A CT-based deep learning model to differentiate between benign and malignant adrenal lesions

Clinical Snapshot

Key Findings

• P Value: Not reported in abstract

• Effect Size: Accuracy 84.2%; AUC 0.93 in the test set (n=113 patients)

• Primary Outcome: Diagnostic accuracy of the best-performing deep learning model (combining clinical, biological, and radiological data) for differentiating benign from malignant adrenal lesions on CT

• Nnt Or Sensitivity: Sensitivity and specificity not reported in abstract; AUC 0.93 (95% CI: 0.899–0.979); segmentation reproducibility Dice coefficient 0.92 ± 0.03 (range 0.72–0.97)

• Confidence Interval: Accuracy 95% CI: 79.9–88.6%; AUC 95% CI: 89.9–97.9%

Clinical Application

Manual segmentation is required, which demands trained radiological personnel and is time-intensive for routine clinical use. Integration into existing PACS/radiology workflows would require significant infrastructure investment. The multimodal input requirement (clinical history, laboratory results, CT data) necessitates coordinated multidisciplinary data access. In Australia, adrenal incidentalomas are commonly encountered in the context of CT performed for unrelated indications, with management guided by RACGP and Endocrine Society of Australia recommendations emphasising biochemical evaluation and size-based surveillance. FDG-PET/CT is available through Medicare for selected indications but not universally funded for adrenal characterisation. Adrenalectomy is performed at specialist centres. This model, if externally validated, could complement existing Australian pathways — particularly in centres managing high-risk patients (known extra-adrenal malignancy, biochemically active lesions). However, TGA approval and prospective Australian validation would be prerequisites for clinical deployment. The model is not currently PBS-listed or TGA-registered as a diagnostic device. Adults with adrenal lesions undergoing CT evaluation at tertiary endocrine or surgical centres, particularly where histopathological confirmation is planned. The model is not yet validated for use in the broader adrenal incidentaloma population managed in general radiology or primary/secondary care settings.

Abstract

OBJECTIVE: To develop a deep learning model to differentiate benign from malignant adrenal lesions. MATERIALS AND METHODS: A total of 380 patients with 385 pathologically confirmed adrenal lesions (101 malignant, 284 benign) were retrospectively included. Adrenal lesions were manually segmented on CT images and analyzed in a deep learning pipeline aimed at differentiating benign from malignant lesions. Four predictive models were developed that incorporated combinations of radiological data (tumor size, and spontaneous attenuation) and non-radiological data (i.e., medical history and laboratory results). Data of 267 patients were used as a training set and those of 113 patients for the test set. The diagnostic capabilities of the four models were estimated using sensitivity, specificity, accuracy, and areas under the receiver operating characteristic curves (AUC) using histopathological findings as the gold standard. The reproducibility of manual segmentation was estimated using the Dice similarity coefficient after blinded resegmentation of 40 adrenal lesions by an independent radiologist. RESULTS: Segmentation reproducibility achieved a mean Dice similarity coefficient of 0.92 ± 0.03 (range: 0.72-0.97). The most accurate model, which combined clinical, biological, and radiological data, achieved 84.2% accuracy (95% confidence interval: 79.9, 88.6) and an AUC of 0.93 (95% confidence interval: 89.9, 97.9) in the test set for diagnosis of malignant adrenal lesion. CONCLUSION: A deep learning model integrating preoperative clinical, biological, and radiological features demonstrates high capabilities in differentiating benign from malignant adrenal lesions on initial CT examination.

References

1. 1.Huang, Z., Dohan, A., Assié, G., Gaillard, M., Violon, F., Jouinot, A., Soyer, P., Bertherat, J., Marini, R., Chassagnon, G., & Barat, M. (2026). A CT-based deep learning model to differentiate between benign and malignant adrenal lesions. European Journal of Radiology, 112907. https://doi.org/10.1016/j.ejrad.2026.112907