Abdominal CT-derived metabolic phenotypes are associated with the MRI-derived brain age gap in women

Figures

• 

  Figure 1. Representative abdominal CT image and automated body-composition segmentation. (A) Representative axial abdominal CT image. (B) Corresponding automated segmentation map illustrating the abdominal body-composition compartments used in this study, including visceral fat, subcutaneous fat, skeletal muscle, and internal organs. FatRatio was calculated as the sum of visceral and subcutaneous fat area divided by the sum of visceral fat, subcutaneous fat, and skeletal muscle area. AVFRatio was calculated as visceral fat area divided by the sum of visceral and subcutaneous fat area. Muscle attenuation was derived from the segmented skeletal muscle compartment. Liver-to-spleen attenuation ratio was calculated as the mean liver attenuation divided by the mean spleen attenuation using automatically segmented organ masks on virtual non-contrast CT images.

• 

  Figure 2. Selected CT-derived metabolic phenotypes associated with the brain age gap. Standardized regression coefficients are shown for the associations between CT-derived phenotypes and BAG. Blue estimates are adjusted for age and parity category. Red estimates were additionally adjusted for BMI, hemoglobin A1c, triglycerides, HDL cholesterol, smoking, and alcohol use. Higher FatRatio and lower liver-to-spleen attenuation ratio were associated with a higher BAG in clinically adjusted models. AVFRatio was defined as abdominal visceral fat area divided by total abdominal fat area, calculated as abdominal visceral fat area/(abdominal visceral fat area + subcutaneous fat area).

• 

  Figure 3. Incremental explanatory value of CT-derived phenotypes. Δ adjusted R2 values are shown for CT-augmented models compared with the clinical base model. The clinical base model included age, parity category, BMI, hemoglobin A1c, triglycerides, HDL cholesterol, smoking, and alcohol use. This figure should be interpreted as an explanatory model comparison rather than evidence of clinical prediction performance.