Methods: Cine EPID images from breast tangents delivered during DIBH were studied. To enhance the heart edge, the power spectrum of each pixel’s variation during beam delivery is calculated. By integrating select frequency components and displaying them in an image, oscillations corresponding to the patient’s heartbeat can be highlighted. EPID images were generated at ~2 Hz and frequency components between 0 Hz and the Nyquist frequency of 1 Hz were integrated to produce the enhanced images. The enhanced images were evaluated against the original EPID images by calculating their respective contrast to noise ratios (CNR) between heart edge and lung pixels.

Results: The contrast between the heart edge and lung was increased for all tested images by a factor of 1.3 – 4.7. The CNR in the unenhanced EPID images ranged from 0.29 to 2.22 while the CNR in the enhanced images ranged from 1.59 to 8.78. The increase in contrast was sufficient for the heart edge to be successfully identified by a reviewer in all enhanced images.

Conclusions: A method for highlighting the heart edge in EPID images from DIBH breast tangents was tested. Differences in the component frequencies of heart edge pixels and lung pixels were successfully used to enhance contrast. More testing is needed to further refine this method and automate the detection of heart edge pixels in frequency images.


Scientific Session 3B:Imaging – Presentation 3

Investigating the spatiotemporal neuroinflammatory response of half-brain irradiation in a murine model using 18F-FEPPA PET.

Sawyer Badiuk, Lise Desjardins, Matthew Fox, Paula Foster, Jonathan Thiessen, Jeff Chen, Eugene Wong
University of Western Ontario, Lawson Health Research institute

Purpose: The aim of this study is to investigate if we can image the neuroinflammatory response from regional brain irradiation using 18F-FEPPA-PET to measure glial activation.

Methods: There are approximately four times as many glial cells in the brain as neurons. Inflammation is initiated when glial cells activate. Half brain irradiation was performed on mice(BALB/c) using a micro-CT/RT system with sham, 4Gy and 12Gy in one fraction. Dynamic 18F-FEPPA-PET was acquired for 90 minutes at 48 hours, 2 weeks, and 4 weeks post-irradiation to quantify level and duration of glial activation. PET kinetic analysis was completed for time activity curves(TACs) and volume of distribution(VT). Immunohistochemistry identifying translocator proteins(TSPO) the specific ligand for 18F-FEPPA was completed.

Results: In all treated mice, the irradiated hemisphere of the brain showed similar 18F-FEPPA uptake as the unirradiated hemisphere; suggesting partial brain irradiation triggers global inflammation in the brain at 48 hours, 2 weeks and 4 weeks. Immunofluorescence histological staining for TSPO also exhibited similar signal between the two hemispheres, confirming our PET imaging findings. In sham and treated mice, we observed that different brain regions exhibited different intensities of TSPO staining which is less obvious in 18F-FEPPA-PET due to PET’s inherent limit in spatial resolution.

Conclusion: This study showed global inflammation with half brain irradiation. Further PET and histological glial activation immunofluorescences data collection and analysis is on-going. Following this, half brain irradiation will be investigated in a breast cancer brain metastasis model to provide a comprehensive understanding of radiation and subsequent glial activation.


Scientific Session 3B:Imaging – Presentation 4

Enhanced Physics Modelling in Radar-based Microwave Imaging for Breast Cancer Detection.

Tyson Reimer, Spencer, Christie,Stephen, Pistorius
University of Manitoba

Purpose: Breast microwave imaging (BMI) is a potential breast cancer screening method, motivated by the relatively large contrast in the microwave properties of malignant and healthy breast tissues. BMI is limited by current image reconstruction methods, which use simplified physics models and have produced low diagnostic specificity estimates of 20-65%. This research aims to develop a more accurate reconstruction method through enhanced physics modelling (EPM). We hypothesize that more accurate physics models will improve the diagnostic accuracy of the technique.

Methods: This work used EPM factors to model the antenna’s beam pattern, frequency-dependent gain, and the distance-based attenuation factor due to the spherical waves produced by scattering points in the breast. Our BMI system was used with MRI-derived phantoms to produce a dataset for the evaluation of the diagnostic performance with respect to tumour detection. Receiver operating characteristic (ROC) curves were produced with a literature standard reconstruction method and the optimization-based approach with and without EPM.

