A MODEL-BASED APPROACH TO INTRAOPERATIVE GUIDANCE OF FLEXIBLE ENDOSCOPY

Thursday 18 Dez. 2003, 12-1pm, (MI 02.13.010)
invited talk by:

Markus Kukuk, PhD
Siemens Medical Solutions
Stanford University Medical Center
Department of Radiology

A Model-Based Approach to Intraoperative Guidance of Flexible Endoscopy

Endoscopy is a minimally-invasive, medical procedure, where a physician uses an optical instrument called an “endoscope'' to obtain a magnified view of the inner surface of hollow organs and to access the tissue through surgical tools. An endoscope is a flexible tube, which is inserted into the body through either natural body openings or small incisions. Flexible endoscopy often results in a better outcome for the patient as opposed to open surgery. However, this technique also presents increased challenges to the physician and therefore often necessitates intraoperative guidance.

This dissertation presents a new approach to the intraoperative guidance of flexible endoscopy. It proposes to calculate a patient specific “protocol'' prior to the intervention to achieve a sensor-less guidance during the procedure. This protocol prescribes in detail how to handle the endoscope and tools in order to successfully perform an endoscopic procedure. During the intervention, the physician executes the protocol, by setting endoscope and tools to the prescribed configuration.

The calculation of the protocol is based on three components: (1) A 3D model of the target anatomy derived from a CT/MRI scan of the patient, (2) a deformable model representing the endoscope and (3) a virtual endoscopy system. These components are combined to simulate an endoscopic procedures and to estimate a set of endoscope parameters. This general approach is implemented and validated for an endoscopic procedure called “Transbronchial Needle Aspiration'' (TBNA), which involves the “blind'' placement of a needle into a target. Based on a set of candidate shapes for the real endoscope obtained from the endoscope model, an “optimal'' needle placement strategy is presented that maximizes the probability of success for TBNAs.