MRI-based deep learning-radiomics dual model predicts placenta accreta spectrum with high accuracy

A fusion model using MRI deep learning, radiomics, and clinical features predicted placenta accreta spectrum; test set AUCs were 0.949 (intraoperative criteria) and 0.936 (histopathologic criteria).

• Model development used pooled data from two centers, with separate models for FIGO intraoperative clinical and histopathologic criteria.
• Fusion models combining radiomics, deep learning, and clinical features achieved high diagnostic performance.
• Clinical and MRI features showed correlations with radiomics and deep learning scores, suggesting biological relevance.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.