Results: The use of EPM improved the area under the curve (AUC) of the ROC curves. The AUC obtained using EPM was (80 ± 1)%, compared to (64 ± 1)% with the literature standard reconstruction method.

Conclusions: The results obtained in this work support our hypothesis and demonstrate that improved physics models in image reconstruction can result in improved diagnostic performance in BMI. Future work will incorporate additional, more complex physics components. This project presents the first step toward physics-informed image reconstruction in BMI to improve the diagnostic accuracy of the modality.


Scientific Session 3B:Imaging– Presentation 5

Surface Guided Radiation Therapy with IDENTIFY vs Cone Beam CT for Intra-fraction motion assessment of Stereotactic Radiation Therapy of Brain Metastases treated with Hyperarcg multi-detector approach.

Claudia Mendez, Ravi Kambo, Lindsay Mathew, Shrinivas Rathod, Fred Hsu, Waseem Sharief
BC Cancer

Purpose: Varian’s surface guided radiation therapy (SGRT) solution, IDENTIFYTM, is highly utilized, yet its relationship to Varian’s image-guided radiation therapy is not fully understood. The purpose of this study was to establish the level of agreement between IDENTIFY and cone-beam CT (CBCT) data acquired pre- and post-treatment.

Methods: Stereotactic brain patients were set-up using IDENTIFY, pre-treatment CBCT was acquired for image matching, then surface monitoring was reset to monitor the patient’s intra-fraction motion in the treatment position. A post-treatment CBCT was acquired for verification of patient positioning. The difference between pre- and post-treatment CBCT deviations and IDENTIFY offsets throughout the treatment were calculated in 6 degrees of freedom. SGRT data was analyzed for trends, maximum, and average deviations for each couch angle.

Results: In total 44 CBCT images and 22 IDENTIFY datasets were analyzed for 7 patients. The mean difference in isocentre offsets of pre-treatment CBCT and IDENTIFY were < 0.01 mm and < 0.14 degrees for translational and rotational shifts. For post-treatment CBCT verification, differences were < 0.02 mm for translational and < 0.13 degrees for rotational shifts. Intrafraction data showed average translational offsets below 0.04, -0.4, and 0.2 mm for couch at 0, 45 and 315 degrees, respectively. Maximum deviations from the mean were 0.3 mm and 0.1 degrees for couch at 45 degrees.

Conclusions: There is excellent agreement between IDENTIFY’s SGRT data and pre- and post-treatment CBCT data. This work shows that the system has sub-millimeter accuracy but is subject to noise in the 0.3 mm range.

Scientific Session 3B:Imaging – Presentation 6

Initial performance evaluation of helical and volumetric dual energy CT acquisitions using the Canon Aquilion Exceed LB.

Samuel Blake, Timothy Yau, Hatem Mehrez, Brandon Disher, Stewart Gaede,
London Health Sciences Centre, Canon Medical Systems Canada

Purpose: The wide-bore Canon Aquilion Exceed LB is capable of helical and volumetric dual-energy data acquisition. CT number linearity and image quality were evaluated for both dual-energy modes and compared to 120 kVp single-energy helical mode used for radiotherapy simulation.

Methods: Tissue characterization, image quality and anthropomorphic phantoms were imaged using single-energy 120 kVp (SE-120), dual-energy 80/135 kVp helical (DEHEL) and volumetric (DEVOL) modes. Data was reconstructed using Adaptive Iterative Dose Reduction 3D. Synthetic 120 kV-equivalent images were reconstructed from the DEHEL (DEHEL-120) and DEVOL (DEVOL-120) data. CT numbers from tissue-mimicking materials and image quality (noise, spatial resolution and low (1%) contrast visibility metrics) were compared across SE-120, DEHEL-120 and DEVOL-120 images.

Results: CT numbers from SE-120, DEHEL-120 and DEVOL-120 images agreed for all tissue-mimicking inserts (maximum differences from nominal values were 5.3, 7.8 and 21.3 HU, respectively). Noise (standard deviation within a uniform region) was 8.9, 13.2 and 8.0 HU for the SE-120, DEHEL-120 and DEVOL-120 images, respectively. Spatial resolution at 80% and 50% of the relative modulation transfer function was 1.7 and 2.6; 1.8 and 2.7; 1.6 and 2.5 lp/cm for the SE-120, DEHEL-120 and DEVOL-120 images, respectively. Low contrast visibility for the 15 and 9 mm inserts were 0.19 and 0.14; 0.16 and 0.10; 0.27 and 0.22 for the SE-120, DEHEL-120 and DEVOL-120 images, respectively.

