Tuberculosis (TB) is a disease caused by the bacterium Mycobacterium tuberculosis, which was discovered by Robert Koch in 1882. The bacterium is rod-shaped, with a width of 0.3–0.6 μm and a length of 1–4 μm. It is transmitted through the air, for example, when an infected person coughs or sneezes. TB diagnosis is typically performed through microscopic analysis of sputum samples. TB is a serious infectious disease and remains a global health concern. Rapid and accurate diagnosis is crucial for effective treatment, yet conventional methods are often time-consuming and less precise. This study developed a TB bacterial image classification system for sputum samples using a Backpropagation Neural Network (BPNN). The system differentiates between single and clustered bacteria using length, endpoints, and branching features. The dataset consisted of 120 images, divided into 60 training and 60 testing samples. All images were processed using preprocessing techniques to enhance image quality. The length, endpoints, and branching features were extracted from the images and used as input to the BPNN. The results showed that the BPNN method could classify TB bacterial images with an accuracy of 86%. The system was also able to distinguish single and clustered bacteria more accurately, potentially contributing to improved TB diagnosis.