The phrases machine learning, deep learning, and artificial intelligence (AI) have entered practically every field of medicine. These approaches have advanced to the forefront of medical imaging research across the board, from image reconstruction to image processing to automated analysis. However, the effects of AI in gynecologic imaging have been less noticeable than in other fields, such as brain and breast imaging. This review article aims to introduce readers to AI ideas with clinical relevance, outline computer vision techniques, and highlight previous research on AI-based picture classification tasks in gynecologic imaging. Multiple databases were searched from their creation to March 18th, 2021 (English language only). Ovid MEDLINE(R), Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, ClinicalTrials.gov, and Epub Ahead of Print, In-Process, and Other Non-Indexed Citations were searched. Researchers conducted a comprehensive literature analysis, selecting 61 papers with the help of 3 reviewers and then having experts classify them according to predetermined inclusion and exclusion criteria. Endometrial, cervical, and ovarian cancers are the most common types of gynecologic malignancies; hence, they provide a literature review organized by these subtypes. Each article is prefaced with a few sentences summarizing the various artificial intelligence (AI) techniques, imaging techniques, and clinical parameters discussed throughout. Investigators wrap up by discussing recent advances, trends, and limitations, and they offer suggestions for future research.
Source: sciencedirect.com/science/article/pii/S0090825822004966
