Poster Session and Exhibits - Campania B, Pre-Function Space, Trentino/Tuscany
le 6 juin 2024 from 18h00 CDT to 21h00 CDT
Scientific Session – Poster Session
Thursday, June 6, 2024, 17:30-20:00
Poster #1 - Characterization of a Dual-Energy CT Protocol for Radiation Treatment Planning
Presenting Author: Heather Young
Princess Margaret Cancer Centre
Purpose: Dual-energy CT (DECT) has many applications in radiation oncology including artifact reduction, iodine contrast visualization, and improved proton dosimetry. The differences between DECT and clinical single-energy CT (SECT) protocols must be characterized before DECT is used for CT simulation.
Methods: A Sun Nuclear multi-energy phantom and a Catphan 500 were imaged on a Canon Aquilion CT simulator using a clinical SECT (120 kV) protocol and two dose-matched DECT (80 and 135 kV) protocols in helical and volume mode which were evaluated using a 120 kV-equivalent scan. Analysis was completed using Matlab code developed in our institution. In the multi-energy phantom, the CT number of each density insert was compared across all scans. In the Catphan uniformity, high-contrast spatial resolution and low-contrast resolution were analyzed.
Results: High-density inserts had a higher CT number in the DECT images than the SECT images. For inserts with CT number below 200 HU, measured CT number agreed within 10 HU (1%) across all protocols. Uniformity was better in the SECT image (1.6 HU) than the helical (1.8 HU) and volume mode (2.4 HU) DECT images. High-contrast spatial resolution was higher in the SECT protocol (7.7 line-pairs/cm) than the helical (7.0 line-pairs/cm) and volume (5.3 line-pairs/cm) DECT images. Low-contrast resolution was visually decreased in the DECT images.
Conclusion: The DECT protocols studied here resulted in lower image quality than a dose-matched SECT protocol. More work is needed to determine the clinical and dosimetric impact of these differences and to optimize DECT protocols for CT simulation.
Poster #2 - Development of an AI-based pathway for breast radiotherapy
Presenting Author: Swapanpreet Kaur
Cancer Care Manitoba
Purpose: This work evaluates the feasibility of AI based contouring to be used to generate VMAT based treatment plans for left breast patients. This is part of a larger goal to develop a semi-automated process for VMAT breast radiation therapy, using AI based contouring, automated treatment plan creation, and automated plan verification.
Methods: Five previously treated left-sided breast patients with the appropriate contours were used in this work. AI contours for the heart, breasts, lungs and spinal cord were generated on the planning CT datasets using OrgansRT software. Contours for the nodal regions (internal mammary and supraclavicular/axillary) were created previously. The breast target volume was created using the OrgansRT breast contour cropped five millimeters from the skin, and expanded using a 5-millimeter PTV margin. Nodal targets were expanded using the same PTV margin. The five patients were then planned using a split field RapidArc technique. AI and manual contours were compared using a Dice coefficient. Various DVH parameters for targets and OARs were compared between RapidArc and hybrid IMRT plans.
Results and Conclusions: There is good agreement found between the contours in clinical plan and the ones generated by organsRT . Automated breast VMAT planning appears feasible, with OrgansRT generating target volumes that are similar to what we currently treat and plans that are preferable in some aspects. For patients with unfavorable anatomy significant increases in target coverage can be attained for modest increases in dose to other structures.
Poster #3 - 3D Conformal Cranial Spine Irradiation Using Field-in-Field Compensation of Overlapping
Presenting Author: Mei Xiangyang
Southeast Regional Cancer Program
Purpose: The 3D conformal technique for cranial spine irradiation (CSI) is still effective due to efficient delivery compared to more complex VMAT/IMRT techniques. However, multiple matching field-borders (MMFB) (jaw based) and multiple plans with spatial junction shifts over the course of treatment are required to reduce dose delivery uncertainties. We demonstrate a better 3D conformal CSI planning technique using MLC based field-in-field compensation of overlapping fields (FIF-COF) in one plan.
Methods: A typical clinical MMFB CSI plan uses 3-isocentres with field boarders matched to the beam divergences at the junctions of the corresponding jaw defined fields, and 3 plans matched at 3 junction shift positions (1cm gap per shift) are typically generated.
The new proposed FIF-COF CSI technique uses similar 3-isocentre beam geometry as the MMFB technique. However, the field junctions are handled by overlapping the open fields by 3cm, and using field-in-field compensation to achieve the desired dose homogeneity across the junction regions in one plan.
Results: The FIF-COF CSI plan shows better dose homogeneity across junction regions while retaining similar overall dose coverage compared to the clinical MMFB CSI plan; spinal canal dose-max in junction regions was reduced from 116% to ~107% using FIF-COF. This efficiently planned FIF-COF technique also reduces dose delivery uncertainties by using more extended junction regions and better accuracy of MLC positions than jaws.
Conclusions: We demonstrate that the FIF-COF technique can be a better alternative approach to the conventional MMFB technique for 3D conformal CSI planning.
Poster #4 - Intestinal Microbiome Interactions in Prediction of Radiation-Induced Lung Injury
Presenting Author: Erin McCurry
University of British Columbia
Purpose: Approximately 15% of patients receiving thoracic radiotherapy will develop radiation-induced lung injuries and while dosimetric factors affect this outcome, individual differences in the intestinal microbiome, which shapes immunity, may also influence the tissue injury response. This project seeks to identify a predictive biomarker for radiation-induced lung injury (RILI) in the intestinal microbiome of laboratory mice.
