Machine Learning Approaches to Support Medical Imaging Diagnosis of Pancreatic Cancer – A Scoping Review

The early diagnosis of pancreatic cancer through medical imaging modalities could be an important breakthrough in the increasing of the survival rate. The purpose of this scoping review was to systematize the application of machine learning models to assist the medical imaging diagnosis of pancreatic cancer. An electronic search was conducted on PubMed, Scopus, Web of Science and Association for Computing Machinery, and 20 studies were included in this review after the selection process. Eleven different machine models were identified in the included studies, and, among these, convolutional neural network (CNN) the most referred (i.e., six studies). In general, the included studies present high values in terms of accuracy, sensitivity, and specificity of the machine learning algorithms. However, the included studies only considered retrospective data. This means that randomized clinical trials are required to translate machine learning implementations to clinical practice.