CT-Based Cerebral Laterality Index and Deep Learning-Assisted Cardiac Morphometry: A Proof-of-Concept Pilot Analysis in a Matched Imaging Cohort
TL;DR
Global CT-derived cerebral laterality was not associated with static cardiac volumetric phenotypes in this small pilot cohort, and the age- and sex-related subgroup findings should be regarded as preliminary, biologically plausible observations that require confirmation in larger standardized cohorts.
Abstract
Background and Objective: This proof-of-concept pilot study aimed to evaluate whether CT-derived structural cerebral asymmetry is associated with deep learning-assisted cardiac volumetric morphometry in a matched imaging cohort. Materials and Methods: The cohort included 17 subjects with paired cranial and thoracic CT data. Brain metrics included intracranial volume (ICV), parenchyma-to-ICV ratio (PAR/ICV), and laterality index (LI). Cardiac metrics were obtained using a pretrained nnU-Net/TotalSegmentator-based inference pipeline and included LV, RV, LA, RA, myocardium, aorta, pulmonary artery volumes, and RV/LV ratio. Spearman correlation was used to test the association between LI and cardiac variables. Results: Mean LI was 0.017 ± 0.059, mean ICV was 1210.831 ± 187.943 mL, and mean PAR/ICV was 0.908 ± 0.028. No statistically significant correlations were found between LI and any cardiac volume metric (all p>0.05). LV and myocardial volumes were significantly higher in males (p=0.027), while PAR/ICV, RV/LV ratio, and aortic volume differed across age bands (p<0.05). Conclusion: Global CT-derived cerebral laterality was not associated with static cardiac volumetric phenotypes in this small pilot cohort. The age- and sex-related subgroup findings should be regarded as preliminary, biologically plausible observations that require confirmation in larger standardized cohorts.