Optical Marker Recognition and Pose Determination Using Neural Networks : Toward Development of A Dental Surgical Navigation System

IINTRODUCTION A new system of computer-assisted surgery for dental implant has been developed and introduced clinically. This navigation system combines preoperative surgery and intraoperative support offered to minimize potential risk of damage to critical anatomic structures of patients. A robust and high accuracy tracking system is a main parameter in navigation system. A set of infrared (IR) based surgical marker emitters are designed in different patterns for tracking the moving surgical instruments. The benefit of the infrared base is that the objects except the surgical markers in the operation room become invisible. IR surgical marker is introduced and developed following the procedures of image capture and processing. Computer-assisted surgery has been introduced in many clinical aims, for example, in skull-base surgery. In 1990, Werner Krybus et al. employed a hand-guided electromechanical 3D-Coordinate digitizer to locate points of interest within the operative field for skull-based surgery. The coordinates measured this method are correlated with a voxel model of the object gained by a preceding CT examination. The accuracy of this way is less than ± 1 mm and this system has been successfully applied in ear-nose-throat (ENT) operations and neurosurgical procedures. Furthermore, a new technique such a Fuzzy logic has been implemented in a current tracking system for navigation purpose. Fang-Chun Huang et al. develop an active vision based space-positioning robot (AVBSPR) by using a Fuzzy inference engine to construct the dynamic tracking the surgical marker movement. They used the fuzzy logic to track and determine the image capture orientation in realtime. The experimental results showed that the robot can track the surgical markers and the positioning error is around 5 mm within 2000 mm distance operating ranges. Sigeru Omatu et al. proposed a rotation-invariant neural pattern recognition system to apply on a rotated coin recognition problem. They used neural network which consisted of a preprocessing network to detect the edge features of input patterns and a trainable multilayered network to recognize rotated patterns and estimate their rotation angles. The results had been proven well compared with the results based on mental rotation via theory of information types which is useful for elucidation of the human perceptual process. This paper proposed surgical marker pattern recognition and rotation angle determination for surgical navigation application. The objective of this study is to implement a new version of tracking algorithm for marker recognition and orientation by using neural network architecture. A rotation-invariant neural pattern …