Abstract
Cigarette smoking may induce oxidative, inflammatory, and microvascular alterations in salivary gland tissue. However, subtle parenchymal changes may not be readily appreciable on conventional ultrasonography. This study aimed to evaluate smoking-related differences in image-derived parotid parenchymal pattern complexity using fractal analysis of B-mode ultrasonographic images. This prospective cross-sectional study included 95 adults: 47 smokers and 48 non-smokers. All participants underwent standardized B-mode ultrasonographic evaluation of both parotid glands. Three predefined regions of interest (ROIs) were analyzed in each gland, and fractal dimension (FD) values were calculated using the two-dimensional box-counting method. Between-group comparisons were adjusted using the Benjamini-Hochberg false discovery rate procedure. Correlation, ROC, multivariable regression, effect-size estimation, and intraobserver reproducibility analyses were also performed. FD values differed between smokers and non-smokers in a region-specific pattern. Significant between-group differences were observed in right ROI 2, left ROI 1, and the mean FD value of the left parotid gland. The mean FD value of the left parotid gland showed the largest effect size and the highest area under the curve. Within the smoker group, left ROI 2 was negatively correlated with age and smoking duration. After adjustment for age and gender, smoking status remained independently associated with right ROI 2, left ROI 1, and the mean FD value of the left parotid gland. Intraobserver reproducibility was high, with intraclass correlation coefficients ranging from 0.87 to 0.92. Quantitative ultrasonographic fractal analysis demonstrated region-specific differences in image-derived parotid parenchymal pattern complexity between smokers and non-smokers. This reproducible, non-invasive approach may complement conventional ultrasonography by providing an objective assessment of subtle smoking-related parotid gland alterations.