AI-Driven Clinical Decision Support System for Automated Ventriculomegaly Classification from Fetal Brain MRI

Area of Science:

• Medical imaging analysis
• Artificial intelligence in healthcare
• Neurodevelopmental disorders

Background:

• Fetal ventriculomegaly (VM) involves enlarged fetal brain ventricles, often leading to developmental issues.
• Manual segmentation and classification of fetal brain MRI are labor-intensive and require specialized expertise.
• Accurate and efficient diagnostic tools are crucial for timely intervention in fetal brain abnormalities.

Purpose of the Study:

• To develop an automated pipeline for fetal brain ventricle segmentation and VM severity classification.
• To improve the efficiency and accuracy of diagnosing fetal ventriculomegaly using deep learning.
• To provide interpretable clinical reasoning for AI-driven diagnostic support.

Main Methods:

• An adaptive slice selection strategy was employed to identify the most informative 2D MRI slices from 3D volumes.
• Deep learning models were utilized for segmenting lateral ventricles and deep gray matter, estimating ventricular width, and classifying VM severity.
• An explainability module integrated MRI slices, segmentation masks, and predicted severity for clinical interpretation.

Main Results:

• The automated system achieved high segmentation performance with Dice scores of 89% for 2D and 87.5% for 3D models.
• The VM severity classification network demonstrated strong performance with 86% accuracy and an F1-score of 0.84.
• The decision support system provided robust and interpretable results for VM analysis.

Conclusions:

• The developed automated pipeline offers a robust and efficient solution for fetal ventriculomegaly analysis.
• This AI-driven approach can significantly aid clinicians in the early detection and management of fetal brain abnormalities.
• The integration of explainability enhances trust and clinical utility of automated diagnostic systems in neuroimaging.