A Low-cost Optical Tracking System for Arthroscopic Surgical Training

Abstract

The minimally invasive surgical procedure is more technically demanding than normal open-joint surgeries because of the limited vision. Thus, preoperative training for surgeons is essential. Current training for arthroscopy uses a fluoroscopy system, but that is costly, and the trainees will be at high risk under X-ray radiation exposure. The purpose of the overall project is to design an affordable arthroscopic surgical training station and no special safety procedures for trainees. Our system combines a virtual imaging system (to replace fluoroscopy) with a physical synthetic model of a hip joint. The purpose of the current project is to develop a 3D visual tracking system using low-cost Raspberry Pi cameras and to test the resolution and accuracy. Two Pi cameras were used to track markers on a surgical tool. The tracked data are intended to be used with a synthetic hip and superimposed on a CT dataset of the hip that can mimic surgery with real-time fluoroscopy. The reconstructed surgical tool can be overlaid on the virtual fluoroscopy to mimic the display in the real arthroscopic surgery. Pi cameras tracked passive coloured markers on a tool from different angles. The markers were tracked independently by colour segmentation, and positions were sent to a central computer simultaneously for 3D reconstruction. The optical tracking system supports 25fps, 1080p live video streaming. The largest errors in the X, Y and Z-axis are 12.46±0.14, 8.55±0.3, 10.09±0.42 mm respectively while the repeatability is in a range from 0.61 to 5.17 mm. These results demonstrate the possibility of using Raspberry Pi camera modules in a low-cost optical tracking system for surgical training purposes. Currently, the frame rate is low (25fps) and the error is still too large (up to 12.46mm) for use in surgical tracking. The resolution of the camera could improve when a better camera module is available.