Design and evaluation of a needle tip measurement system based on binocular vision

Abstract

The precise location of the needle tip is critical in robot-assisted needle-based percutaneous interventions. An automatic needle tip measuring system based on binocular vision technology with the advantages of non-contact, excellent accuracy and high stability is designed and evaluated. First the measurement requirements of the prostate intervention robot are introduced. A laser interferometer is used as the reference for measuring the position of the needle tip whose relative position variation is described as the needle tip distance in the time domain. The parameters of the binocular cameras are obtained by Zhang’s calibration method. Then a robust needle tip extraction algorithm is specially designed to detect the pixel coordinates of the needle tip without installing the marked points. Once the binocular cameras have completed the stereo matching, the 3D coordinates of the needle tip are estimated. The measurement capability analysis (MCA) is used to evaluate the performance of the proposed system. The accuracy of the system can be controlled within 0.3621 mm. The agreement analysis is conducted by the Bland–Altman analysis, and the Pearson correlation coefficient is 0.999847. The P/T ratio value is 16.42% in the repeatability analysis. The results indicate that the accuracy and stability of the binocular vision needle tip measuring system are adequate to meet the requirement for the needle tip measurement in percutaneous interventions.