Search

Leveraging the development of artificial intelligence, in recent years several medical sectors have integrated automated solutions to segment anatomical structures from bioimaging with deep learning approaches. Musculoskeletal System segmentation is key for studying anatomical tissue alterations and supporting medical interventions. The clinical usage of such tools requires understanding the proper way to read their results and evaluate their performance. The current systematic review aims at presenting the common bottlenecks for musculoskeletal structures analysis (e.g., small sample size, data inhomogeneity) and strategies faced by different Authors for similar tool development. Research has been performed on the PUBMED database with the following keywords: deep learning, musculoskeletal system, segmentation. A total of 140 articles published until February 2022 were obtained as query output and analyzed according to PRISMA framework in terms of: anatomical structures, bioimaging techniques, pre/post-processing operations, training/validation/testing subset creations, network architectures, loss functions, performance indicators and so on. Even if from this survey emerged that there are some common trends, the comparison between different methods needs to be discussed based on each specific case-study (anatomical region, medical imaging acquisition setting, study population, etc.). These findings can guide clinicians (as end users) to better understand the potential benefits and limitations of these tools.