Key Technologies Of Binocular Camera For Intervention Surgery Navigation

Intervention Surgery Navigation uses medical images helping doctors make surgical instrument into nidus accurately.There are many techniques included such as surgical instrument calibration and tracking,processing of medical images,displaying surgical instrument and medical images together in the same coordinate.This paper focus on image processing efficiency,surgical instrument and image segmentation calibration.Main job are as follows:Firstly,multithread technique is proposed to process images of surgical instrument from binocular camera.Surgical instrument needs to be tracked instantaneously in surgery navigation.Computer must process surgical instrument images taken from binocular camera continually to get 2D coordinates of surgical instrument to restrict 3D surgical instrument.Considering that processing images of surgical instrument at the same time may cause the loss of instantaneity and thus doesn't meet the demand of instantaneous system,this paper use multithread techniques to make full use of CPU resource to promote the efficiency of image processing.Experiments show that multithread techniques can promote the time efficiency of image processing.Multithread technique is faster and efficient when compared with non-multithread.Secondly,a new method of surgical instrument is proposed.During instrument calibration,impulse noise is introduced because of trembling of instrument tip,LMS(Least Mean Square)algorithm losses calibration accuracy under impulse noise.Considering this problem,this paper proposes a new algorithm that combines CG(Conjugate Gradient)algorithm and MCC(Maximum Correntropy Criterion)algorithm.CG converges fast and MCC is very robust under impulse noise.Experiment results shows that compared with MCC,the proposed algorithm converges faster.The proposed algorithm is also more accurate than LMS algorithm and CG algorithm.Then,a new method of image segmentation is proposed.Region-scalable fitting based on local entropy is robust to noise and initialization of contour,however,calculation of local entropy increases the computational cost.To solve this problem,this paper proposed a new method merging CV model and RSF based on local entropy,defining an energy functional with local intensity fitting and global intensity fitting to promote the segmentation efficiency.Experiments results show that the proposed algorithm is robust and can promote the segmentation efficiency.Lastly,we talk about surgical navigation demo system.Hardware system and software system of surgical navigation system is introduced;Also,we take some experiments like camera calibration,instrument calibration and 3D coordinate display.