Photon-counting CT characterization of carotid perivascular adipose tissue: a layer-by-layer quantitative analysis. A preliminary analysis in an asymptomatic population

Figures

• 

  Fig. 1. PVAT segmentation using concentric annular rings from the carotid wall. Axial cross-sections of the right carotid artery demonstrate the sequential expansion of perivascular concentric layers, ranging from 1 mm to 5 mm (a–e). Each image shows a jet color heatmap superimposed on the segmented PVAT, where attenuation values are visualized in a color gradient. The red circular contour marks the vessel wall, and the colored halo depicts the voxel-wise HU distribution. f The entire PCCT matrix

• 

  Fig. 2. PVAT mean HU and Circle HU values. a Boxplot representation of the overall PVAT mean attenuation values across five concentric annular thicknesses (1–5 mm). The red dashed line represents the mean value at each layer. b Comparison of mean PVAT attenuation values stratified by side (right vs left). c Boxplot distribution of Circle HU values, defined as the mean HU within each discrete 1-mm annular layer. d Comparison of Circle HU values by side

• 

  Fig. 3. Subject-wise variability in Circle HU values. a Heatmap representing the ΔHU between each concentric layer (2–5 mm) and the 1 mm reference layer for each subject. b Radar plots showing HU values across the five layers for four subsets of subjects (1–5, 6–10, 11–15, 16–20). Distinct attenuation profiles emerge, with some subjects exhibiting linear trends and others showing nonlinear or inverted behavior

• 

  Fig. 4. Voxel count analysis across circles. a Boxplot of voxel count distribution for each concentric circle (1–5 mm). b Line plot of the mean voxel count with superimposed 95% confidence intervals and standard deviation. c Comparison of voxel count distribution between right and left sides. d Combined bar and line plot showing the variance (blue bars) and coefficient of variation (orange dashed line) of voxel counts for each circle. Both metrics increase with concentric thickness, suggesting growing heterogeneity in segmentation

• 

  Fig. 5. Example of PVAT segmentation with adjacent lymph node. Axial cross-sections illustrate how a lymph node (indicated by white arrows) adjacent to the carotid artery can interfere with perivascular segmentation, particularly at larger concentric thicknesses. In the panel (a), the CTA axial scan with the outer border of the plaque traced. As the segmented layers increase from 1 mm to 5 mm (b–f), the inclusion of non-adipose tissue, specifically the lymph node, results in visible distortion of the perivascular region. The jet heatmap demonstrates heterogeneous attenuation patterns that differ from the typical PVAT signature, especially in outer circles (d–f). This highlights how anatomical complexity can influence voxel composition and impact on quantitative measurements of attenuation and volume