Related Experiment Video
Updated: Sep 17, 2025
13:35
Segmentation and Linear Measurement for Body Composition Analysis using Slice-O-Matic and Horos
Published on: March 21, 2021
10.8K
Improved muscle and fat segmentation for body composition measures on quantitative CT
Jianfei Liu1, Praveen Thoppey Srinivasan Balamuralikrishna2, Sovira Tan3
1Imaging Biomarkers and Computer-Aided Diagnosis Laboratory, Clinical Center, National Institutes of Health, 10 Center Dr, Bethesda, MD, 20892, USA. jianfei.liu@nih.gov.
Summary
A new segmentation method accurately analyzes body composition on bone densitometry CT scans, improving muscle and fat segmentation for better health risk assessment.
Area of Science:
- Medical Imaging
- Radiology
- Artificial Intelligence in Medicine
Background:
- Body composition analysis using abdominal CT scans aids in opportunistic screening and prognostic insights for mortality and cardiovascular risk.
- Current muscle and fat segmentation methods struggle with quantitative CT scans used for bone densitometry, hindering osteoporosis diagnosis and monitoring.
- Accurate body composition analysis is crucial for comprehensive patient assessment and risk stratification.
Purpose of the Study:
- To develop and evaluate an accurate segmentation method for muscle and fat on quantitative CT scans used for bone densitometry.
- To compare the performance of the proposed method against existing segmentation techniques.
- To enhance the utility of bone densitometry scans for body composition analysis.
Main Methods:
- An nnU-Net framework was employed for segmenting muscle, subcutaneous fat, visceral fat, and a general 'body' class.
- Training data utilized CT scans with bone densitometry phantoms, with annotations refined manually from a previous method.
- The method was validated on 980 CT scans across internal and external datasets, including phantom scans, and compared with TotalSegmentator and a prior approach.
Main Results:
- The proposed method demonstrated superior accuracy in segmenting muscle and subcutaneous fat across all tested datasets (p < 0.05).
- Comparable accuracy was achieved for visceral fat segmentation.
- Crucially, no false segmentations occurred within the densitometry phantom on patient scans when compared to TotalSegmentator and the previous method.
Conclusions:
- The developed method significantly improves segmentation accuracy for muscle and subcutaneous fat on bone densitometry CT scans, while maintaining high accuracy for visceral fat.
- This advancement enables reliable body composition analysis directly from scans intended for osteoporosis assessment.
- The findings suggest the method's potential to broaden the clinical applications of quantitative CT in body composition analysis and patient risk stratification.
