ESTRO 2022 - Abstract Book

S1120

Abstract book

ESTRO 2022

Our results support the feasibility of SIB-IMRT in treatment of SCCAC, with acceptable toxicity incidence and favorable outcome results. Active smoking and omission of chemotherapy negatively impacted LRC, while N staging correlated with metastatic relapse risk. Increased severe skin toxicity was associated with the primary tumor stage, requiring careful treatment surveillance in patients with T3-4 disease.

PO-1325 Automated rectal tumor segmentation with inter-observer variability-based uncertainty estimates

L. Weishaupt 1 , T. Vuong 2 , A. Thibodeau-Antonacci 3 , A. Garant 4 , K. Singh 5 , C. Miller 6 , A. Martin 7 , F. Schmitt-Ulms 8 , S.A. Enger 1 1 McGill University, Medical Physics Unit, Department of Oncology, Montréal, Canada; 2 Jewish General Hospital, Department of Oncology, Montréal, Canada; 3 McGill University , Medical Physics Unit, Department of Oncology, Montréal, Canada; 4 UT Southwestern Medical Center, Department of Radiation Oncology, Dallas, USA; 5 McGill University Health Centre, Division of Gastroenterology, Montréal, Canada; 6 McGill University Faculty of Medicine, Department of Medicine, Montréal, Canada; 7 Centre hospitalier universitaire de Québec, Department of Radiation Oncology, Quebec City, Canada; 8 McGill University, Department of Computer Science, Montréal, Canada Purpose or Objective Without biopsies, the task of tumor detection carries an intrinsic uncertainty. However, deep learning models that are used for automatic tumor detection, are typically trained to classify pixels as either tumor or non-tumor, disregarding the uncertainty. This study aims to develop a deep learning-based method that can model this uncertainty. Materials and Methods Three gastrointestinal physicians and radiation oncologists from three different institutions contoured the tumor regions in 1704 endoscopy images from 21 patients comprising 101 endoscopic exams. Not all images contained tumors. A deep learning model was trained to classify pixels that all observers classified as tumor or non-tumor. Regions with inter- observer variability were considered uncertain. For training and testing purposes images from 80 exams were used for training, while 21 exams were used for testing, which made up 1392 and 312 images respectively. A soft dice loss function was used to train a deeplabv3 model in PyTorch. Images were resized to 512x512 pixels and normalized before being passed into the model. To increase the model’s robustness, the training set was augmented during each epoch using random rotations and random horizontal and vertical flipping. The model’s performance was evaluated for both the tumor and non-tumor class on the test set. Regions with inter-observer variability were not considered in these scores. Thus, for each pixel outside of the regions of inter-observer variability, the ground truth classification was compared to the class with the greater probability from the model's prediction. Results A representative example of the model’s performance is displayed in Figure 1 and the model’s performance on the test set is shown in Table 1.