A Scoping Review of the Methodologies and Reporting Standards in Recent Applications of Artificial Intelligence in Radiomics for Chronic Subdural Hematoma Imaging.

Chronic subdural hematoma (cSDH) is the accumulation of blood in the subdural space, primarily affecting older adults. Radiomics is a rapidly emerging field that integrates artificial intelligence (AI) with imaging to improve diagnostic precision and prognostic predictions, including hematoma expansion and recurrence. However, the heterogeneous study designs, endpoints, and reporting standards limit its clinical application. This scoping review queried PubMed for studies published before or on December 25, 2024, using terms related to cSDH and AI-based imaging analysis. Inclusion criteria required primary research applying AI to cSDH imaging and reporting prognostic endpoints such as recurrence, expansion, or treatment response. Extracted data included methodological variables, imaging modalities, endpoints of interest, and performance metrics. Most studies used computed tomography (CT) imaging for analysis, with hematoma recurrence being the most frequently evaluated endpoint of interest. However, there was wide inconsistency in the reporting of model performance metrics. Thus, radiomics offers opportunities to improve outcome prediction and treatment planning in cSDH. Future work should focus on defining clinically meaningful endpoints, standardizing metrics, and validating models prospectively to facilitate integration into practice.