Session III ; Young investigators Presentations
Objectives
9:25 7-01
Deep learning-based Auto-segmentation of Organs at Risk in High-Dose Rate Brachytherapy of Cervical Cancer
Reza Mohammadi, Iman Shokatian, Mohammad Salehi, Hossein Arabi, Isaac Shiri and Habib Zaidi
Introduction: Delineation of organs at risk (OARs), such as the bladder, rectum and sigmoid, plays an important role in the delivery of optimal absorbed dose to the target owing to the steep gradient in high-dose rate brachytherapy (HDR-BT). In this work, we propose a deep convolutional neural network-based approach for fast and reproducible auto-contouring of OARs in HDR-BT.
Materials and Methods: Images of 113 patients with locally-advanced cervical cancer were utilized in this study. We used ResU-Net deep convolutional neural network architecture, which uses long and short skip connections to improve the feature extraction procedure and the accuracy of segmentation. Seventy-three patients chosen randomly were used for training, 10 patients for validation, and 30 patients for testing. Well established quantitative metrics, such as Dice similarity coefficient (DSC), Hausdorff distance (HD), and average symmetric surface distance (ASSD), were used for evaluation. D5cc, D2cc, D1cc, and D0.1cc for OARs, were compared between ResU-net and Unet models to assess the effect of predicted segmentation inaccuracies on dose-volume parameters.
Results: The DSC values for the test dataset were 95.7± 3.7%, 96.6±1.5% and 92.2 ± 3.3% for the bladder, rectum, and sigmoid, respectively. The HD values (mm) were
4.05±5.17, 1.96±2.19 and 3.15±2.03 for the bladder, rectum, and sigmoid, respectively. The ASSDs were 1.04±0.97, 0.45±0.09 and 0.79±0.25 for the bladder, rectum, and sigmoid, respectively. The ResU-Net model achieved better results compared to Unet model for all dosimetric parameters. The differences between all dosimetric parameters for the sigmoid were statistically significant between the two models (p-value < 0.05). There is no proof of statistical difference for the bladder and rectum considering all dosimetric parameters.
Conclusion: The proposed deep convolutional neural network model achieved a good agreement between the predicted and manually defined contours of OARs, thus improving the reproducibility of contouring in brachytherapy workflow.
9:32 7-02
Development of a dynamically rotating shield for endorectal intensity-modulated brachytherapy
Alana Thibodeau-Antonacci, Té Vuong, Hamed Bekerat, LiHeng Liang, Shiring A. Enger
Purpose:
Total mesorectal excision is the primary curative treatment for rectal adenocarcinoma, often preceded by chemoradiotherapy. External beam radiation therapy is the best-studied form of neoadjuvant radiotherapy but is associated with significant toxicities. An alternative is high-dose-rate brachytherapy (HDR-BT), with initial results indicating fewer toxicities and similar perioperative outcomes. However, the rotationally-symmetric dose distributions provided by brachytherapy sources deliver a high dose to the tumor, but often with poor target conformity resulting in dose spillage to the organs at risk (OAR). By incorporating dynamically rotating metallic shields, intensity-modulated brachytherapy (IMBT) opens the possibility to deliver more conformal dose distributions. The goal of this study was to develop a novel MRI-compatible dynamic-shielding IMBT applicator for the treatment of rectal cancer and compare the results with conventional HDR-BT with and without static shielding.
Methods:
For dynamic-shielding IMBT, a single-grooved, MRI-compatible, 15 mm diameter tungsten shield with a 180° emission window was designed. A cylindrical intracavitary applicator (length 28 cm, diameter 2 cm) (Elekta Brachytherapy, Veenendaal, The Netherlands) with a central lumen that fits an 8 mm diameter tungsten rod for static shielding was also modeled. For both applicator models, four rectal cancer patients were retrospectively planned using an in-house Monte Carlo based treatment planning system called RapidBrachyMCTPS and a generic Ir-192 source. The plans were optimized using a fast mixed-integer optimization method. The prescribed dose was 3 fractions of 10 Gy. The resolution of the scoring grid was 3x3x3 mm3 and 106 decay events were simulated.
