Wires segmentation in fluoroscopic images during cerebral aneurysm endovascular intervention

Endovascular intervention is currently considered the treatment of choice for cerebral aneurysms. Nowadays imaging technology for real-time endovascular tool navigation consists of digital fluoroscopy detectors mounted on C-shaped arms with multiple orientations capability. In theory, a 3D representation of interventional devices (guide wire, coils) can be done using biplane fluoroscopic acquisitions. This approach opens the door to a real 3D road-map guidance of endovascular procedure. However, it is important that the tool be precisely identified during an intervention on the low contrast and low resolution fluoroscopic images. Therefore, a new method for segmenting guide wires and coils on fluoroscopic images is presented. A line enhancement algorithm based on Hessian matrix is applied and line candidates are marked by simple thresholding. Line artifacts are suppressed according to geometric criterion. The remaining line segments are linked when appropriately aligned to assemble continuous tools regardless of crossing from the projection of a planar view. This identifying method is evaluated on in vivo fluoroscopic sequences acquired during endovascular interventions on patients with cerebral aneurysm.