Methods: We have previously developed two unique mouse strains: one manifesting enhanced RILI and one protected from RILI. Herein, 8-week-old mice of these strains were given 14 Gy whole thorax irradiation, and the intestinal contents of these mice, and of age-matched controls, were collected one week post-irradiation. Supplemental fecal samples were collected from untreated mice of our lab-specific strains and the progenitor strain (C57BL/6J) at 8 weeks of age. Bacterial DNA was extracted and 16S rRNA sequenced. The constituents of the intestinal microbiome in each strain and treatment group were revealed using the NIAID Nephele pipeline sequence analysis with the SILVA taxonomy database.
Results: Preliminary analysis of the intestinal microbiome of female mice has shown a difference between the microbiomes of our lab-specific mouse strains. Specifically, bacterial features which differ between strains with different radiation responses and which correlate with lung disease have been revealed.
Conclusions: Differences that may predict for the development of RILI have been identified in the microbiomes of female mice. Work to identify similar microbiome traits appearing in male mice is currently ongoing. This work may reveal additional bacterial species of interest among RILI resistant and susceptible strains.
Poster #5 - Re-designing PTV Margins for Proactive Adaptive Radiation Therapy for Bladder Cancer
Presenting Author: Je (Jane) He
University of Toronto/Trillium Health Partners
Purpose: In this study, we analyzed the inter-fraction bladder variations using daily CBCT images in bladder radiation treatment and derived a set of PTV margin options based on the statistics. This will enable us to prepare multiple plans proactively with different PTV margins for adaptive selection and ultimately aim to reduce bowel dose as much as possible.
Methods: Thirteen bladder cancer patients who received 44-50 Gy in 20-25 fractions between 2022-2023 at our center were included. Patients were instructed to empty bladder as much as they could prior to CT simulation and each treatment. Daily CBCT was registered with planning CT by bony match (278 CBCT images in total). In each CBCT, the bladder was contoured and compared to that of the planning CT. The difference in 3D was calculated in six directions: Sup-Inf, Ant-Post and Rt-Lt.
Results: Analyzed over the population, mean plus two standard deviations (µ+2σ, 95% confident level) are 20 mm, 11 mm, 10 mm, 14 mm, 11 mm and 12 mm in Sup., Inf., Ant., Post., Rt., and Lt., respectively. The magnitude of variations is patient-dependent, some within 8 mm in all directions and some over 15 mm in at least one direction.
Conclusions: Significant changes in bladder volume and shape are observed in bladder radiotherapy. The variations are asymmetrical due to pelvic anatomy and also patient dependent. Our next step is to retrospectively validate the PTV margin options we derived prior to clinical application.
Poster #6 - Comparing clonogenic survival and cellular proliferation of prostate tumor spheroids and monolayer culture treated with docetaxel chemotherapy
Presenting Author: Stephanie Swanson
University of Waterloo
Purpose: In recent decades, 3D tumor spheroids have become the preferred in vitro tumour model over conventional 2D monolayer cell cultures. This study compared clonogenic survival and cellular proliferation following treatment with docetaxel chemotherapy between prostate cancer cells cultured as 3D tumour spheroids and 2D monolayers. 3D computer simulation was used to further investigate the spatial effect of treatment within spheroids.
Methods: Spheroids and monolayer cultures were exposed to various doses of docetaxel chemotherapy for 24 and 48 hours before measuring clonogenic survival and cellular proliferation. Clonogenic assay was performed by disaggregating and trypsinizing spheroids and monolayer culture respectively, plating cells in 2D for two weeks of observation, then staining and counting colonies. Cellular proliferation was measured with Alamar Blue assay by incubating spheroids and monolayer cultures with resazurin for three hours, then removing the agent and measuring fluorescent intensity. A 3D agent-based computer simulation of tumor spheroid growth and docetaxel treatment was developed to probabilistically employ the survival measured by proliferation assay.
Results: With increasing docetaxel dose and duration of exposure, 3D cell simulation demonstrated decreasing survival and proliferation of 3D prostate tumour spheroids. Clonogenic survival is substantially lower than survival estimated via cellular proliferation, likely by including cell death that occurs after instantaneous measurement of cellular proliferation.
Conclusions: We will present the clonogenic survival and cellular proliferation of 3D tumour spheroids and 2D monolayers following 24 and 48 hours of treatment with docetaxel chemotherapy.
Poster#7 - Failure mode and effects analysis of the liver stereotactic ablative radiotherapy (SABR) process at BC Cancer Centre Surrey
Presenting Author: Karanjit Singh
BC Cancer - Surrey
Introduction: Goal of this study was to use a risk-based approach to assess our center's existing treatment workflow for the Liver Stereotactic Ablative Radiotherapy (SABR) program. This was accomplished using failure mode and effect analysis (FMEA) of our existing procedures, with the aim of detecting and reducing potential shortcomings in the process and enhancing patient care.
Methods: The process was assessed from initial patient consultation to follow-up and a process map was created. A survey was conducted to collect information on the workflow. Any observations during the treatment procedures and relevant patient safety learning system data were included in the FMEA. Chart QA procedure was audited, and interviews were conducted with various staff members. The FMEA results were reviewed using methodology proposed in literature.
Results: This multi-disciplinary project involved physicists, dosimetrists, nurses, CT sim and treatment radiation therapists, and radiation oncologists. The survey results identified the strengths and areas for improvement for the program. A high modulation factor for a treatment plan was an identified failure mode with a risk priority number (RPN) value of 151.2. A check for the modulation factor was added to the plan-checking procedure, as it could lead to issues with the plan or treatment.
