Automation in Sputum Microscopy

Abstract

This chapter describes an automatic intelligent diagnostic system for Tuberculosis. Sputum microscopy is the most common diagnostic technique to diagnose Tuberculosis. In Sputum microscopy, Sputum are examined using a microscope for Mycobacterium tuberculosis. This manual process is being automated by image processing, where classification is performed by using a hybrid approach (color based and shape based). This hybrid approach reduces the false positive and false negative rate. Final classification decision is taken by a fuzzy system. Image processing, soft-computing, mechanics, and control system plays a significant role in this system. Slides are given as input to the system. System finds for Mycobacterium tuberculosis bacteria and generates reports. From designing point of view ARM11 based, 32 bit RISC processor is used to control the mechanical units. The main mathematical calculation (including image processing and soft computing) is distributed between ARM11 based group and Personal Computer (Intel i3). This system has better sensitivity than manual sputum microscopy.