TB-CARE: A Novel Convolutional Autoencoder-based Tuberculosis Classification System with Enhanced EfficientNet

The novel framework for TB classification using Convolutional AutoencodeR with EfficientNet (TB-CARE) involves the utilization of a convolutional autoencoder for feature extraction, an affinity propagation clustering method for selecting templates, and an enhanced EfficientNet (EEffNet) for classification. Extensive tests are performed on datasets that are freely accessible. The results of our methodology surpassed those of previous approaches, demonstrating its practicality for real-world applications. By leveraging deep learning models within the ensemble method, TB classification achieves a notable area under the receiver operating characteristic of up to 0.99, outperforming other tested classifiers and setting a new benchmark. EEffNet exhibits outstanding performance with an accuracy of 99.8%, sensitivity of 99.8%, and specificity of 99.7% on the NIH chest X-ray dataset and accuracy of 99.6%, sensitivity of 99.9%, and specificity of 99.4% on TBX11 k Dataset. These results indicate that employing features extracted from various image sources can significantly enhance the detection rate.