Microcatheter Shaping for Intracranial Aneurysm Coiling Using Artificial Intelligence-Assisted Technique

Abstract

Abstract INTRODUCTION A precise microcatheter shaping is critically necessary for the successful coiling of intracranial aneurysms. Currently, the development of artificial intelligence (AI)-based on technologies in medicine in advancing rapidly. The use of AI may have practical values in aiding the endovascular treatment of intracranial aneurysms. In this study, we tested the hypothesis that application of AI for analyzing parameters from three dimensional images, could warrant more accuracy and stability in microcatheter shaping during the treatment of intracranial aneurysms. METHODS Based on the 3-dimensional (3D) images of the digital subtraction angiography (DSA), we first segmented the vessels and the target aneurysms using a deep-learning-based method. Then the centerline of all vessels was extracted using a 3D thinning algorithm and the surface was calculated from the segmented binary maps. The simulated microcatheter path was then displayed in a 3D fashion together with the segmented vessels and aneurysms to assist the surgical planning and training process. After path simulation, the tip of the path was extracted and used to guide the steam-based shaping. The surgeons could then refer to the guidance during actual shaping. RESULTS A total of 22 patients with 22 aneurysms were treated with this technique. The average diameter of these aneurysms was 5.2 ± 2.9 mm. All of preplanned microcatheters matched vessel structures and aneurysm anatomy. Appropriate positioning was achieved in 15 (68.2%)patients without microguidewire assistance. In 5 cases (22.7%), microguidewire was used for microcatheter navigation and microcatheter reached the appropriate position. In 2 cases (9.1%) of ophthalmic aneurysm in the dorsal side of internal carotid artery, the microcatheter required modification for the initial shaping with reference to AI. During the operation, only 1 microcatheter (4.5%) required repositioning due to kick back during coil deployment. CONCLUSION AI-assisted microcatheter shaping techniques may be promising for faciliating easier and safer procedures in endovascular coiling of intracranial aneurysms.