Yan Zhuang, Abhinav Suri, Tejas Sudharshan Mathai, Brandon Khoury, Ronald M Summers
J Imaging Inform Med. 2025 Apr 30. doi: 10.1007/s10278-025-01473-y. Online ahead of print.
CT-based imaging biomarkers can be derived from the pancreas for detecting pancreatic pathologies. However, current approaches using full pancreas segmentations are unable to provide region-specific biomarkers that are crucial in predicting disease severity for many conditions, such as pancreatic adenocarcinomas. This study aims to develop an automated 3D tool to detect and segment the pancreatic sub-regions (the head, body, and tail) on CT volumes. This retrospective study used a subset of 549 CT volumes from the publicly available TotalSegmentator (TS) dataset. The dataset was randomly split into training (n = 440) and testing (n = 109) subsets. Additionally, 30 CT volumes from the TCIA NIH Pancreas-CT dataset were used for external validation. A 3D full-resolution nnUNet model was trained with a custom loss function to detect the landmarks corresponding to the pancreas's head, body, and tail. Based on the detected landmarks, a post-processing algorithm generated the sub-region segmentations. We evaluated the predicted segmentation against the ground truth masks using the Dice similarity coefficient (DSC) and Normalized Surface Distance (NSD). The mean�std of DSC (%) and NSD (%) for the head, body, and tail were 90.8�4.1 and 94.5�4.6, 83.3�7.6 and 87.2�7.4, and 85.1�9.8 and 89.7�8.8, respectively. On the external dataset, the mean�std of DSC and NSD for the head, body, and tail were 83.4�2.6 and 89.7�4.1, 79.4�5.9 and 88.5�6.0, and 81.2�5.5 and 91.4�5.3, respectively. The proposed model can accurately segment three pancreas sub-regions and enables imaging biomarkers to be derived from each sub-region and the pancreas as a whole.