JMIR Publications

ABSTRACT

Background:

The incidence of adrenal incidentalomas is increasing annually, and most types of adrenal masses require surgical intervention. Accurate classification of common adrenal masses based on tumor CT images by radiologists or clinicians requires extensive experience and is often challenging, which increases the workload of radiologists and leads to unnecessary adrenal surgeries. There is an urgent need for a fully automated, non-invasive, and precise approach for the identification and accurate classification of common adrenal masses.

Objective:

Enhance diagnostic efficiency and transform the current clinical practice of preoperative diagnosis of adrenal masses.

Methods:

This study is a retrospective analysis that includes patients with adrenal masses who underwent adrenalectomy from January 1, 2021, to May 31, 2023, at Center 1 (internal dataset), and from January 1, 2016, to May 31, 2023, at Center 2 (external dataset). The images include non-contrast, arterial, and venous phases, with 21,649 images used for the training set, 2,406 images used for the validation set, and 12,857 images used for the external test set. We invited three experienced radiologists to precisely annotate the images, and these annotations served as references. We developed a deep learning-based adrenal mass detection model, named Multi-Attention YOLO (MA-YOLO), which can automatically localize and classify six common types of adrenal masses.

Results:

A total of 516 patients were included. In the external test set, the MA-YOLO model achieved mean average precision (mAP) for the localization of six types of adrenal masses in unenhance, arterial, and venous phase CT images of 0.838, 0.885, and 0.890, respectively. The corresponding mAP for classification were 0.885, 0.913, and 0.915, respectively. Additionally, with the assistance of this model, the classification performance of six radiologists and clinicians for adrenal masses significantly improved.

Conclusions:

The MA-YOLO model shows promise for extensive screening of common adrenal masses in CT examinations, enabling efficient, accurate, and non-invasive preoperative localization and classification.