Automated CT Body Composition Analysis

We developed an automated deep learning system for comprehensive body composition analysis from computed tomography (CT) scans, enabling accurate segmentation of muscle and fat tissues throughout the body.

Key Features

• Automated segmentation: End-to-end deep learning pipeline for muscle and fat segmentation in CT images
• Comprehensive analysis: Supports both 2D and 3D body composition measurements
• Clinical integration: Enables automated quantification of body composition metrics for clinical decision support
• High accuracy: Robust segmentation performance across diverse patient populations and imaging protocols

Clinical Applications

• Surgical risk prediction: Integration of body composition metrics improves risk stratification for surgical procedures (e.g., colectomy)
• Nutritional assessment: Automated quantification of muscle mass and fat distribution for nutritional status evaluation
• Longitudinal monitoring: Track body composition changes over time for disease progression and treatment response
• Research applications: Enable large-scale studies on body composition and its relationship to clinical outcomes

Impact

This work demonstrates that integrating automated body composition analysis into clinical models significantly improves surgical risk prediction, highlighting the value of AI-driven body composition metrics in healthcare.