Conclusion: Development of the FMEA and process map for our liver SABR program allowed for improved understanding of the process and helped participants to better understand their role in the overall process. Multiple approaches were used to identify failure modes and classify them within the scope of the appropriate personnel.
Poster # 8 - Investigating Associations Between Radiomics Features and Biomarkers in an Orthotopic Non-Small Cell Lung Cancer Mouse Model
Presenting Author: Gabriel Giampa
McGill University
Purpose: To investigate the correlation between radiomics features and histopathological biomarkers expressed by non-small cell lung cancer (NSCLC) tumors in response to treatment.
Methods: Cells from a human NSCLC line were orthotopically implanted into the lungs of 38 male athymic mice. The mice were split into four treatment groups: a control, receiving vehicle saline injections; a group receiving a single radiotherapy dose of 20 Gy; a group receiving injections of carnosine, a potent antioxidant and radiosensitizer; and a group receiving both the radiotherapy dose and carnosine injections. The mice were followed for at least two weeks and underwent cone-beam computed tomography (CBCT) scans two to three times per week. The mice were subsequently euthanized and their tumors were resected and immunohistochemically stained for biomarkers associated with cellular proliferation, hypoxia, and other parameters describing tumor response to the treatment. We used pyRadiomics to extract over 100 radiomics features from the CBCT images.
Results: Thus far, we have seen statistically significant differences in several feature values and biomarker expressions across treatment groups after two weeks of treatment. Several features show linear correlation with the expression of HIF1-alpha, a marker of hypoxia, and further investigation is underway.
Conclusion: The expressions of both radiomics features and certain biomarkers change in response to treatment. If a correlation between these expressions is established, radiomics could present a non-invasive way to obtain biomarker information from a tumor, providing important information on its response to treatment.
Poster #9 - Integrating Remote Locum Physicists in a clinical department to manage physicist workload
Presenting Author: Kristin Marchant
Saskatchewan Cancer Agency
Our department had a positive experience with introducing remote locum physicists as a way of moving forward successfully during a prolonged shortage of medical physicist staffing. In October 2021, funding was approved for 2 new permanent full-time medical physicist positions, increasing the FTE allocation to 6 clinical medical physicists at each of our 2 centres. Over the following 2.5 years, we navigated 4 extended leaves of absence, 3 departures and 3 new hires for clinical physicists, however the 2 new positions remained vacant. We made the decision beginning in September 2022 to hire locum physicists to work remotely, supporting our clinical work by doing physics chart checks. This was done through a professional radiation oncology staffing agency. We continued to schedule an on-site physicist as well on chart-checks to deal with any issues or questions that came up. Both of the physicists we contracted have extensive experience and were able to bring an outside perspective on potential improvements of our processes. Because of this added support, during this time we were still able to successfully implement significant new programs and commission 2 new brachytherapy afterloaders and one linac. In a changing working and staffing environment, remote locum physicists may be a beneficial option for other Canadian clinics to consider.
Poster #10 - Impact of magnetic susceptibility distortions on cranial radiation treatment planning structures
Presenting Author: Tigris Joseph
BC Cancer Vancouver
Purpose: MR is beneficial for radiation treatment planning (RTP) due to its soft tissue contrast, however, differing tissue susceptibilities introduce geometric image distortions. Our objective was to investigate these shifts in clinically contoured brain RTP structures.
Methods: Susceptibility maps of 5 patients from the Glioma Image Segmentation for Radiotherapy: RT targets, barriers to cancer spread, and organs at risk dataset in the Cancer Imaging Archive were made by assigning bulk susceptibility values to bone, air, and tissue CT segmentations (–11.31, 0.36, and –9.1 ppm respectively). MR anterior/posterior spatial shift maps were calculated using a Fourier method on the susceptibility maps at 1.5 T. A global tissue shift was seen in spatial shift maps, that would be accounted for in radiofrequency MR calibration and CT registration. To study the residual shifts in OARs and target structures, the average shift in tissue was subtracted from the spatial shift maps.
Results: Mean structure shifts after subtracting the average tissue shift ranged from -0.22 – 0.06 mm. Maximum structure shifts after subtracting the tissue shifts were largest in the eyes, cochlea, GTV, and CTV (0.26 – 0.42 mm, 1 MR pixel).
Conclusion: Magnetic susceptibility distortions are error sources that may warrant inclusion in PTV/PRV margins depending on a structure’s size, proximity to susceptibility interfaces, and the treatment technique. In the context of MR-only planning, further work is needed to evaluate the impacts of susceptibility distortions on MR segmentation, especially regions close to tissue interfaces.
Poster #11 - Optimization strategies in external beam radiation treatment planning of cervical cancers: a comparative analysis of normal tissue objective versus concentric ring structures
Presenting Author: Susan Dang
University of Victoria/BC Cancer Victoria
Purpose: This study compares the effectiveness of normal tissue objectives (NTO) and concentric ring structures (CRS) in plan optimizing radiation therapy plans for cervical cancer, aiming to optimize the compromise between target coverage and organs at risk (OAR) sparing.
Methods: Three cervical cancer radiation treatment plans (45Gy) were optimized using NTO and CRS in the Eclipse v18.0. NTO settings were fixed (start dose 100%, end dose 40%, and start distance at 0cm) with fall-off values from 0.1 to 0.4. Four-millimeter-wide concentric rings were used around the planning target volume (PTV) to set the desired dose fall-off from gradients of 1 Gy/mm to 3 Gy/mm. The study evaluated priority levels 0, 100, and 200 for each method, with plans normalized to 95% of the PTV being covered by 95% of the prescribed dose. Metrics such as target coverage, maximum dose, and dose volume histograms (DVH) for OARs were evaluated.
