Abstract
Atrial fibrillation (AF) is a common supraventricular arrhythmia, characterized by irregular and uncoordinated contraction of the atrium. Although AF itself is not lethal, it accounts for secondary diseases and morbidity and mortality. The minimally-invasive radiofrequency ablation (RFA) procedure is a commonly performed treatment option for AF, guided by Xray fluoroscopy. While it provides high temporal and spatial resolution, X-ray fluoroscopy has poor soft-tissue contrast. In order to enhance image guidance, volumetric roadmaps overlaid onlive X-ray fluoroscopy may be used. However, these roadmaps are often static and do not reflect cardiorespiratorymotion. Real-time roadmap update is important to compensate for this motion and maintain the accuracy of the guidance information, thereby allowing accurate determination
of catheter-based ablation treatment sites. Motion gating is crucial for achieving this accuracy. Four novel and clinically applicable robust-to-noise motion gating techniques are developed and presented in this thesis.
The first proposed technique, Tracked-principal component analysis (PCA), is based on the formation of a novel statistical model of the motion of a coronary sinus (CS) catheter using PCA of tracked electrode locations from monoplane X-ray fluoroscopy images. Motion gating was later further extended to be applicable to more types of minimally invasive procedures, such as
Cardiac resynchronisation therapy, where the CS catheter is not present in the X-ray images.
This is achieved by the development of two robust to varying image-content techniques, the hierarchical manifold learning based and the Masked-PCA techniques. To avoid the limitation of the requirement to build a separate model for each X-ray view, an X-ray system View-angle independent technique was developed based on learning CS catheter motion using PCA and then
applying the derived motion model to unseen images taken at arbitrary projections.
Applications of the motion gating techniques spanning from basic research to clinical tasks are demonstrated. These include catheter reconstruction in 3D from catheter tracking in gated sequential biplane X-ray images that can effectively achieve 2D/3D registration of 3D cardiac data (CT or MRI) to X-ray fluoroscopy; motion gating of 3D rotational X-ray angiography sequences
where the angulation of the scanner is changed between frames; and motion compensation on unseen images taken at any arbitrary projection, by integrating cardiorespiratory motion into MRI-derived roadmaps fused with live X-ray fluoroscopy. This thesis is devoted to the development of robust-to-noise motion gating techniques and their use for improved procedure guidance
while at the same time minimising patient and staff exposure to radiation, by allowing the use of lower-dose fluoroscopy. Results indicate that, within the constraints of acceptable accuracy, the achievable dose reduction factor is 1/25 for the HML-based and Masked-PCA techniques, indicating a dose reduction of more than 25 times, between 1/10 and 1/25 for the Tracked-PCA technique
and around 1/10 for the View-angle independent technique.
| Date of Award | 2015 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Kawal Rhode (Supervisor) & Andrew King (Supervisor) |