Results:
For the same target coverage, the contralateral healthy rectal wall D2cc, rectum D50 and pelvis D50 decreased on average by 38.93%, 18.82% and 18.75% when using dynamic-shielding IMBT compared with HDR-BT with the static shield and by 53.04%, 68.35% and 36.62% compared with HDR-BT without the static shield. However, the inferior and superior dose spill regions D2cc were 11.14% and 9.57% greater compared with static-shielding HDR-BT. The treatment times were below 30 minutes for all cases.
Conclusion:
The designed dynamically-rotating IMBT applicator for the treatment of rectal cancer shields the OAR more effectively with the possibility to increase the delivered dose to the tumor.
9:39 7-03
RapidBrachyTG43: A TG-43 Parameter Calculation Engine for Brachytherapy Applications
Jonathan Kalinowski, Gabriel Famulari, Shirin A. Enger
Introduction: Dosimetry calculations in clinical brachytherapy are performed according to the American Association of Physicists in Medicine's Task Group 43 (TG-43) formalism, approximating the patient as a uniform water phantom with the dose rate distribution equal to a product of source-specific parameters. While TG-43 parameter data is widely available in the literature for many commercial brachytherapy sources, presently there is no user-friendly software package which can facilitate accurate and flexible TG-43 parameter calculations for novel brachytherapy source models and isotopes. The aim of this project was to develop an open source, user-friendly Monte Carlo (MC) based software package for calculation of TG-43 parameters of various combinations of brachytherapy source models and isotopes, including novel source models implemented by the user. This software package would then provide an independent library of TG-43 parameters useful for benchmarking against other MC codes and clinical treatment planning systems.
Materials and Methods: RapidBrachyTG43 is an open-source software package based on the Geant4 MC toolkit, and is integrated with the MC-based treatment planning system RapidBrachyMCTPS. Air kerma strength (per unit source activity), dose rate constant, radial dose function, and 2D anisotropy function are calculated for each source using dose/kerma-rate maps determined through simulations. Radioactive decay is explicitly simulated. Initial testing of this package was done with the microSelectron-v2 (Ir-192), benchmarked against calculations from RapidBrachyMC and BrachyDose. RapidBrachyTG43 will be further verified with additional source models, including low energy isotopes, and the complete RapidBrachyTG43 package will be released to the public by Fall 2021.
Results: TG-43 parameters were calculated for the microSelectron-v2 using RapidBrachyTG43. Data was prepared with sub-percent type A uncertainties in each dose rate map at all points. Air kerma strength and dose rate constant were calculated to be (9.683±0.006)*108 U/Bq and 1.116±0.005 cGy/h/U respectively, within 1% agreement to published data. Radial dose function and 2D anisotropy function evaluated at r=1 cm also generally agreed to published data within 1%.
Conclusion: Upon release, users of RapidBrachyMCTPS will therefore be able to perform accurate, fast, and user-friendly calculations of TG-43 parameters using RapidBrachyTG43. TG-43 parameters for the microSelectron-v2 calculated with this package agreed within 1% of published data. While most other MC codes utilize pre-calculated photon spectra for simulations, the simulation of decay events directly in RapidBrachyTG43 will allow users to experiment with any isotope of their choosing without prior spectrum calculations. This package will also facilitate the creation an independent dataset of TG-43 parameters for commercial brachytherapy sources, providing a benchmark for other TG-43 parameter calculations.
9:46 7-04
Safety and efficacy of 2D Brachytherapy vs 3D Image-Guided Adaptive Brachytherapy for locally advanced cervical Cancer – A single institution retrospective study
Mame Daro Faye, MD-PhD, Mariana P. Araujo, MD, Michel D. Wissing, MD, Khalid Alrabiah, MD, Joanne Alfieri, MD
Introduction: Treatment paradigms for locally advanced cervical cancer have shifted from 2D- brachytherapy (2D-BT) to 3D-image-guided adaptive BT (3D-IGABT), with the goal of delivering doses more precisely to clinical targets while sparing organs-at-risk. In this retrospective study, we aimed to report our institution’s experience with 3D-IGABT for cervical cancer.
