Computer‐Assisted Path Planning for Minimally Invasive Vascular Surgery

Abstract

Path planning assisted by two-dimensional medical images is an essential part of minimally invasive diagnosis and treatment for cardiovascular diseases. Due to the complex background of angiography images and intricate vascular structure with multi-branch and stenoses, creating accurate pathways from angiography image is a challenge task. We present a new path planning methodology based on angiography medical images using the steady fluid dynamics. Our novel approach is useful in many medical applications, such as for computer-assisted medical images analysis and the follow-on image-guided interventions. A graph-cuts based energy function was applied to the vessel segmentation of angiography images in order to obtain boundary information. We have adopted Finite volume method (FVM) to simulate the Newtonian fluid inside the segmented blood vessels, and a set of isobars under the steady fluid condition are obtained by Meandering Triangles algorithm. The selected center points of isobars are organized to generate the directed vessels-tree, from which the vascular stenoses are automatically detected and the final surgical path is generated with branches. Our method can be used for quantitative path analysis, and we show experimental results to demonstrate that the versatility and applicability of the algorithm in obtaining single-pixel surgical path with good performance, high accuracy and less manual interventions, especially it is robust on complex vascular structures.