Prediction of Whole-Body Tissue Composition from Regional Sub-Body CT Scans

Area of Science:

• Radiology and Medical Imaging
• Biomedical Engineering
• Human Physiology

Background:

• Accurate body composition assessment is crucial for understanding human physiology and health risks.
• Traditional methods like L3 vertebral segmentation have limitations in capturing whole-body tissue distribution complexities.
• Factors such as genetics, metabolism, and environment influence body composition.

Purpose of the Study:

• To develop and validate a novel methodology for estimating whole-body composition using computed tomography (CT) scans of sub-body anatomical regions.
• To quantify skeletal muscle (SKM), subcutaneous adipose tissue (SAT), visceral adipose tissue (VAT), and intermuscular adipose tissue (IMAT) using CT data.
• To assess the predictive performance of models using different anatomical region combinations.

Main Methods:

• Utilized whole-body CT scans from 101 cancer patients.
• Developed in-house segmentation software to quantify SKM, SAT, VAT, and IMAT in defined sub-body regions (e.g., chest, abdomen, pelvis).
• Employed a multivariate linear regression model with tenfold cross-validation to predict whole-body tissue volumes, evaluating with MAE, RMSE, MAPE, and R².

Main Results:

• The novel method accurately predicted whole-body tissue volumes, with higher R² values and lower errors compared to single-region methods.
• Visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT) showed significant improvements with broader anatomical coverage (e.g., combined regions).
• Skeletal muscle (SKM) demonstrated strong performance across regions, particularly in combined chest and abdomen (CHA) scans (R²=0.924).

Conclusions:

• Estimating whole-body composition using CT scans of sub-body anatomical regions offers a more comprehensive and accurate approach than traditional methods.
• Broader anatomical coverage in CT scans significantly enhances the prediction accuracy for key body tissues like VAT and SAT.
• This methodology holds promise for improved personalized medical strategies and a deeper understanding of human physiology.