Results: NTOs are effective in protecting OARs from high doses while ensuring target conformity, making them suitable for treatments intended to preserve adjacent healthy tissues. In contrast, CRS offer precise dose control and predictability by utilizing predefined dose fall-off rates, benefiting plans involving larger OAR volumes or that require steeper dose gradients.
Conclusion: By integrating the strengths of each method (reliability and predictability) into treatment planning, therapists can create highly optimized treatment plans uniquely customized to the anatomical configuration of each patient. CRSs can also predict potential DVH metrics for OARs even before plan optimization, enhancing NTO’s effectiveness in protecting healthy tissues.
Poster #12 - Licensing Made Simpler - The CNSC Lifecycle Licence
Presenting Author: Rachel Timmins
Canadian Nuclear Safety Commission
Purpose: To update the community on the enhancements made to the regulatory process for medical linear accelerator facilities and what they have achieved.
Methods: In 2020, the Canadian Nuclear Safety Commission (CNSC) undertook an assessment of the regulatory scheme for medical linear accelerator facilities and following the assessment, a new type of licence to streamline the licensing process for routine equipment replacements was created in 2021. Now, after three years using this type of licensing, an analysis has been done on what the efficiency gains provided by this approach look like.
Results: Of the 49 radiotherapy centres licensed by the CNSC, there are 20 centres with lifecycle licenses. These 19 licensees have decommissioned 15 linacs and commissioned 14 linacs. With the lifecycle license these licensees have eliminated amendments and/or new licences associated with 15 decommissioning licences, 14 operate to commission” licences, 14 operation licences, and the revocation of these licences.
Conclusions: If your radiotherapy centre does not have a lifecycle licence and is planning to replace a linac soon, consider applying for a lifecycle licence beforehand. The lifecycle licence can simplify the licensing process eliminating multiple licence amendments and licensing transactions.
Poster #13 - A 3D Printed Device for the Quality Assurance of 3D Printers Used for Bolus Fabrication in Radiation Therapy
Presenting Author: Hunter Lowe
BC Cancer Centre for the North
Purpose: To design a phantom and implement a procedure for the routine or ad-hoc quality assurance (QA) of Fused Deposition Modeling (FDM) 3D printers used for custom patient bolus fabrication.
Methods: A 3D model was designed using Fusion 360 software to test parameters important for fabricating both rigid bolus and silicone moulds, such as infill consistency, dimensional accuracy, and overhang quality. The phantoms were printed on Prusa Mk3S+ and Prusa XL printers with optimized print settings using polylactic acid (PLA), and then filled with EcoflexTM 00-30 type 2-part silicone. Computed Tomography (CT) scans with 1.25 mm slice thickness were used to perform measurements.
Results: The 3D printed QA phantom successfully tested key stresses for FDM printing. Infill consistency values ranged from 87 to 105 Hounsfield units for the solid PLA part of the phantom. Overhangs from 55° to 80° were filled with silicone to test for leaks, and 80° was found to be an achievable upper limit for FDM printing of silicone bolus moulds. Concentric arches assessed the dimensional accuracy of internal cavities, and diameters from 3mm to 30mm were measured to within a half millimeter of precision.
Conclusions: The QA phantom stresses the limits of FDM 3D printers but passes all tests within acceptable tolerances when a printer is configured correctly and in good condition. The phantom may be used for routine quality assurance for any FDM 3D printer used for bolus fabrication in radiation therapy.
Poster #14 - The need for comprehensive characterization of ion collection efficiency during ion chamber commissioning: A case study
Presenting Author: Islam El Gamal
Nova Scotia Health Authority
Purpose: The commissioning of a PTW- Pinpoint 31006 chamber is presented. The impact of not performing a comprehensive ion collection efficiency investigation is demonstrated.
Methods: The ion chamber settling, linearity, leakage and stem effect were assessed by performing irradiations using a 6 MV linac in solid water, using the manufacturer recommended bias and orientation. The ion collection efficiency and polarity effect were assessed using both the two-voltage technique and the Jaffè plot method, as a function of relative dose per pulse. Chamber angular dependence was assessed, and KV images of the chamber electrode acquired to validate the results.
Results: The ion chamber performed as expected when assessing settling, linearity, leakage and stem effect. The two-voltage assessment of ion collection efficiency showed agreement within 0.2 % for all Dpp values when compared to the Jaffè plot method. However, the variation of Pion as a function of Dpp was non-linear and showed anomalous behavior particularly when the chamber polarity was reversed, indicating potential damage to the chamber. KV images acquired of the chamber showed a potentially bent electrode. The chamber showed an angular dependence of 0.65 % confirming the damage to the electrode.
Conclusion: In the absence of a comprehensive assessment of a chamber’s ion collection efficiency damage to the chamber could potentially be overlooked. Measurements performed with the chamber would have an increased uncertainty of 1.11 % compared to an estimated uncertainty of 0.57 % for an undamaged chamber. An assessment of Pion as a function of Dpp is strongly recommended.
Poster #15 - Open-source Picture Archiving and Communication System (PACS) Server Network for Research-Specific Applications
Presenting Author: David DeVries
London Health Sciences Centre
Purpose: Clinical Picture Archiving and Communication Systems (PACSs) are often not suitable for research due to not enough file permissions for researchers or too broad of file permissions for all users, allowing for potential data loss. Research-specific PACSs can suffer from the same file permission issues and are expensive to implement. We propose a research PACS solution to address these needs consisting of a PACS server network built on open-source software.
