Artificial intelligence for the detection of pancreatic lesions

• Published in: International journal for computer assisted radiology and surgery 2022-10, Vol.17 (10), p.1855-1865
• Main Authors: Anta, Julia Arribas, Martínez-Ballestero, Iván, Eiroa, Daniel, García, Javier, Rodríguez-Comas, Júlia
• Format: Article
• Language: English
• Subjects: Artificial Intelligence; Artificial neural networks; Cancer; Computed tomography; Computer Imaging; Computer Science; Health Informatics; Humans; Imaging; Lesions; Magnetic resonance imaging; Medical imaging; Medicine; Medicine & Public Health; Neoplasms; Pancreatic cancer; Pancreatic Cyst - diagnosis; Pancreatic Cyst - pathology; Pancreatic Neoplasms; Pancreatic Neoplasms - diagnostic imaging; Pancreatic Neoplasms - pathology; Pattern Recognition and Graphics; Prospective Studies; Radiology; Review Article; Surgery; Tomography, X-Ray Computed; Vision
• ISSN: 1861-6429; 1861-6410; 1861-6429
• DOI: 10.1007/s11548-022-02706-z

Purpose Pancreatic cancer is one of the most lethal neoplasms among common cancers worldwide, and PCLs are well-known precursors of this type of cancer. Artificial intelligence (AI) could help to improve and speed up the detection and classification of pancreatic lesions. The aim of this review is to summarize the articles addressing the diagnostic yield of artificial intelligence applied to medical imaging (computed tomography [CT] and/or magnetic resonance [MR]) for the detection of pancreatic cancer and pancreatic cystic lesions. Methods We performed a comprehensive literature search using PubMed, EMBASE, and Scopus (from January 2010 to April 2021) to identify full articles evaluating the diagnostic accuracy of AI-based methods processing CT or MR images to detect pancreatic ductal adenocarcinoma (PDAC) or pancreatic cystic lesions (PCLs). Results We found 20 studies meeting our inclusion criteria. Most of the AI-based systems used were convolutional neural networks. Ten studies addressed the use of AI to detect PDAC, eight studies aimed to detect and classify PCLs, and 4 aimed to predict the presence of high-grade dysplasia or cancer. Conclusion AI techniques have shown to be a promising tool which is expected to be helpful for most radiologists’ tasks. However, methodologic concerns must be addressed, and prospective clinical studies should be carried out before implementation in clinical practice.