Improving System Accuracy in Computer Aided Robotic ORL Surgery B Bell 1 , N Gerber 1 , J Salzmann 1 , E Nielsen 3 , G Zheng 1 , C Stieger 2 , LP Nolte 1 , M Caversaccio 2 , S Weber 1 ARTORG Center for Biomedical Engineering Research, University of Bern 1 Institute for Surgical Technologies and Biomechanics, University of Bern 2 Department ENT Surgery, University Hospital Bern 3 NTB Buchs, University of Applied Sciences Bern INTRODUCTION Reduction of comorbidity and cost has recently ex- tended implementation of minimally invasive surgical techniques into oto-rino-laryngology (ORL) [1]. How- ever, minimally invasive computer aided micro-surgery (CAMS) interventions require an unprecedented level of overall system accuracy. For example, inserting a coch- lear implant in a single drill pass requires the definition of a drill trajectory through the facial recess to access the middle ear. This trajectory (approximately 1.0 – 3.5 mm in width) is bounded posteriorly by the facial nerve and anteriorly by the chorda tympani [2-4]. In order to successfully avoid these structures, an overall CAMS system accuracy of 0.2 -0.3 mm is required. Additional benefits of integrating robotic manipulators in ORL surgery include tremor reduction [5], and augmentation of force dependent tissue interaction [6]. Table 1: Accuracy comparison in CAS and CAMS Error Source Imaging Tracking Regist. Calib. Overall Requirement Voxel size RMS TFE RMS TRE Unit [mm] [mm] [mm] [mm] [mm] CAS 0.50 0.25 1 – 2 0.1 2 – 3 CAMS 0.15 - 0.1 – 0.2 0.01 0.2 – 0.3 In this contribution we discuss ways of achieving this accuracy level, which is an order of magnitude better than current CAS approaches (table 1). Specifically we discuss and present recent findings in: Imaging: A higher image resolution would decrease the partial volume effect and improve the recognition of small anatomical structures. Registration: Hand-picked fiducial marker selections are highly variable and dependent on user skill. Additio- nally, sub-pixel accuracy is required for high precision image registration. Relative Cranial Tracking: Current tracking systems lack sufficient accuracy for ORL CAMS. Herein we dis- cuss alternative approaches to relate cranial movement relative to the robot base. Through this approach, we bypass navigation errors completely. Surgical Manipulator / Robot: Because of the small size scale and proximity of critical anatomic structures, free hand drilling techniques exceed the perception and dexterity of the surgeon to safely circumnavigate these structures. The use of a mechatronic manipulator will greatly enhance surgeon abilities and facilitate precise placement of the implant cavity. MATERIALS AND METHODS Imaging study: Investigation of a suitable imaging modality with respect to available image contrast and reproduction quality was performed with cadaver head scans using Computed Tomography (CT) and Digital Volumetric Tomography (DVT) imaging modalities [7]. Automatic landmark detection: A semiautomatic marker detection algorithm was implemented and com- pared to manual marker detection by quantifying the standard deviation from the results of the two tech- niques. Manually detected marker centers were chosen by identifying the marker center in appropriate scan planes. In contrast, semiautomatic marker locations were chosen by registering 3D solid CAD model data with voxel intensity gradients. The geometric center of the 3D model was then used as the marker location. Both methods were repeated and their results compared to address repeatability. Development of surgical manipulator / robot: To achieve the required accuracy [2, 3] and to maintain small positional and rotational deviation throughout the drilling process, a surgical manipulator was developed. RESULTS Imaging: Fig. 1 Comparison of image quality Cochlea in CT (left) and DVT (right). Bar 3mm. We have identified a Newtom 3G DVT scanner as a sui- table imaging means. It allows for scanning of patients with a resolution of (0.15 mm) 3 and an effective dose of The Hamlyn Symposium on Medical Robotics (2010) 11