Path planning optimization for the TRUS guided prostate biopsy robot
장도영 · Seoul National University Open Repository (Seoul National University) · 2012
Prostate cancer (PCa) is most commonly diagnosed by a freehand, extended sextant biopsy with trans-rectal ultrasound (TRUS) guidance. There has been much research to optimize the outcomes of freehand TRUS biopsy by expanding the number of biopsy. However, it is unknown how accurately the freehand, TRUS-guided biopsy cores are placed within the prostate gland and how biopsy core distribution may affect the PCa detection rate. In this thesis a robotic assisted trans-rectal ultrasound guided prostate biopsy system is developed. The system is based on an ultrasound probe tandem robot which has four degrees of freedom, designed and manufactured in the URobotics Laboratory at Johns Hopkins University. The robot has a remote center of motion(RCM) which is a spatial pivoting point designed to be placed in the sphincter of the rectum. This particular mechanism enables full access of the prostate through the limited entry point of the anal sphincter. The biopsy robot control program consists of 4 steps; i.e. scan, surface extraction (segmentation), path planning, biopsy. First, the prostate is scanned and the volume is reconstructed using the position and orientation information of the collected ultrasound images. Next, the surface of the prostate is extracted and based on this a biopsy path planning optimization is performed. In order to maximize the efficiency, that is to maximize cancer detection rate with a given number of biopsy cores, a probability based significant cancer detection rate model is proposed, which quantifies the volume sampling rate and false-negative detection probability. The significant PCa is assumed as a sphere with size larger than 0.5 cc (D=10mm). Based on this model, a pattern search algorithm is applied to maximize the prostate volume sampling ratio. To determine the accuracy and precision of biopsy cores placement in-vitro, we developed a biopsy simulation system with a gelatin-based pelvic mockup and optical tracking system. We then determined the exact geometric distribution of biopsy cores by five experienced urologists and the TRUS Robot. Finally, the detection rate by freehand versus TRUS Robot-assisted biopsy has been determined using the probability-based model. The results show that the biopsy quality performed by the robot is superior compared to the result of 5 anonymized urologists. First, the mean targeting detection error is which stands for accuracy in the initial biopsy has been improved from 9 mm to 0.96 mm. The accuracy and precision of biopsy which indicates the quality in the case of repeated biopsy shows an improvement of 10.1 mm to 1.7 mm and 23.6 mm to 0.6 mm each respectively. Finally, the significant cancer detection rate of a standard 12 core biopsy performed by urologists was 43.5% where the detection rate of the robotic system was 75%. In conclusion, systematic biopsy with freehand TRUS guidance does not closely follow the sextant biopsy plan and may result in suboptimal sampling and cancer detection. The newly developed TRUS robot biopsy system provides an effective alternate means of accurate and precise sampling, and may significantly enhance detection rate coupled with the probability based optimized core placement plan.