A new method of automatic recognition for tuberculosis disease diagnosis using support vector machines.

Abstract

Tuberculosis is an infectious disease caused by a bacillus called Mycobacterium tuberculosis. It can lead to death in untreated and inappropriately treated patients particularly in countries with low income. Therefore, early diagnosis of the disease not only increases treatment success, but also reduces death rates. Today, due to high classification and diagnosis rates, specialist systems have become an important tool in diagnosis of the disease. In this study, support vector machines (SVM), which is a machine learning technique was used for preliminary diagnosis of tuberculosis disease for the first time. A recognition system that was developed with the properties included in patient reports obtained from a local hospital was tested for its performance. The results indicated performance of the designed system was quite successful and that it could be used in diagnosis of the disease. Obtained diagnostic results were compared with similar studies using different specialist systems on this disease, and it was observed that our results were better.