Deep Learning Radiomic Analysis of MRI Combined with Clinical Characteristics Diagnoses Placenta Accreta Spectrum and...

Area of Science:

• Radiology
• Artificial Intelligence
• Medical Imaging

Background:

• Placenta accreta spectrum (PAS) subtypes present varied surgical risks.
• Machine learning (ML) models show promise for diagnosing PAS disorders.

Purpose of the Study:

• To develop a cascaded deep semantic-radiomic-clinical (DRC) model for diagnosing PAS and its subtypes.
• Utilize T2-weighted MRI for enhanced diagnostic accuracy.

Main Methods:

• Retrospective study of 361 pregnant women with suspected PAS.
• Developed a DRC model integrating radiomics, deep semantic features, and clinical data.
• Evaluated model performance against radiologists and other ML models using AUC and ACC metrics.

Main Results:

• The DRC-1 model demonstrated superior performance in PAS diagnosis (AUC 0.850 internal, 0.841 external).
• DRC-2 model achieved comparable performance to radiologists in classifying PAS subtypes.
• The model incorporates clinical data like uterine surgery history and placenta previa presence.

Conclusions:

• The DRC model provides efficient and highly sensitive diagnostic capabilities for PAS.
• This AI-driven approach can significantly aid in preoperative surgical planning for PAS cases.