A Novel Dense-Net Deep Neural Network Method for Classify Tuberculosis Stages Using Chest X-Ray Images with Enhanced Features

Although chest X-ray (CXR) imaging is critical for the detection of potentially lethal disorders, interpretation may arise because to similarities between distinct lung conditions. In response, researchers proposed CBAMWDnet, a new deep learning model for tuberculosis (TB) diagnosis based on computed tomography (CXR) images. To successfully collect geographical and contextual data, the model includes the Convolutional Block Attention Module (CBAM) into its architecture, which is based on the Wide Dense Net (WDnet). Pre-processing using an enhanced Gabor filter improves image quality prior to applying the model. We test the model using three well-known datasets: Shenzhen, Montgomery, and Indiana. The results indicate that the Shenzhen dataset performs the best. CBAMWDnet beats previous techniques by better classifying tuberculosis (TB) into normal, primary, latent, and active stages. This complex classification engine improves both accuracy and dependability.