Ruptures of cerebral aneurysms present high mortality and morbidity rates. The current preferred treatment for rupture prevention is the embolisation of the cerebral aneurysm. This endovascular treatment requires the navigation of a guidewire and microcatheter within cerebral arteries. Platinum coils are then released within the aneurysm to induce thrombosis. This delicate procedure is performed with the use of a fluoroscopic imaging system, which provides clinicians with only two two-dimensional views. However, clinicians must mentally reconstruct the 3D scene. In order to offer a 3D real-time visualisation software for use during interventions, a method for robust guidewire tracking in near real-time on fluoroscopic images is proposed.
The main steps of the method consist of line structure enhancement filtering using an adapted version of Frangi’s vessel enhancement filter, mapping of the image space to a vertical space, minimal path segmentation and endpoint localisation. The method was tested on nine fluoroscopic sequences with a total of 506 images. We obtained a precision of 0.08 ± 0.09 mm in lateral guidewire movement and 1.09 ± 2.24 mm for tip localisation. Guidewire tracking success rate is at 95 % and tip tracking success rate is at 82 %. Mean computational time is 0.006 ± 0.007 seconds per image. These results show that robust guidewire tracking in near real-time with good tracking success is possible. Nonetheless, tip tracking can be improved to achieve better performance. An additional validation process including on-line guidewire tracking should be performed to confirm real-time capabilities.
| Date | 21 May 2010 |
|---|
| Original language | French |
|---|
| Awarding Institution | - École de technologie supérieure
|
|---|
| Supervisor | Jacques A. de Guise (Supervisor) & Gilles Soulez (Co-supervisor) |
|---|
Lau, C. (Author),
De Guise, J. (Supervisor) & Soulez, G. (Co-supervisor),
21 May 2010Student thesis: Master's thesis › Master in Engineering: Engineering