Rapid, sensitive and accurate tests are critical for contemporary health care. Tuberculosis has been with us since time immemorial, and although being treatable and curable, it remains one of the world’s most infectious malady, killing millions each year. Significant advances have been made in tuberculosis diagnostic technologies however, they are not adequately portable, rendering them inapplicable in remote, rural areas where they are most required. They are costly, lack the essential sensitivity or precision and are frequently susceptible to error. In this regard, an automated system that can analyse and diagnose pulmonary TB by using Chest X-Rays (CXR) can prove to be a cost effective, dependable and quick alternative to manual diagnostic methods. In our work, we propose a customized CNN architecture having three convolution layers and three max pooling layers for accurately classifying the CXRs into TB-infected and normal classes. To deal with the issue of unbalanced classes in the TB CXR dataset, we use different oversampling techniques such as weighted averaging, SMOTE, ADASYN and Borderline SMOTE. To ensure a comprehensive assessment, we extended our research to further employ two widely recognized pre-trained CNNs; DenseNet169 and ResNet50 to gauge their performance on the same task. Grad-CAM has been applied to interpret the decision-making process of the models and correctly identify the regions of interest. Various evaluation metrics employed show very encouraging results. The proposed customized CNN model along with ADASYN outperforms DenseNet and ResNet models in terms of computation time as well.

Pulmonary disorders, including conditions such as Pneumonia, Tuberculosis, and COVID-19, affect millions worldwide, presenting a major global health challenge. Accurate and timely detection of these diseases is crucial for effective treatment and patient care. This research paper introduces a novel CNN-LSTM model for effectively differentiating between and categorizing multiple lung disorders. Leveraging a dataset comprising 4529 chest X-rays (CXRs) collected from Shenzhen, Montgomery, Belarus, and COVID-19 Radiography datasets, the study aims to enhance disease diagnosis accuracy. Contrast-limited Adaptive Histogram Equalization and median filtering techniques have been applied to enhance image quality and reduce noise. Although LSTM networks are typically used for sequential data, this study employed them to capture spatial dependencies within 2D medical images. This approach allowed the model to learn complex representations and retain context from different image regions, enhancing feature extraction and overall performance. LSTM worked by exploiting the neighborhood relationships between pixels in CXRs by treating image patches as sequences. This approach enabled the model to better understand and utilize spatial information, thereby improving accuracy. Furthermore, baseline CNN and CNN-Bi LSTM models were also executed for a comparative study. Among the implemented models, CNN-LSTM emerged as the most effective, achieving an accuracy of 96.26% and precision, recall, F1 score of 96.44%, 96.62%, and 96.49%, respectively. It was followed by CNN-BiLSTM and CNN models with accuracy rates of 95.58% and 94.24%.