Soft Tissue Radiodensity Asymmetry as Clinical Indicator for Diabetes and Hypertension in Aging.
Authors
Affiliations (7)
Affiliations (7)
- Institute of Biomedical and Neural EngineeringReykjavik University 102 Reykjavik Iceland.
- Department of Biomedical ScienceHumanitas University Pieve Emanuele 20072 Italy.
- IRCCS Humanitas Research Hospital Rozzano 20089 Italy.
- Department of Electrical Engineering and Information TechnologyUniversity of Naples Federico II Naples 80125 Italy.
- Biomedical DepartmentUniversity of Padova Padua 35131 Italy.
- Icelandic Gerontological Research InstituteLandspitali University Hospital 101 Reykjavik Iceland.
- Department of ScienceLandspitali University Hospital 101 Reykjavik Iceland.
Abstract
Chronic conditions like diabetes mellitus (DM) and hypertension (HTN) significantly impair physical functioning in older adults, leading to a reduced quality of life and increased risk for disability. This study evaluates the utility of soft tissue radiodensity asymmetry, derived from mid-thigh computed tomography (CT) imaging, as a biomarker for DM and HTN. Using data from the AGES-Reykjavik study, the Nonlinear Trimodal Regression Analysis (NTRA) method was employed to extract 11 geometrical features from fat, muscle, and connective tissue radiodensity distributions derived from cross-sectional CT images of the mid-thigh. Asymmetry indices were calculated as the absolute differences between corresponding left and right leg parameters and were analyzed for their associations with DM and HTN. Statistical analyses demonstrated that DM and HTN status altered the relationship between age and radiodensity asymmetry: DM significantly influenced muscle-specific asymmetry ([Formula: see text] musc), while HTN showed directional trends across multiple tissue types that did not survive correction for sex, BMI, and multiple comparisons. Logistic regression identified fat tissue location asymmetry as a key predictor for DM, while connective tissue asymmetry parameters were significantly associated with HTN status. Machine learning models validated these findings, with Random Forest achieving 89.3% accuracy for DM classification and 88.1% for HTN, highlighting the robustness of these asymmetry features. These results establish soft tissue asymmetry as a promising novel biomarker for DM and HTN, reflecting metabolic and vascular dysfunctions. Integrating such measures into clinical practice could enhance predictive models and inform targeted interventions, improving health outcomes for aging populations.