Radiomics: a new tool to differentiate adrenocortical adenoma from carcinoma

Area of Science:

• Radiology
• Oncology
• Medical Imaging

Background:

• Differentiating benign from malignant adrenal tumors is challenging, especially for incidentally discovered masses.
• Imaging techniques often lack precision in defining the nature of adrenal lesions without clear invasion or metastases.

Purpose of the Study:

• To investigate the potential of radiomics to predict malignancy in adrenocortical tumors.
• To assess the efficacy of CT texture analysis combined with machine learning for adrenal tumor characterization.

Main Methods:

• CT images from 19 patients with resected adrenocortical tumors and 11 patients with incidentalomas under surveillance were analyzed.
• Radiological features (first-order and second-order) were extracted from unenhanced, arterial, and venous phases.
• Unsupervised machine learning (K-means clustering) was applied to the extracted features.

Main Results:

• Several radiomic features in different CT phases significantly differentiated adrenocortical carcinoma from adenoma (P < 0.050).
• The unsupervised machine learning approach correctly predicted malignancy in 7 of 8 adrenocortical carcinomas across all phases, after excluding outliers.
• CT texture analysis demonstrated a high capability in discriminating malignant from benign adrenocortical tumors.

Conclusions:

• Radiomics, utilizing CT texture analysis and unsupervised machine learning, effectively discriminates between malignant and benign adrenocortical tumors.
• This pilot study suggests radiomics is a promising tool for improving the diagnostic accuracy of incidentally discovered adrenal masses.