Methods: We used the open-source Orthanc PACS server software to construct a network of multiple PACS servers all running on a single Linux virtual machine. A head node PACS server would receive data from clinical systems via the Digital Imaging and Communications in Medicine (DICOM) protocol. Based on the patient anonymization scheme, data would be automatically forwarded to additional PACS servers created for each research project. Access permissions to each project’s PACS server was controlled to restrict access to only users associated with each project. A redundant secondary PACS network was also constructed and synchronized with the primary network.
Results: The described PACS network was successfully utilized for multiple research projects to store and provide access to over 0.5 million DICOM files exported from clinical systems. Data security and permissions were tested and operated correctly. System uptime was excellent, with the redundant PACS network providing continued access during minor outages on the primary network.
Conclusions: Our proposed research PACS solution successfully met our objectives of being free, reliable, secure, and able to finely control user access to research data.
Poster #16 - Inverse Planning Strategies for HDR Cervix Brachytherapy
Presenting Author: Claudia Mendez
BC Cancer
Purpose: This study introduces an inverse treatment planning optimization strategy for High Dose Rate (HDR) cervix brachytherapy. The objective is to assess the dosimetric performance of various inverse planning strategies against forward planning, focusing on enhancing planning time efficiency and maintaining planning consistency among planners.
Methods: HDR forward planning can be time-consuming, particularly when using interstitial needles for large target volumes. Inverse planning can expedite the optimization step but requires additional control structures to achieve the traditional pear shape and minimize high-dose volumes. The Elekta Oncentra® Brachy treatment planning system offers two inverse planning algorithms: Inverse Planning by Simulated Annealing (IPSA) and Hybrid Inverse Planning Optimization (HIPO). We developed inverse planning strategies methods that minimize the contouring required for optimization. Plan evaluation included D90%, V100%, and V200% for the high-risk CTV, and D2cc for bladder, rectum, bowel, and sigmoid. Optimization objective sets were evaluated for each method by reporting how many constraints were met by the initial optimization and adjusted to complete the final optimization. The time required for different planning methods was also recorded.
Results: Fifteen cases have been replanned. The mean CTV volume was 33.0 cc (16.1-92.9 cc), with needle usage ranging from none to six. Strategies that minimize the number of control structures have resulted in similar target coverage and overall average improvements in organ-at-risk (OAR) doses while minimizing planning time.
Conclusions: Similar conformity and OAR doses to forward planning were achieved with the proposed IPSA method, accompanied by a notable reduction in the planning optimization time.
Poster #17 - Software for the import of electroanatomic maps into treatment planning systems in the cardiac radioablation target definition process
Presenting Author: Sarah Konermann
McGill University Health Centre, Medical Physics Unit
Purpose: To create a semi-automated software for electroanatomic cardiac map (EAM) conversion to the DICOM standard for cardiac radioablation (CRA) treatment planning, enabling EAM import into the treatment planning system (TPS).
Methods: Software was developed to allow the import of EAMs into a TPS. Voltage and spatial data are sorted into voxels and exported in DICOM format, with each voxel containing the average voltage value of the data that fall within it. The effect of this averaging was evaluated by examining nine different sets of patient data. The standard deviation in each voxel and the difference between the voltage of a given point and its corresponding voxel were analyzed to ensure that data loss is kept to a minimum throughout processing. Five EAMs were used to re-evaluate targets for previously-treated patients with their EAMs now directly registered to their planning CTs.
Results: The mean coefficient of variation for voxels containing two or more data points ranges from 0.02 to 0.1 across the sets of data tested. The mean difference between the actual voltage of a given datum and the voltage in its voxel is less than 0.07% of the mean voltage value in all datasets. Of five target volumes from previously-treated patients, adjustments were made to four (80%) retrospectively using images generated with this tool.
Conclusions: This software provides a reliable solution for EAM import into a TPS in CRA target definition, streamlining target delineation.
Poster #18 - Use of Kilovoltage Triggered Imaging for Motion Management in Stereotactic Ablative
Presenting Author: Maryam Rostamzadeh
BCCA- Abbotsford Centre
Purpose: This study introduces and assesses an intrafraction motion monitoring approach during spine stereotactic body radiotherapy (SBRT) using triggered kilovoltage (kV) imaging. The aim is to enhance treatment precision by directly tracking patient anatomy, crucial for conformal dose distributions and minimizing risks to adjacent radiosensitive organs, notably the spinal cord.
Methods: Utilizing the Varian TrueBeam advanced imaging package, a method was developed to identify vertebrae adjacent to the target volume on triggered kV imaging and alert the radiation therapist if vertebral motion exceeds a pre-defined threshold equal to the planning target volume margin. Phase one involves software development and testing on an anthropomorphic phantom simulating spine SBRT treatments. Phase two entails employing triggered kV imaging in patients undergoing spine SBRT.
Results: Evaluation with the anthropomorphic phantom demonstrates the feasibility and precision of the triggered kV imaging approach for intrafraction motion monitoring during spine SBRT. Triggered kV imaging effectively detects vertebral motion, enabling real-time intervention if necessary. The method shows promise in providing accurate feedback on target structure position during treatment delivery.
Conclusions: This study validates triggered kV imaging as an effective tool for intrafraction motion management in spine SBRT. By directly tracking patient anatomy during treatment delivery, this approach offers a practical and accessible solution for clinics aiming to optimize the accuracy of spine SBRT treatments. Successful implementation of triggered kV imaging could significantly improve treatment precision, thereby enhancing patient outcomes while minimizing risks to adjacent critical structures.