Materials and Methods: Patients with cervical cancer who received chemoradiation between February 2004 and August 2019 were studied, including 98 3D-IGABT and 48 2D-BT patients. Data was retrospectively collected until December 2019. Odds ratio (OR) for treatment-related toxicities and hazard ratios (HR) for locoregional control (LRC), distant control (DC), failure- free survival (FFS), cancer-specific survival (CSS) and overall survival (OS) are reported with their 95% confidence interval.
Results: Median follow-up was 50.3 months (35.6-57.6 months). There was a significant decrease in late toxicities in the 3D-IGABT group compared to the 2D-BT group (OR 0.22[0.10- 0.52]), specifically for late gastrointestinal (OR 0.31[0.10-0.93]), genitourinary (OR 0.31[0.09- 1.01]) and vaginal toxicities (0% vs 29.6%). Grade≥3 toxicity was not significantly (NS) different but low (2D-BT: 8.2% acute, 13.3% late vs 3D-IGABT: 6.3% acute, 4.4% late; NS). There were no grade 5 events. The 5-year LRC, DC, FFS, CSS and OS for the 3D-IGABT group were 92.0%, 63.4%, 61.7%, 75.4% and 73.6%; compared to 87.3%, 71.8%, 63.7%, 76.3% and 70.8% respectively in the 2D-BT groups (NS). Factors significantly predicting for worse FFS, CSS and OS on multivariate analysis included advanced disease stage and lymph node status.
Conclusion: 3D-IGABT for the treatment of cervical cancer is associated with a decrease in late gastrointestinal, genitourinary and vaginal toxicities in this group, confirming 3D-IGABT as a safer treatment approach for cervical cancer. We did not find any significant difference in disease control or survival outcomes, but rates were favorable and comparable to historical controls.
9:53 7-05
BIOCHEMICAL RECURRENCE-RATE AND PRESERVATION OF ERECTILE FUNCTION AFTER PROSTATE SEED BRACHYTHERAPY IN EARLY-ONSET PROSTATE CANCER
Cédric Charrois-Durand, Daniel Taussky, Guila Delouya, Daniel Liberman
Introduction:
Decisions regarding treatment modality in localized prostate cancer (PCa) is often based on treatment-related toxicity. There is very little contemporary data on outcomes and toxicity in early- onset (≤55 years) PCa. We analyzed the biochemical recurrence-free survival and erectile dysfunction (ED) in younger men treated with prostate seed brachytherapy (PB).
Methods:
We questioned our prospectively maintained institutional database for patients ≤55 years treated with PB. Erectile function at baseline and after treatment were assessed using the standardized physician-reported measure called the Common Terminology Criteria for Adverse Event Scale (CTCAE) version 4.0. Biochemical recurrence (BF) was defined according to the Phoenix Consensus definition (PSA nadir + 2 ng/mL). The log-rank test (Kaplan-Meier method) and cox- regression analysis was used to calculate BF-free survival.
Results:
Between July 2005 and November 2020, a total of 137 patients ≤55 years (range 44-55 years old) were treated with PB. Median follow-up was 72 months. 20% had Gleason 3+4 disease, 6% a PSA >10 ng/mL. Median prostate volume was 34 cc. Actuarial biochemical recurrence free survival at 5, 7, and 10 years, were 98%, 95% and 89%, respectively. On multivariate analysis, CAPRA-score (HR 4.46, 95%CI 1.76-11.33, p=0.002 and the dosimetric measure D90 >130 Gy (p=0.03) were predictive of BF. Five deaths occurred in our cohort, two due to cardiovascular reasons and three due to another malignancy. At baseline, all patients were able to have erections with or without medication. At 5 years and 7 years after PB, 80% (n=46) and 64% (n=32) of patients had little or no ED (erections without the need for medication) respectively.
Conclusion:
In young-onset patients treated with PB, recurrence rates are similar to their older counterparts. Sexual function decreases with time, even in patients with good sexual function.
Evaluation: