Search

The recognition of tuberculosis (TB) bacilli base on a color image processing is one of the most challenging tasks in image processing. The problems increase if the captured images are blurred or have low contrast. This paper proposes partial contrast and saturation technique methods to enhance color images of Ziehl-Neelsen stained smear for the detection of TB bacilli. The partial contrast and saturation techniques have been tested using seven different images. It is shown that the TB bacilli are easier to be detected after applying these techniques.

Tuberculosis infection is a serious disease which could be controlled by early diagnosis. A commonly used technique for detecting the TB bacilli is by analyzing sputum smear. Now days, image recognition systems have several applications in enormous fields. This paper uses an artificial neural network to enhance color images of Ziehl-Neelsen stained smear for the purpose of detecting TB bacilli. The first necessary step is the captured images are converted into usable format (RGB values) and pass the RGB values to neural network for training to emulate the contrast enhancement technique. The training is based on back-propagation algorithm. It is found that the proposed neural network approach could emulate contrast enhancement technique quite well.

The Ziehl-Neelsen stain, also known as the acid-fast stain, was first described by two German doctors; Franz Ziehl (1859 to 1926), a bacteriologist and Friedrich Neelsen (1854 to 1894), a pathologist.It is a special bacteriological stain used to identify acid-fast Mycobacteria. The preparation of Ziehl-Neelsen slides require several procedures and the slide should be examined under oil immersion by using the 100X magnification. There are some factors that may degrade the image quality such as under expose, over expose, under stain and thick samples. Thus, the images should be enhanced to improve the image quality in term of contrast and intensity. This paper proposes a contrast enhancement technique based on bright stretching and dark stretching algorithms for color images. Although the adopted image processing technique is quite simple, the results indicate that our method may have some potential to be used for improving the quality of Ziehl-Neelsen slide images.