High Precision Cranio-Maxillofacial Navigation System

With the fast development of science and medical technologies, mini-mally invasive and computer-aided surgeries become the paramount trend inmedicine services. Surgical navigation system (SNS) is an application ofcomputer assisted surgeries and has attracted increasing attentions. SNS sup-plies real-time display of the relative location and orientation between surgi-cal instruments and the anatomical structures of the lesion site for doctors. Itmakes surgeries with higher operation accuracy, smaller operation traumaand less surgical errors. According to the needs of cranio-maxillofacial sur-gery, a new generation of high-precision cranio-maxillofacial navigationclinical prototype system is developed in this Ph.D. research.This paper studies key technologies of cranio-maxillofacial navigation,including stereo reconstruction, surgical instruments positioning, registrationof cranio-maxillofacial surgery navigation system and so on. The main inno-vative results of this study are summarized as follows:1. Study of calibration, recognition and reconstruction algorithms ofsurgical instruments. In this paper, a high precision reconstruction algorithmof surgery instruments is proposed. Usually, the problem of surgical instru-ments positioning is considered as the basic task in computer vision to esti-mate the3-D location of an object from three image points. However, the er- rors from camera calibration, imaging quantization and light variation stillaffect the positioning accuracy. This paper proposes a method which takes themechanical structure of surgical instruments as constraints, and reconstructsthe whole marks on one surgical instrument together. This method can notonly improve accuracy effectively but also reduce the jitter of surgical in-strument greatly.2. Design of new navigation prototype for auto-registration. We de-signed auto-registration and tracking device according to craniofacial ana-tomical characteristics. We also developed calibration, registration and track-ing software for the device applications. The skull specimen experimentsverified the accuracy of the device. The device will avoid causing additionaltrauma, and can be used repeatedly.3. Analysis of registration error of cranio-maxillofacial navigation sys-tem. After study the errors of fiducial localization, fiducial registration, targetregistration, we find the configuration of fiducials is an important factor inminimizing target registration error and its prediction is a good guide for fi-ducial marker placement.4. Study3D shape reconstruction and tracking of deformable surface inSNS. For the case of the binocular camera, we propose an effective approachfor3D shape reconstruction of deformable surface using linear programming(LP) from an image pair taken with a stereo rig. More generally, we recoverythe deformable surface in monocular camera using LP and solve the LPproblem with an algorithm which converges superlinearly. It execute fasterthan the state-of-the-art method.Based on the key technologies above, this paper develops the clinicalprototype system of cranio-maxillofacial surgery navigation. The hardwareconsists of binocular camera, surgical instruments, focus objects and comput- er. The software includes digital image processing, surgery instruments re-construction, medical image reconstruction, intraoperative navigation, and soon. A large number of experimental results on the skull model and skullspecimens show that the clinical prototype system of surgery navigation isvery promising. The system conducted several cases of clinical operation inthe Shanghai Ninth People’s Hospital, which verified that it meet the clinicaloperations requirements.