A non-invasive approach for improved differentiation of adrenocortical adenoma subtypes using combined radiomic and clinico-radiological modeling on abdominal contrast-enhanced CT images

In this retrospective multicenter study, 169 patients who underwent adrenalectomy and contrast-enhanced abdominal CT were analyzed. A comprehensive clinico-radiological evaluation was performed for all adrenal lesions, and radiomic features were extracted from volumetric segmentation. Feature selection was conducted using SVM and LASSO algorithms, followed by the construction of three different models: radiomic, clinico-radiological, and combined. The combined model demonstrated the highest accuracy in differentiating adrenocortical adenoma subtypes, highlighting its potential value as a non-invasive preoperative diagnostic tool.