Automatic identification of Mycobacterium tuberculosis in sputum smears with Acid-Fast stain

• Main Authors: Cheng-Jia Tsai, 蔡承家
• Other Authors: S. P. Fang
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/43344873730898004977

碩士 === 南台科技大學 === 電子工程系 === 98 === Tuberculosis is the most serious infectious disease in Taiwan and causes more than one thousand deaths every year, half of all infectious death cases. This thesis uses Matlab and Matlab image processing toolbox to develop automatic identification software of Mycobacterium tuberculosis in sputum smears with Acid-Fast Stain from microscope images. The identification algorithm is based on the shape and color of Mycobacterium tuberculosis. The images were taken by a digital microscope camera with 1000 magnification. The candidate features are red features after filtering the origin image, and processed further by parameters according to Mycobacterium tuberculosis shape. The results are sensitivity 96% and specificity 98%. The identification consuming time is around 0.7 second with 5 candidate features and 2.77 seconds with 51 candidate features. We have demonstrated that the software can improve the current identification technology in accuracy and speed.