Image guided navigation for minimally invasive surgery

In this thesis, we present a catheter navigation system for minimally invasive surgery. The core of such a navigation system is to register pre-operative high resolution 3D or 4D heart model with infra-operative bean shape model to build a combined heart shape. With such a combine shape model the intra-operative catheter motion can be tracked and displayed with it Then clinicians can easily navigate catheters inside human heart. The most difficult part of such a registration is how to quickly and reliably capture infra-cardiac heart shape measurement. We designed a 3D ultrasound catheter "virtual touch" technique which can capture heart shape measurement during operation in a much faster and more accurate way than conventional manual collection method. Our system can use 4D model and points which captures the heart shape change during one cardiac cycle for registration. To address non-rigid shape changes from breathing cycles and other sources, our system has a local non-rigid registration component to further correct the high resolution pre-operative shape model based on infra-operative realtime shape measurement from "virtual touch".