Abstract

Tuberculosis is one of the leading causes of mortality worldwide, but its impact can be reduced through timely diagnosis and treatment. The Ziehl–Neelsen staining method is commonly used for tuberculosis detection, where specialists manually examine microscopic images to identify bacilli. However, this process is time-consuming and requires expert knowledge. This work proposes an automatic tuberculosis bacilli segmentation system to accelerate diagnosis. The input images are first preprocessed using an adaptive mean filter to remove impulse noise, followed by power-law transformation to enhance image quality. The color space is then converted from RGB to HSV, which better isolates image components for processing. A multi-level thresholding algorithm is applied to accurately segment bacilli from microscopic samples. Experimental results demonstrate an improvement of 2.13% in segmentation accuracy compared to state-of-the-art techniques, highlighting the effectiveness of the proposed approach.

Additional Comments