Poster #19 - EZFluence Evaluation for Automated Breast Planning
Presenting Author: Mithunan Modchalinga
Trillium Health Partners - Credit Valley Hospital
Purpose: The plan quality and time savings of automated breast planning using Radformation’s EZFluence was evaluated.
Materials and Methods: Manually planned breast patients were retrospectively reviewed and re-planned using EZFluence. A total of 32 two-field breast tangents, 15 four-field breasts, and 10 high-tangents were re-planned. Plan quality was evaluated by comparison of the breast and nodes PTV V95%, along with the V103% and global max dose. Each plan was timed and compared to an average time of segmentation obtained through a survey of 10 planners to determine the time savings.
Results and Discussions: The two-field breast tangents produced by EZFluence showed an insignificant change in the breast PTV V95% of +0.3% (p=0.15). However, the V103% and max dose showed significant reductions of 9.0% (p<0.001) and 0.6% (p<0.001). Segmentation time was reduced by 81%, from 25.3 minutes to 4.8 minutes. The breast and nodes PTV V95% of four-field plans also showed insignificant changes of +0.0% (p=0.43) and -0.2% (p=0.16). Four-fields had a reduction in V103% of 188.9cc (p<0.001) and a reduction in max dose of 0.6% (p=0.002). Segmentation time was reduced by 71%, from 58.8 minutes to 16.9 minutes. High tangents also showed insignificant changes in breast and nodes PTV V95% of -0.5% (p=0.16) and +0.1% (p=0.77). The V103% and max dose were again significantly reduced by 74.4cc (p=0.049) and 1.1% (p=0.009). Segmentation time was reduced by 64%, from 30 minutes to 10.8 minutes.
Conclusions: EZFluence produced plans of equivalent coverage and improved homogeneity in significantly reduced times.
Poster #20 - Patient-reported outcome prediction models: a head and neck case-study
Presenting Author: Owen Paetkau
University of Calgary
Purpose: Evaluate the robustness of machine-learning-based predictive models to changing clinical demographics over time, as applicable to patient-reported outcomes (PRO) for head and neck cancer.
Methods: Two cohorts of head-and-neck cancer patients, treated with radical chemoradiotherapy were recruited to this retrospective study (2019: 155 patients; 2023: 70 patients). At routine follow-up clinics, patients completed three surveys: MDASI assessing general patient-reported symptom severity; MDADI for dysphagia-specific symptoms; and XQ for xerostomia-specific symptoms. Medical chart review and treatment planning features, including organ-at-risk dose-volume data, provided input data for machine-learning-based regression model development. Composite scores of each PRO survey served as the endpoint to be predicted. We developed regression models for acute (6 months since treatment completion) and late (>6 months) symptom reporting. Input variables with statistically significantly differences between cohorts were identified and removed from model development. Five-fold cross validation optimized feature selection (KBest) for the selected machine learning model types (random forest, XGBoost, lightBGM), and enabled hyperparameter tuning. Mean absolute error assessed training (2019) and testing (2023) model performance over bootstrapped samples.
Results: Statistically significant differences between cancer site, nodal stage, and patient age occurred between the cohorts, indicating potential fluctuations in patient demographics. Removing these variables from model development, mean absolute error was comparable to PRO survey cutoffs for clinical significance, suggesting that further model development may be able to identify clinically significant changes in symptom scores over time.
Conclusions: Our machine learning models, developed on input variables without significant changes over time, demonstrated robust temporal performance.
Poster #21 - Enhancing Radiation Therapy Dosimetric Outcomes for Pancreatic Tumors: Utilizing Gold Nanoparticles with 2.5 MV Photons
Presenting Author: Navid Khaledi
CancerCare Manitoba
Purpose: This study explores the potential of computing dose enhancement with gold nanoparticles (GNPs) in pancreatic tumor radiation therapy with a 2.5MV photon beam compared with a conventional 6MV, in order to improve the dosimetric outcomes.
Methods: To employ the impact of GNPs in pancreatic cancer radiotherapy, five cases were considered for comparison. By employing custom MATLAB scripts and an open-source treatment planning software, matRad, relative biological effectiveness (RBE) was computed in each voxel of the target based on dose distributions of a 6MV clinical beam and a 2.5 MV photon beam. The doses and dose-volume histograms (DVHs) were computed by comparing various scenarios with and without gold nanoparticles (GNPs).
Results: Integration of GNPs with 2.5 MV photon beams substantially enhanced tumor control probability (TCP) from 58% to 97%, while maintaining low normal tissue complication probability (NTCP) (<4%). On the other hand, 6 MV photon beam with a hypofractionated dose schedule yielded higher TCP rates however elevated NTCP from 2% to 77% as well.
Conclusions: The preliminary findings highlight the potential of combining GNPs with a 2.5MV therapy to improve radiation therapy outcomes for pancreatic cancer. Further exploration into optimal concentrations and sizes for diverse tumor types and organs at risk are essential to confirm the preliminary findings. This study contributes to advancing nanoparticle-assisted radiation therapy, offering avenues for enhanced therapeutic efficacy while mitigating normal tissue complications and improving clinical outcomes in pancreatic cancer patients.
Poster #22 - Prone Breast Radiation Therapy: A simple solution for CT simulation challenges, large pendulous breast planning
Presenting Author: Janos Juhasz
Miami Cancer Institute
Purpose: A simple solution to overcoming challenges for large pendulous breast radiation therapy is presented. The CT couch interferes with the natural shape of the breast and results in an undesired shape for achieving dose homogeneity across the target region. In addition, it does not reflect the actual treatment position for setup assuming conventional linear accelerator delivery systems.
