Journal of NeuroInterventional Surgery, E-107 Multi-center study of a deep learning model for intracranial aneurysm detection in computed tomography angiography

This study investigates the use of advanced techniques and technologies for the detection and treatment of intracranial aneurysms (IAs). The research presents a multi-center evaluation of a deep learning algorithm (DLA) designed for IA detection in computed tomography angiography (CTA) scans, emphasizing its robust performance across diverse imaging protocols and scanner types. The DLA demonstrated a lesion-level sensitivity of 0.72, with variations influenced by scanner training data. Results highlight improved detection of aneurysms larger than 3 mm in internal carotid and anterior communicating arteries. Additionally, the study reviews the technical nuances of telescoping pipelines for aneurysm treatment, identifying key procedural challenges, including access, sizing, and apposition, that contribute to higher-than-expected complication rates. This comprehensive analysis underscores the potential of DLAs in enhancing IA diagnosis and the need for technical expertise in complex aneurysm treatments, paving the way for improved patient outcomes in both diagnostic and therapeutic settings.