Implementation of Fully Automated AI-Integrated System for Body Composition Assessment on Computed Tomography for...

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Healthcare
• Radiology and Imaging Informatics

Background:

• Opportunistic computed tomography (CT) screening for sarcopenia and myosteatosis is increasingly important.
• Fully automated artificial intelligence (AI) systems are crucial for widespread opportunistic screening.
• Real-world clinical implementation of AI for body composition assessment in routine health check-ups remains unevaluated.

Purpose of the Study:

• To evaluate the performance and clinical utility of a fully automated AI system for body composition assessment.
• To assess the AI system's integration and effectiveness during opportunistic CT scans in routine health check-ups.
• To determine the system's accuracy, speed, and success rate in a clinical setting.

Main Methods:

• Prospective multicenter study involving 537 patients across 3 institutions.
• AI algorithm for L3 slice selection and end-to-end muscle/fat segmentation integrated with Picture Archiving and Communication System (PACS).
• Assessment of technical success rate, processing time, and segmentation accuracy (Dice similarity coefficient).

Main Results:

• The AI system demonstrated 100% technical success rate with no manual adjustments needed.
• Mean processing time from CT acquisition to report generation was 4.12 seconds.
• Segmentation accuracy reached 97.4% compared to human experts, revealing age-related body composition changes.

Conclusions:

• The fully automated AI system significantly improved opportunistic sarcopenia screening.
• High technical success and segmentation accuracy were achieved without manual intervention.
• The AI system has the potential to revolutionize routine health check-ups with rapid, accurate body composition analysis.