Artificial Intelligence Supported Analysis of Anal Sphincter and Levator Ani Muscle Using Medical Imaging Techniques: A Systematic Review.
Authors
Affiliations (5)
Affiliations (5)
- Department of Development and Regeneration, Cluster Urogenital Surgery, Biomedical Sciences, KU Leuven, Leuven, Belgium. [email protected].
- Department of Development and Regeneration, Cluster Urogenital Surgery, Biomedical Sciences, KU Leuven, Leuven, Belgium.
- Clinical Department of Obstetrics and Gynaecology, UZ Leuven, Leuven, Belgium.
- Institute for the care of the mother and child, Prague, Czech Republic.
- School of Biomedical Engineering and Imaging Sciences, King's College London, London, UK.
Abstract
Injuries to the anal sphincter (AS) and levator ani muscles (LAM) are linked to pelvic floor disorders and can be assessed through medical imaging. Artificial intelligence (AI), particularly machine learning and deep learning, is increasingly used in pelvic floor imaging to support clinicians in their diagnostic assessment. However, the quality of AI research in this domain has not been systematically evaluated. Reviewing current AI research and its applications is essential to understand its benefits, as well as its limitations and knowledge gaps. We conducted a systematic search of PubMed, Scopus, Web of Science, Embase, and IEEExplore to identify papers on AI development and/or validation for any AS or LAM-related health conditions. Reporting quality was evaluated using the AI version of the Transparent Reporting of a Multi-variable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD-AI) checklist. We identified 40 studies: 24 using ultrasound and 16 magnetic resonance imaging. Most studies (n = 27) focused on pelvic floor disorders, followed by oncology (n = 8). AI tools often surpassed traditional methods in time efficiency and clinical acceptability. Segmentation was the most common task (n = 25), with UNet-based models dominating (n = 22). Quality assessment showed moderate adherence to reporting standards, with 31 studies meeting 50-75% of TRIPOD-AI items. AI shows potential in assisting pelvic floor imaging, especially for segmentation tasks. However, most models are not yet ready for widespread clinical adoption. Future studies should focus on improving dataset diversity, external validation, and reporting transparency to enable safe and effective clinical translation. PROSPERO (CRD42023464383).