1. A NEW ALGORITHM FOR THE VISUAL TRACKING OF SURGICAL INSTRUMENT IN ROBOT-ASSISTED LAPAROSCOPIC SURGERY 1 Interdisciplinary Program for Bioengineering, Graduate School, Seoul National University, Seoul, Korea 2 Center for Biomedical Engineering, Asan Medical Center, Seoul, Korea 3 Department of Biomedical Engineering, College of Medicine and Institute of Medical and Biological Engineering, Medical Research Center, Seoul National University, Seoul, Korea Dong Heon Lee 1, Jae Soon Choi 2,*, and Hee Chan Kim 3 Ⅰ. INTRODUCTION [1] J. Ryu, J. Choi, and H. C. Kim, "Endoscopic Vision-Based Tracking of Multiple Surgical Instruments During Robot-Assisted Surgery," Artificial Organs, vol. 37, pp. 107-112, 2013. [2] J. Zhou and S. Payandeh, "Visual Tracking of Laparoscopic Instruments," Jounal of Automation and Control Engineering, vol. 2, pp. 234-241, 2014. [3] T. Xiaoyang and B. Triggs, "Enhanced Local Texture Feature Sets for Face Recognition Under Difficult Lighting Conditions," Image Processing, IEEE Transactions on, vol. 19, pp. 1635-1650, 2010. Ⅴ. REFERENCES ACKNOWLEDGEMENTS Research was supported by the Industrial Strategic Technology Development Program, (10035145, Development of Multi-arm Surgical Robot for Minimally Invasive Laparoscopic Surgery) funded by the Ministry of Knowledge Economy, Korea. Ⅱ. MATERIALS AND METHODS Ⅲ. RESULTS Fig1. Image processing procedures in the proposed algorithm Ⅳ. DISCUSSION & CONCLUSION Fig2. (a) Instrument segmentation without background removal step (b) Instrument segmentation with background removal step Fig3. (a) Tool tip position error of the proposed algorithm in comparison to errors from other two algorithms in one sample image. (b) Mean RMSE (SD) from 7 samples  Mean results from 7 samples The performance of each method was compared by calculating the RMSE between the estimated tool tip position and the manually measured true position. B. Proposal A. Background  To automatically prevent this type of unwanted situation, previous approaches based on color characteristics, geometric information and/or tool shape classifiers achieved partial success in automatic tool detection, but all those algorithms exposed limitation such as intermittent loss of tool tip information under typical time-varying illumination conditions.  In this study, we proposed an improved automatic tool detection algorithm, which normalize illumination and remove background which interferes with tool segmentation. These steps are the most efficient way to remove noise from the images under difficult lighting conditions [3], we verified its applicability in test images.  The use of robot-assisted laparoscopic surgery(RALS) has various advantages such as smaller incisions and thereby shortened recovery time. Nevertheless, there exists some chance of exerting unintended excessive force on fragile tissue by the instruments [1].  Vision-based surgical instruments’ position detection is expected to play an important role in the realization of automatic prevention of possibly hazardous conditions such as tool collision and unintended tissue damage.  As compared to the noise removed image without background removal step Fig. 2(a), the one followed by background removal Fig. 2(b) clearly has further reduced noise.  As shown in Fig. 3, the RMSE values of the tip position are 1.07±0.83mm, 1.34±0.83mm, 0.08±0.63mm for the tool #1, #2, and #3, respectively, which represents significantly smaller average as well as SD of error compared to the other two previously published methods [1][3].  Stomach laparoscopic surgery image from a pig model, The proposed algorithm consists of several steps of image processing as shown in fig. 1, among which so-called illumination normalization, and background removal step. The last step shows the results of tool segmentation. The red dots represent the true position measured manually and the blue dots represent the calculated tool tip position.  In conclusion, the proposed instrument segmentation algorithm is a good candidate for the automatic tool tracking which can be usefully applied to the detection of emergency situations during robot-assisted laparoscopic surgery.