CTA-based directional texture analysis of calf muscle in peripheral arterial disease: association with runoff score–defined lower-limb arterial disease severity

To investigate whether CTA-based directional texture features of the posterior calf muscles are associated with runoff score–defined lower-limb arterial disease severity in patients with peripheral arterial disease. This retrospective study included 30 patients with peripheral arterial disease and 60 lower limbs. Six consecutive axial slices centred on the maximum posterior calf muscle cross-sectional area were selected from the 15-s and 30-s phases of dynamic CTA. The gastrocnemius and soleus muscles were segmented, and 70 Gy-level co-occurrence matrix texture features were extracted for each phase. Phase-specific L1-penalized logistic regression models were developed using patient-grouped cross-validation and the one-standard-error rule. Internal validation was performed using 500 patient-cluster bootstrap resamples, with the complete model-development procedure repeated within each resample. The final 1-SE LASSO models retained 10 features at 15 s and 9 at 30 s. After bootstrap correction for optimism, the AUC was 0.769 for the 15-s model and 0.695 for the 30-s model. The 15-s model showed numerically higher discrimination, although no formal comparison between phases was performed. CTA-based directional texture features of the posterior calf muscles were associated with runoff score–defined lower-limb arterial disease severity. These findings support further evaluation of directional calf-muscle texture analysis in larger independent cohorts.