Quality Assurance Tools for Intraoperative 3D-Navigation

3D-navigation intra-operatively provides a surgeon with information about the position of a tool or a probe inside the patient with respect to some preoperative radiologic imagery. Clinical intraoperative 3D-navigation is severely impeded by the lack of adequate and efficient means to assess the usability and the clinical performance of the system. Costly 3D-navigation equipment is often operated under suboptimal conditions and surgeons hardly do trust the 3D-navigation. Therefore, systems are rarely used, considerable expertise and manpower is wasted and the patient does not benefit from this technology. Quality assurance is necessary to control the whole process of 3D-navigation, comprising a stable and reliable patient-to-image-registration, the goodness of intraoperative information presentation and means to provide a "figure of merit" for the whole process for the surgeon: the spatial orientation of the application error. Thus central goals of this project are: Creation of a temporal continuity of the patient-to-image-registration User-friendly and efficient visualization of surgical information Provision of first intraoperative quality assurance. Based on our own experience in clinical 3D-navigation, a new paradigm for patient-to-image-registration needs to be developed. Intraoperative ultrasound-A mode surfaces of the cranium are promising candidates that avoid human operator errors in the physical registration process, i.e. digitizing and correlating landmarks of the patient with appropriate points in the imagery, respectively. An open-source 3D-navigation system will be made available as a test-bed for the creation and evaluation of "useful" information visualization strategies and quality assurance. A stochastic estimator for quality assurance will be investigated. The lateral and anterior skull bases are, due to the absence of soft-tissue-shifts, ideally suited for navigation. This demonstrator will benefit from the close cooperation with the in-house surgeons to provide a surgeon-driven technological development, in-depth technological and surgical evaluation in a laboratory setting by cadaver experiments. Monoscopic and stereoscopic image overlays of surgically relevant information in Ear-, Nose-, and Throat surgery for 3D-navigated endoscopic operations will be analyzed by surgeons and technologically. Psycho-physical principles of awareness-attraction and computer graphics will be investigated to optimize the quality of intraoperative information presentation and perception. The project is set up in the Innsbruck ENT clinic with abundant surgical expertise in the group and aims at increasing the impact of 3D-navigation in endoscopic navigated surgery.

For intraoperative navigation it is crucial to precisely pinpoint registration loci on the patient and in the corresponding radiologic imagery. This project quantitatively revealed that only implanted registration fiducials can exclude the human error in the registration step. Anatomical landmarks can only be localized with a precision of ~4 mm, limiting the use of Computer-Assisted Surgery (CAS) systems to rough orientation guides. During operations in delicate anatomical regions surgeons benefit from the guidance provided by CAS systems. Such systems show the current position of a tracked tool in the three-dimensional medical imagery of the patient as a cross-hair. Ear-, Nose- and Throat surgery, specifically operations at the anterior skull base and the paranasal sinuses, and Neurosurgery are key disciplines that frequently use CAS systems to aid surgeons. In order to be of value, the surgeons need to know how "accurate" the navigation is. The clinical application accuracy needs to be determined fast and reliably so as to permit a safe use of the system: the positions as shown by a CAS system have to be correct for surgical guidance in anatomical areas, where disease can mask the border between safe und dangerous zones. The patient and the tracked probe (or tool) are localized in space typically with optical position sensing triangulation of active or retro-reflective spheres. The surgery at the anterior skull base is a unique setting for navigated surgery, as the surgical site is everywhere enclosed by bony walls. These walls may be very thin, but permit the use of simplest registration algorithms that guarantee correspondence of intraoperative anatomy with the preoperative radiologic imagery. Computerized Tomography is the imaging modality of choice in surgery of the paranasal sinuses; it shows these bony borders of the surgical with high spatial fidelity and without soft-tissue deformation. A translation and a rotation as the simplest mathematical approach maps patient anatomy as localized with the optical tracker to the preoperative 3D image stack. This registration step depends on the properties of the tracker, the imagery used and of the user. The latter issue is known but has up to now an unknown influence on the application accuracy. It is frequently suspected that human errors in localizing features in the 3D-imagery and on the patient is crucial, but are still unknown. The application accuracy can be predicted theoretically in closed form and is frequently used in image-registration and navigation. It is widely used but to date no experimental validation of this first-order perturbation theoretic approach was done. In order to study this prediction in an intraoperative setting it is necessary to replicate surgery in an experimental operating room and to possess detailed knowledge of all technological parameters that govern the application accuracy of a CAS system.