Conclusions: Dual-energy synthetic helical and volumetric 120 kV-equivalent images exhibited comparable CT number linearity and image quality relative to single-energy helical 120 kVp images acquired on the Canon Aquilion Exceed LB.


Scientific Session 3B:Imaging – Presentation 7

Stereoscopic and monoscopic real-time x-ray image guidance for markerless lung tumor tracking.

Zakary McLure, Mike Sattarivand
Dalhousie University

Purpose: The motion of lung tumors during breathing poses challenges in stereotactic body radiotherapy, warranting improved monitoring techniques. This study aims to combine dual-energy imaging bone suppression with ExacTrac's potential for stereoscopic and monoscopic 3D localization to develop a non-invasive tracking technique for precise lung tumor motion monitoring.

Methods: A motorized programmable breathing phantom combined with an anthropomorphic phantom simulated a lung tumor's respiratory motion, with various tumor models (6.4-25.4 mm). The ExacTrac stereoscopic system acquired real-time images at high and low energies (140 & 60 kVp). Weighted logarithmic subtraction and an anti-correlated noise reduction algorithm generated dual-energy images. Conventional single-energy images (120 kVp) were acquired for comparison. Digital reconstructed radiographs of the tumor served as templates for template-matching algorithms. In the monoscopic case, when only one x-ray view was available, a probability density function (PDF) utilized correlations between 3D motions for 3D position estimation.

Results: Dual-energy techniques demonstrated sub-millimeter accuracy, especially where the tumor was obstructed only by the ribcage. The view obstructed by the spinal cord showed lower success rates. Dual-energy consistently outperformed single-energy methods, successfully localizing a minimum of 92% of positions for all tumor sizes, compared to a 50% minimum for single energy.

Conclusions: Stereoscopic/monoscopic tumor tracking indicates that dual energy significantly increases accuracy of successful tumor localization as compared to the conventional single energy approach, especially for smaller tumors. The use of PDFs may be a viable approach to monoscopic estimates when only one view is available.


Scientific Session 3B:Imaging – Presentation 8

Toward Magnetic Resonance Oximetry: Development of Multi-Parametric Mapping in Fat-Water Mixtures.

Jorge Campos, Renée-Claude Bider, Véronique Fortier, Cristian Ciobanu, Marc-Antoine,Fortin, Ives Levesque
McGill University

Purpose: Investigation of two MRI-based methods to jointly measure magnetic resonance (MR) relaxation (R1f, R1w, R2*), and proton density fat fraction (PDFF), aiming towards tissue oximetry.

Introduction: Hypoxia is a marker of treatment outcome and tumor progression. MR-oximetry enables non-invasive mapping of blood and tissue oxygenation by measuring changes in the relaxation rates R2* and R1, respectively. The sensitivity of R1 to dissolved oxygen in tissue with sufficient fat content can be increased by measuring fat-specific R1 (R¬1f), while water-specific R1 (R1w) can be used in tissue with low fat content. Tissue fat content can be measured through the PDFF.

Methods: Two MR-oximetry approaches were tested for measurement of R1f, R1w, R2*, and PDFF: Fat Driven Equilibrium Single Pulse Observation of T1 (Fat DESPOT) and Multi-Echo Magnetization-Prepared 2-Rapid Gradient Echo (ME-MP2RAGE). We characterized the accuracy and precision of the methods and optimized the acquisition parameters using numerical simulations. We tested the feasibility of these methods in phantoms.

Results: Numerical simulations suggest that ME-MP2RAGE is more accurate and precise than Fat DESPOT. In phantoms, R1f, R1w, R2*, and PDFF measured by Fat DESPOT and ME-MP2RAGE agreed within 5%, 4%, 12% and 3%, respectively, with similar standard deviations. The divergence between simulation and experiment might be explained by mismatched SNR (lower in ME-MP2RAGE measurements). The larger difference in R2* is likely due to a difference in echo times between techniques.

Conclusion: Results in phantoms suggest that both ME-MP2RAGE and Fat DESPOT are appropriate techniques for MR-oximetry.