Tuberculosis Classification Using EfficientNet B3 Deep Learning Architecture

Tuberculosis is a communicable illness that is transmitted through the inhalation of airborne pathogens between individuals. The main purpose of this study is to categorize Tuberculosis disease using X-ray images. Utilizing the EfficientNetB3 transfer learning model enables the attainment of this objective. The potential of EfficientNetB3 in automatically diagnosing tuberculosis (TB) inside medical images has been highlighted by a recent study. This study introduces a novel methodology that employs EfficientNetB3 to classify photos of tuberculosis (TB) disease into two distinct categories. The methods adopted in this study utilize a dataset including 4,200 photographs. The input photos are used for training the model throughout 40 epochs, utilizing a batch size of 30 and a learning rate of 0.001. The findings derived from our study reveal substantial findings, wherein the proposed EfficientNetB3 model showcases an impressive accuracy rate of 99%. The findings presented in this study show the potential efficacy of EfficientNetB3 as a reliable instrument for the automatic categorization of tuberculosis in X-ray pictures. The implications of the research findings are significant, as achieving high levels of accuracy can significantly enhance the rapid and accurate identification of tuberculosis (TB), leading to enhanced patient outcomes and more successful strategies for disease management. Moreover, the successful implementation of this model underscores the importance of employing advanced deep learning techniques in examining medical images, hence opening up new possibilities for developing intelligent diagnostic tools in the healthcare industry.