Placenta accreta spectrum: Detection using quantitative analysis of diffusion MRI-based myometrial tractography.

OBJECTIVES: Prenatal diagnosis of the placenta accreta spectrum (PAS) relies on subjective anatomical MRI interpretation and lacks objective quantitative biomarkers. This study aimed to validate diffusion MRI myometrial tractography for PAS detection and compare its performance with conventional visual MRI diagnosis. MATERIALS AND METHODS: This prospective study enrolled 154 pregnant women with suspected PAS. All subjects underwent routine MRI and DTI. Tractography-derived FA, fiber density and ADC were measured. Diagnosis was confirmed by 2023 FIGO criteria; participants were stratified into a 70% training set and independent 30% test set. Statistical analyses included group comparisons, ROC analyses with DeLong's test, and ICC for interobserver reliability. RESULTS: PAS patients exhibited lower FA (p < 0.001), reduced fiber density (p < 0.001) and higher ADC (p = 0.044). The clinically adjusted FA + fiber density model reached an AUC of 0.86 in the training set (conventional MRI AUC = 0.81) and maintained an AUC of 0.82 in the test set, comparable to conventional MRI. Combined parameters yielded an AUC of 0.82 for differentiating PAS grade 1 and grade 2. ICCs (0.788-0.948) indicated excellent interobserver agreement. CONCLUSION: Quantitative diffusion MRI tractography provides objective biomarkers for prenatal PAS evaluation. The combined FA and fiber density model yields stable diagnostic performance and delivers complementary quantitative information to mitigate subjectivity in conventional anatomical-MRI assessment. This proof-of-concept work requires further external multicenter validation before clinical adoption.