A Vision-based Tracking Method of Flexible Surgical Instrument

Abstract

In clinical minimally invasive surgery, flexible surgical instruments are widely used for intervention treatment. For a better clinical characteristics, the requirements of tacking accuracy, visual image recognition rate and less lag of real-time tracking are increasing. However, in practical conditions, the environment is under uneven illumination and produce uneven image quality. The recognition rate of flexible devices is low and the tracking is inaccuracy. Therefore, this paper presents a vision-based tracking method of flexible surgical instrument in order to promote the situations above. First, the entrance region of the flexible surgical instruments in the field will be separated to determine an initial local area. Then, the real-time end position detection of the flexible surgical instrument in this local will be captured, and a new local will be generated once the end of the instrument leaves the local area. In every new local area, the process of detecting end position-generating a new local area will be repeated until the operation of the surgical instrument and the tracking task are completed. The experiment results illustrate that the proposed method can effectively solve the problems which include poor image quality, low recognition rate under uneven illumination and inaccuracy tracking. Thus, this tracking method can greatly improve tracking precision of the end of instrument.