Evaluation of surrenal lesions with histogram analysis on computed tomography

ABSTRACT EVALUATION OF SURRENAL LESIONS WITH HISTOGRAM ANALYSIS ON COMPUTED TOMOGRAPHY The aim of this study was to evaluate the effectiveness of histogram-based radiomic parameters derived from contrast-enhanced computed tomography (CT) images in differentiating adrenal lesions, including lipid-poor adenomas, pheochromocytomas, adrenocortical carcinomas and metastases. This retrospective study included a total of 232 patients diagnosed with adrenal lesions who underwent contrast-enhanced abdominal CT scans between 2013 and 2023. The study group comprised 76 metastases, 62 lipid-poor adenomas, 50 pheochromocytomas, and 44 adrenocortical carcinomas. Regions of interest (ROIs) were selected on axial CT images, excluding areas of necrosis, cystic components, calcifications, and major feeding vessels. Histogram-based texture analyses were performed using a custom MATLAB algorithm, and the following parameters were extracted: pixel count, mean, standard deviation, variance, entropy, skewness, kurtosis, uniformity, and percentiles (1st–99th). Statistical analyses were conducted using SPSS version 22, employing Kruskal-Wallis and Chi-square tests, with p<0.05 considered statistically significant. Histogram parameters showed statistically significant differences among the four lesion groups. In particular, the metastasis group exhibited significantly higher HU values, whereas standard deviation and variance values were lower. All percentile values from the 1st to the 99th percentiles reflected the high-density nature of metastases. Entropy values were found to be higher in pheochromocytomas, adrenocortical carcinomas, and metastases, indicating a more complex internal structure. However, skewness and uniformity did not differ significantly between the groups. Histogram-based CT texture analysis appears to be a valuable non-invasive method for the differential diagnosis of adrenal lesions. Specifically, standard deviation, variance, entropy, and percentile-based HU values may serve as distinguishing parameters in identifying metastases from other benign and malignant lesions. These findings suggest that incorporating radiomic data into routine diagnostic workflows may be beneficial; however, further large-scale studies supported by advanced artificial intelligence models are necessary to enhance clinical applicability and standardization. Keywords: Surrenal lesions, computed tomography histogram analysis