Objectives: Metabolic Syndrome (MetS) is a cluster of metabolic risk factors, which elevate the risk of cardiovascular diseases and mortality.Body composition, especially the muscle and adipose tissue, plays a critical role in MetS development.The objectives were to explore fractal analysis to quantify the spatial distribution pattern of body composition from computed tomography (CT) and combine with clinical data to develop and validate a diagnostic model for MetS.Methods: In this two-center retrospective study, participants were classified into MetS and control groups based on International Diabetes Federation criteria.Clinical and CT images of lower extremities were analyzed.A 3D box-counting fractal analysis on CT images was applied to assess the microstructural complexity of body compositions.Feature selection involved univariate analysis, variance inflation factor assessment and multivariate logistic regression.Model performance was evaluated using receiver operating characteristic, decision curve analysis, and calibration curves.Results: The cohort included 184 subjects from center 1 (split 7:3 into training/internal test sets) and 74 from center 2 for external validation.The final combined model identified five independent predictors: triglyceride (odds ratio (OR) =2.136), history of diabetes (OR=7.774),fractal dimension of intramuscular adipose tissue (IMAT) (OR=3.100),and IMAT multifractal range (OR=3.613),IMAT/muscle (OR=1.927).Model combined clinical and radiological features demonstrated robust discrimination, with area under the curves of 0.932 (training), 0.861(test) and 0.855 (validation).Conclusion: Fractal properties of IMAT derived from CT scans are potent, non-invasive biomarkers for MetS.A diagnostic model integrating radiological features with clinical factors provides excellent, externally validated performance for MetS identification.
Hou et al. (Wed,) studied this question.