Deep learning for pancreas segmentation on computed tomography: a systematic review

Pancreas segmentation has been traditionally challenging due to its small size in computed tomography abdominal volumes, high variability of shape and positions among patients, and blurred boundaries due to low contrast between the pancreas and surrounding organs. Many deep learning models for pancreas segmentation have been proposed in the past few years. We present a thorough systematic review based on the Preferred Reporting Items for Systematic Reviews and Meta-analyses statement. The literature search was conducted on PubMed, Web of Science, Scopus, and IEEE Xplore on original studies published in peer-reviewed journals from 2013 to 2023. Overall, 130 studies were retrieved. We initially provide an overview of the technical background of the most common network architectures and publicly available datasets. Then, the analysis of the studies combining visual presentation in tabular form and text description is reported. The tables group the studies specifying the application, dataset size, design (model architecture, learning strategy, loss function, and training protocol), results, and main contributions. We first analyze the studies focusing on parenchyma segmentation using datasets with only pancreas annotations, followed by those using datasets with multi-organ annotations. Then, we analyze the studies on the segmentation of tumors, cysts, and inflammation. The studies are clustered according to the different deep learning architectures. Finally, we discuss the main findings from the published literature, the challenges, and the directions for future research on the clinical need, deep learning and foundation models, datasets, and clinical translation.