EP183 - ECE_3184 - Artificial intelligence in adrenal imaging: detection and segmentation of adrenal glands

Abstract

Accurate visualization of the adrenal glands is crucial for the evaluation of adrenal pathology. However, their small size, variable shape, and proximity to other retroperitoneal structures make automated analysis challenging. Artificial intelligence (AI) tools capable of detecting and segmenting adrenal glands could significantly improve workflow efficiency and lay the groundwork for future tumor characterization. At the current stage of the project, we aim to develop and validate an AI-based algorithm for automated detection and 3D segmentation of adrenal glands in non-contrast CT scans. This constitutes a foundational step for subsequent modules that will analyze adrenal lesions. A multidisciplinary team from the Medical University of Warsaw and Warsaw University of Technology is collaborating to build a comprehensive dataset of anonymized abdominal CT scans. Using state-of-the-art deep learning architectures, we are training models to precisely identify and segment both adrenal glands, generating accurate three-dimensional reconstructions. The performance of the model is evaluated based on segmentation accuracy and anatomical consistency. The AI algorithm successfully detects and segments adrenal glands in three dimensions, offering a robust and scalable tool for future applications in adrenal imaging. This stage represents a critical foundation for further development of AI systems aimed at lesion detection and classification in adrenal pathology.