Intracranial aneurysm (IA) is a potentially fatal cerebrovascular disorder related to abnormal arterial dilation and rupture leading to subarachnoid bleeding in case it is not diagnosed. The diagnosis and treatment planning of aneurysms require proper detection, segmentation, and rupture risk evaluation. Deep learning (DL) has revolutionized neurovascular imaging, but to ensure its performance in the real world, it is necessary to move beyond simple detection methods toward more sophisticated and high-precision analysis. This survey offers an in-depth analysis of DL-based innovations in IA analysis across several imaging modalities, such as digital subtraction angiography (DSA), 3D rotational angiography (3DRA), computed tomography angiography (CTA), and magnetic resonance angiography (MRA). In this review, we give a brief summary of key publicly available IA datasets and the latest state-of-the-art methodologies ranging from convolutional neural networks (CNNs) to emerging architectures such as vision transformers (ViTs), state space models (Mamba), and geometric deep learning (GDL), with a special emphasis on their major contributions and methodological innovations to IA analysis. In addition to conventional architectures, this review examines the transition to next-generation architectures, such as foundation models (FMs), multimodal rupture risk prediction frameworks and physics-informed neural networks (PINNs) that incorporate vascular hemodynamics. This survey outlines the existing issues and possible gaps in the field of IA analysis. We conclude by describing future research directions for each of the challenges identified with the aim of reducing the gap between algorithmic advances and real-world clinical deployment.
Kanadath et al. (Mon,) studied this question.