Methods: Five patients were simulated in CTSIM in the prone position by allowing the breast to touch the CT table, and they were transferred for set up in a LINAC bunker. A cone-beam CT was acquired in the same position, allowing the breast to naturally fall into an unhindered position as it would be situated on the treatment days. Fusion between the two data sets was performed. External contours and overrides for differences in the breast shapes between the two data sets were defined. The choice of density override for breast tissue and soft tissue was analyzed and compared. Treatment planning was prepared using a 3D conformal technique, except for one VMAT case. Daily setup images were
acquired and analyzed to measure setup reproducibility.
Results: Optimized treatment plans passed all in-house diametric criteria. Density overrides of soft tissue versus breast tissue did not have any clinical significance, and both plans for each patient showed minimal differences in dose coverage and DVH parameters. Setup reproducibility was excellent and comparable to standard-prone breast setups at our institution.
Conclusion: All five patients undergoing this technique were successfully simulated, planned, and treated with excellent dosimetric outcomes and setup reproducibility.
Poster #23 - Dose metric for assessing the dosimetric impact of a geographic miss
Presenting Author: Baochang Liu
Juravinski Cancer Centre
Purpose: Defining the dosimetric impact of a geographic miss error in radiation therapy is complicated due to the heterogeneity of the dose difference, which also makes the classification of the severity of the error difficult. Metrics based on geometric differences have been proposed but they are not generally applicable if a few fractions of a course of treatment are affected. Therefore metrics based on volumetric dose differences were investigated to identity the optimal metric to describe dosimetric impact.
Methods: Geographic errors were simulated mathematically for existing optimized patient treatment plans, and the resultant dose distributions were compared with the planned dose distributions. Dose metrics such as mean, median, Dmax2cc, Dmin2cc, EUD, V95, D95 were calculated for the planned dose distribution and the composite dose distribution including the offset. The dosimetric impact was calculated as the percentage difference of a dose metric from the planned value for both target volumes and for organs at risk (OAR).
Results: When examining target coverage, among all metrics, the D95 matric was the most stable metric and is recommended for calculation and reporting the dosimteric impact of the error. The dosimetric impact to OARs are more complicated and it is recommended that dosimetric impact be based on key DVH parameters used in developing the treatment plan.
Conclusions: The optimal dose metric to assess the impact of a geographic miss for target volumes is the D95. For OARs key DVH values used for plan optimization are appropriate.
Poster #24 - Morphological assessment of synthetic CT images generated from a Flexible Ultra Short Echo Time (FUSE) Sequence: Beyond using the DICE similarity coefficient
Presenting Author: Asieh Tavakol
CancerCare Manitoba/University of Manitoba
Purpose:This study aims to use different morphological metrics to validate sCT obtained by the novel Flexible Ultra Short Echo Time (FUSE) Sequence, which may have a significant effect on ensuring the quality of sCT generated to use in treatment planning systems (TPS).
Methods:The sCT derived from the FUSE sequence and CT images were acquired using a dried cadaver skull phantom. To compare the FUSE images with CT imaging, image registration, segmentation, and geometric similarity assessment were performed. The CT images were registered to the FUSE images using 3D Slicer software, and bone segmentation was conducted on both image sets. The Fuzzy C-means clustering method was used to extract bone from the FUSE difference images, generating probability maps for the skull bone and background. Geometric similarity between the segmented masks was evaluated using metrics such as DSC, sensitivity, volume ratio, HD, MASD, and ICC.
Results: The results of the geometric validation showed the following metrics: a Dice similarity coefficient (DSC) of 0.69, a Sensitivity of 0.77, a volume ratio of 0.82, a Hausdorff distance difference of 13.8 mm, a mean absolute surface distance of 0.7, and an Interclass Correlation coefficient of 0.67. Additionally, we demonstrated the practical applicability of the sCT data generated using the FUSE sequence in radiation therapy planning.
Conclusion: Various morphological metrics were used to assess the similarity between FUSE-based pseudo-CT and CT, Overall, our study confirms the feasibility and potential of the FUSE sequence for generating clinically equivalent sCT data for radiation therapy planning.
Poster #25 - On the use of Fricke gel dosimeters for the delivery quality assurance of whole-brain irradiation with hippocampal sparing
Presenting Author: Seyedeh Mar Taghavi
University of Montreal
Purpose: This study focuses on the fabrication and application of Fricke gel dosimeters for quality assurance in complex radiotherapy treatments, specifically whole-brain radiation therapy with hippocampal sparing (HS-WBRT).
Methods: Two one-liter samples of Fricke gel were prepared, composed of 6% w/w gelatin, 65 mM sulfuric acid, and 1 mM ferrous ammonium sulfate in PET containers. One container underwent irradiation according to a designed calibration plan, with uniform dose sections of 1, 3, 5, 8, and 10 Gy delivered within the container, while the other container was exposed to the HS- WBRT plan. Irradiations were conducted with the containers positioned in the middle of a plexiglass cylindrical phantom filled with water, employing a VMAT technique, with E = 6 MV and Dose rate = 500 MU/min. Turbo spin echo sequences with an inversion recovery pre-pulse were used to acquire T1 maps for dose distribution analysis, with imaging conducted one hour post-irradiation.
