Search

Purpose: Convolutional Neural Networks (CNNs) have emerged as transformative tools in the field of radiation oncology, significantly advancing the precision of contouring practices. However, the adaptability of these algorithms across diverse scanners, institutions, and imaging protocols remains a considerable obstacle. This study aims to investigate the effects of incorporating institution-specific datasets into the training regimen of CNNs to assess their generalization ability in real-world clinical environments. Focusing on a data-centric analysis, the influence of varying multi- and single center training approaches on algorithm performance is conducted. Methods: nnU-Net is trained using a dataset comprising 161 18F-PSMA-1007 PET images collected from four distinct institutions (Freiburg: n = 96, Munich: n = 19, Cyprus: n = 32, Dresden: n = 14). The dataset is partitioned such that data from each center are systematically excluded from training and used solely for testing to assess the model's generalizability and adaptability to data from unfamiliar sources. Performance is compared through a 5-Fold Cross-Validation, providing a detailed comparison between models trained on datasets from single centers to those trained on aggregated multi-center datasets. Dice Similarity Score, Hausdorff distance and volumetric analysis are used as primary evaluation metrics. Results: The mixed training approach yielded a median DSC of 0.76 (IQR: 0.64–0.84) in a five-fold cross-validation, showing no significant differences (p = 0.18) compared to models trained with data exclusion from each center, which performed with a median DSC of 0.74 (IQR: 0.56–0.86). Significant performance improvements regarding multi-center training were observed for the Dresden cohort (multi-center median DSC 0.71, IQR: 0.58–0.80 vs. single-center 0.68, IQR: 0.50–0.80, p < 0.001) and Cyprus cohort (multi-center 0.74, IQR: 0.62–0.83 vs. single-center 0.72, IQR: 0.54–0.82, p < 0.01). While Munich and Freiburg also showed performance improvements with multi-center training, results showed no statistical significance (Munich: multi-center DSC 0.74, IQR: 0.60–0.80 vs. single-center 0.72, IQR: 0.59–0.82, p > 0.05; Freiburg: multi-center 0.78, IQR: 0.53–0.87 vs. single-center 0.71, IQR: 0.53–0.83, p = 0.23). Conclusion: CNNs trained for auto contouring intraprostatic GTV in 18F-PSMA-1007 PET on a diverse dataset from multiple centers mostly generalize well to unseen data from other centers. Training on a multicentric dataset can improve performance compared to training exclusively with a single-center dataset regarding intraprostatic 18F-PSMA-1007 PET GTV segmentation. The segmentation performance of the same CNN can vary depending on the dataset employed for training and testing. [ABSTRACT FROM AUTHOR]

Poster: Clinical track: Prostate PO-1186: Impact of a low FODMAP diet on rectal gas and rectal volume during radiotherapy of prostate cancer C. Schaefer, C. Zamboglou, N. Volegova-Neher, C. Martini, N.H. Nicolay, N. Schmidt-Hegemann, P. Rogowski, M. Li, C. Belka, A. Grosu, T. Brunner. [Extracted from the article]

Debate: This house believes that Elective nodal radiotherapy should be performed for high risk prostate cancer SP-0021: For the motion (for ENI) C. Zamboglou. [Extracted from the article]

The implementation of synthetic data (SD) might enhance clinical research in radiation therapy (RT) while protecting patient privacy. Especially, in the era of artificial intelligence, big datasets are needed for outcome prediction modelling. The best methodology to create SD in the field of RT and to assess its performance is unknown. Consequently, the aim of this work was to create a framework for SD creation and evaluation with machine-learning (ML) based models and to assess the application for data-sharing scenarios. Five retrospective survival related datasets (n = 1038 recurrent prostate cancer, n = 109 primary localized prostate cancer, n = 46 metastasized prostate cancer, n = 1072 head and neck cancer, n = 298 gliomas) with patients undergoing RT in different scenarios were collected. Four different ML-based models (Tabular Variational Autoencoder (TVAE), Conditional Tabular General Adversarial Networks, Gaussian Copula, Hybrid Copula General Adversarial Network) were applied to create multiple iterations of SD for each dataset, all being the same size as the original datasets. Subsequently, SD datasets were compared with their original counterpart, with initial exclusion criteria including p<0.05 in the log-rank test and >5% of exact data row matches. From the remaining candidates, the most suitable SD were chosen based on a variety of metrics, including the concordance index and comparison of hazard ratios from multivariate Cox Proportional Hazards. There was no significant difference (p value = 0.704, Student-t test) in the Concordance indexes between the best performing SD and their respective original counterpart. The hazard ratios found from fitting the SD were consistently within the corresponding 95% confidence intervals of the original data's' confidence intervals, suggesting strong explainability. Most SD sets had 0% exact matches (with numeric tolerance) compared to their original counterpart. The Tabular Variational Autoencoder (TVAE) method consistently produced optimal results, with all final SD sets chosen stemming from its usage. We present a framework, comprising of methods and metrics, to create and to evaluate the performance of SD in the field of RT. The final SDs that were chosen achieved a high level of privacy, explainability and predictive efficacy. This showcases the strong recent advancements of synthetic data-generating ML models, especially TVAEs. [ABSTRACT FROM AUTHOR]

