Computed tomography based morphometry of muscle and adipose tissue in a generally healthy Moscow population: a cross-sectional study

Introduction: Alterations in body composition (muscle and adipose tissue) can influence clinical outcomes of underlying diseases. Comprehensive assessment of body composition via instrumental diagnostics includes bioelectrical impedance analysis, dual-energy X-ray absorptiometry, computed tomography (CT), and magnetic resonance imaging (MRI). Although CT and MRI are considered gold-standard methods, their use is limited by challenges in manual segmentation and the lack of universally accepted diagnostic thresholds. Artificial intelligence (AI) can fully automate this process and facilitate the calculation of diagnostic threshold values in cross-sectional population studies. Objective: To obtain data on the sex and age distribution of morphometric parameters of muscle and adipose tissue, assessed using AI-based software on CT scans capturing the region of interest (L3 vertebra level), in a generally healthy population in Moscow. Materials and methods: A retrospective, observational, cross-sectional descriptive study was conducted across 65 medical institutions of the Moscow Department of Healthcare. CT data from generally healthy patients were selected through a multistage process, including assessment of anamnestic data. Participants were grouped by sex and age. Native CT images were analyzed using AI-based software for segmentation and calculation of the following parameters: skeletal muscle area (SMA), skeletal muscle index (SMI), radiological density of skeletal muscle (RDSM), subcutaneous adipose tissue area (SATA), visceral adipose tissue area (VATA), intramuscular adipose tissue area (IMATA), and its proportion (PIMAT). For each parameter, quantile regression models were constructed (5th, 25th, 50th, 75th, and 95th percentiles) with 95% confidence intervals. Statistics: Cut-off values for morphometric parameters were determined using a quantile regression model. T-test, one-way ANOVA, and Mann-Whitney tests were used for comparative analysis. Statistical analysis was performed using jamovi version 2.3.28 and RStudio 2024.12.1. Results: After multi-stage analysis of CT images, text of their corresponding reports and electronic medical records according to inclusion and exclusion criteria, the final sample consisted of 900 patients (579 women (64.3%) and 321 men (35.7%)). SMA and SMI showed similar trends: a gradual increase to peak values followed by a decline, with minima in the «80 years and older» group. SMA, SMI, and RDSM were statistically significantly higher in men (p