Results: The dosimetric properties of the Fricke gel, including dose response linearity, were assessed. For the validation, dose distribution from the TPS plan was used as reference. The gamma analysis passing rate using a criteria of 5%/3mm between the dose distributions from Fricke gel measurement and TPS calculation was 97.5%.
Conclusion: This study establishes the viability of Fricke gel dosimeters for quality assurance in HS-WBRT treatments.
Poster #26 - Parameter optimization for a 3-Dimensional Treatment Planning System based Model of Moist Desquamation Risk for Breast Radiotherapy
Presenting Author: Aria Malhotra
BC Cancer Vancouver
Purpose: Moist desquamation (MD) is a severe skin response to radiotherapy that can cause pain and discomfort, and in severe cases, treatment interruptions. Previously, a model predictive of moist desquamation was developed on 2-dimensional in-vivo film dosimetry, with promising results. In this study we explore the optimization of this model in a more clinically relevant 3-dimensional treatment planning system context.
Methods: The Eclipse treatment planning system’s Analytical Anisotropic Algorithm was used to calculate treatment planning dose for 20 whole breast patients treated with different dose prescriptions and photon beams. of 42.5Gy/16 fractions and 50Gy/20 fractions with a combination of 6MV, 10MV and 15MV photon beams. The presence of MD was classified based on a combination of staff assessments and patient reported outcomes. The surface rind of the 20 body contours are extracted and each voxel is run through a modified logistic model that assigns a new value based on the spatial distribution of dose surrounding each voxel. The inflection point and slope of the function are optimized to differentiate between regions of the breast where moist desquamation occurred and regions that had no moist desquamation.
Results: Testing of the 3D predictive risk model with various parameters on 3D TPS data showed promising results. Further testing on a clinical trial dataset is ongoing.
Conclusions: This model has been shown to predict specific regions of skin expected to develop MD in breast radiotherapy, which could allow targeted action to be taken prior to treatment. Expanded testing in a clinical context is underway.
Poster #27 - Elekta Hounsfield Unit Calibration and 3D Printed Phantom for CBCT-Based Dose
Presenting Author: June Cheng Baron
Durham Regional Cancer Centre
Purpose: Evaluate the Elekta XVI Hounsfield unit (HU) calibration and develop a process for dose calculation using cone-beam imaging.
Methods: HU calibration was performed according to the manufacturer’s procedure and image quality parameters were evaluated using the Catphan 504. CBCT to electron density (CBCT-to-ED) curves were generated using a custom 3D printed phantom that can accommodate the Gammex 467 tissue characterization inserts. The 3D printed phantom was designed to have dimensions similar to a human head and body to simulate in vivo scatter. CBCT-derived dose was compared to planning CT-derived dose using gamma analysis in the head, thorax and pelvis.
Results: Implementation of the HU calibration showed an improvement in CBCT-derived HU versus reference HU in the Catphan. It did not affect image quality parameters of low contrast visibility, spatial resolution and spatial geometry. CBCT images of the 3D printed phantom produced a CBCT-to-ED curve significantly different from our standard planning CT-to-ED curve. Applying the standard CT-to-ED curve resulted in passing rates mostly >95%, but as low as 80% in some cases. The use of CBCT-to-ED curves did not reliably show improvement over the standard CT-to-ED curve with passing rates as low as 65% in some cases.
Conclusions: Implementation of the XVI HU license and calibration produces more consistent CBCT values. However, accurate dose calculation using a standard CT-to-ED curve cannot necessarily be achieved. Further work is necessary to elucidate the relationship between CBCT value and phantom/patient geometry to improve the CBCT-to-ED curve and to achieve robust CBCT-based dose calculations.
Poster #28 - Development of a high-accuracy experimental benchmark of in-air off-axis ratios for validation of Monte Carlo radiation transport codes
Presenting Author: Claudiu Cojocaru
National Research Council Canada (NRC)
Purpose: Off-axis in-air ratios can be used during linear accelerator commissioning to develop TPS beam models. Using a research linac, with very well-defined beam parameters and a verified beam transport geometry allows these in-air ratios to be used as a strict test of the highest-accuracy Monte Carlo simulation codes. This comprehensive study aimed to investigate possible sources of systematic error in such radial beam profiles to yield the necessary high-precision experimental data sets.
Methods: Electron beams with well defined energies and geometries produced by a research LINAC were stopped in various targets (Be, Al, Cu, Ta and Pb). In-air scans of the bremsstrahlung distribution were performed at ~ 1m SAD, perpendicular to the beam axis, and went from -55 cm to + 55 cm from the origin (corresponding to an angular opening of ± 26°). Influence quantities, including ion chamber type and geometry, build-up cap material, polarity effects, positioning and scan orientation were investigated.
Results: Over 100 beam profiles were collected by permutating targets, build-up caps and ionization chambers. The FWHM of the distribution correlated with the Z of the target (Be being the narrowest, Pb the widest). The estimated standard uncertainty on the measurements is 1.0 %, with little variation from the central axis to the edges of the distribution.
Conclusions: A comprehensive set of experimental data has been collected for off-axis bremsstrahlung profiles. Given the high accuracy of the data it can be used for benchmarking Monte Carlo codes used for radiotherapy applications.
Poster #29 - Update on the proposed regulatory review of the Class II Nuclear Facility and Prescribed Equipment Regulations (C2NFPER)
Presenting Author: Julie-Anna Benjamin
Canadian Nuclear Safety Commission
In the autumn of 2022, CNSC briefed stakeholders on the proposed changes to the C2NFPER via public workshops and a discussion paper. This presentation will provide an update on the project, with a focus on what we heard from commenters following the 2022 workshops and discussion paper.