The prospective, 2-armed non-randomized HypoFocal phase II trial investigates the safety and feasibility of focal dose escalated external beam radiotherapy (EBRT) and high-dose-rate brachytherapy (HDR-BT) for prostate cancer (PCa) patients based on PSMA-PET and multiparametric MRI. This approach improves tumor coverage and thus putatively treatment effectiveness but leads to larger boost volumes. Here we present toxicity and patient reported quality of life (QoL) results after 2 years follow-up (FU) Patients with intermediate- or high-risk PCa and cN0/cM0 stage were included. Patients in arm A received 60 Gy in 20 fractions to the prostate with an integrated boost of up to 75 Gy. Patients in arm B received one session HDR-BT with 15 Gy to the prostate and a boost of up to 19 Gy, followed by EBRT of 44 Gy in 20 fractions. Boost volumes were defined by PSMA-PET and mpMRI based on validated approaches. Genitourinary (GU) and gastrointestinal (GI) toxicity (CTCAE v5.0) and QoL with IPSS and EORTC questionnaires (QLQ30 and PR25) were assessed. Fifty patients were treated in both arms in two centers (Freiburg and Berlin). Table 1 shows patients characteristics. In arm, A grade 2 GU and GI toxicity rates after 2 years were 8% and 4%. There were no grade 3 GU toxicities. Two patients experienced grade 3 GI toxicities due to multifactorial causes. In Arm B grade 2 GU and GI toxicity rates after 2 years were 17% and 0%. No grade 3 toxicities were observed in arm B. Toxicities were not statistically significantly different between baseline and 2y FU (p>0.055). QoL analysis was performed with patients with available questionnaires at baseline and 2y FU (12-15 in Arm A and 13-15 in Arm B). Only bowel function (p = 0.0005, median 4 vs 25 points) in Arm A and sexual- (p = 0.016, median 25 vs 50 points) and bowel function (p = 0.004, median 0 vs 8 points) and dyspnea (p = 0.031, median 0 vs 0 points) in Arm B decreased significantly after 2 year FU. Other QoL items were not significantly different. Bowel symptoms were significantly worse in Arm A compared to Arm B (p = 0.003). Median PSA values after 2 years were 0.23 ng/ml in Arm A and 0.33 ng/ml in Arm B. Despite large boost volumes, the 2 years FU of the HypoFocal-Phase II trials shows no significantly increased GU and GU toxicities compared to baseline symptoms. Patients reported about a good QoL but increased bowel symptoms after 2 years, particularly if treated with EBRT only. Implementation of PSMA-PET into focal dose escalated radiotherapy approaches appears safe and feasible. However, radioproctitis demands careful management. The current PSA values suggest a highly effective therapy, but longer FU is needed to evaluate oncological outcomes. The HypoFocal-SBRT phase III trial will evaluate the PSMA-PET and mpMRI-based focal dose escalated SBRT. [ABSTRACT FROM AUTHOR]

Poster: Physics track: Intra-fraction motion management PO-1653: Dosimetric effects of interfractional variations in prostate cancer radiotherapy T. Bostel, I. Sachpazidis, M. Splinter, N. Bougatf, T. Fechter, C. Zamboglou, O. Jäkel, P. Huber, D. Baltas, J. Debus, N. Nicolay. [Extracted from the article]

Poster: Clinical track: Prostate PO-1157: Short-term results of coverage probability radiotherapy of MRI-staged prostate cancer J. Marzec, R. Eisenmann, F. Paulsen, D. Wegener, B. Frey, C. Zamboglou, A. Othman, K. Nikolaou, M. Alber, O. Dohm, D. Thorwarth, D. Zips, A. Müller. [Extracted from the article]

Proffered Papers: Proffered papers 10: Prostate OC-0208: Comparison between 68Ga-PSMA-PET/CT and mpMRI for radiotherapy planning in primary prostate cancer S. Spohn, C. Jaegle, A.S. Bettermann, S. Kiefer, C.A. Jilg, J. Kranz-Rudolph, T.F. Fassbender, P. Bronsert, N.H. Nicolay, M. Bock, J. Ruf, M. Benndorf, A.L. Grosu, C. Zamboglou. [Extracted from the article]