A Deep-Learning Method for the End-to-end Prediction of Intracranial Aneurysm Rupture Risk

Peiying Li,Yongchang Liu,Jiafeng Zhou,Shikui Tu,Bing Zhao,Jieqing Wan,Yunjun Yang,Lei Xu

DOI: https://doi.org/10.1016/j.patter.2023.100709

2023-01-01

Patterns

Abstract:It is critical to accurately predict the rupture risk of an intracranial aneurysm (IA) for timely and appropriate treatment because the fatality rate after rupture is 50%. Existing methods relying on morphological features (e.g., height-width ratio) measured manually by neuroradiologists are labor intensive and have limited use for risk assessment. Therefore, we propose an end-to-end deep-learning method, called TransIAR net, to auto-matically learn the morphological features from 3D computed tomography angiography (CTA) data and accu-rately predict the status of IA rupture. We devise a multiscale 3D convolutional neural network (CNN) to extract the structural patterns of the IA and its neighborhood with a dual branch of shared network structures. Moreover, we learn the spatial dependence within the IA neighborhood with a transformer encoder. Our ex-periments demonstrated that the features learned by TransIAR are more effective and robust than hand-crafted features, resulting in a 10% -15% improvement in the accuracy of rupture status prediction.