Applied Artificial Intelligence and Prospect of Internet of Everything (IoE) for the Detection of Tuberculosis

Abstract

ABSTRACT Mycobacterium tuberculosis is a bacterium that causes disease known as Tuberculosis. Tuberculosis is highly contagious and can result in high mortality rate if left untreated. In order to screen individuals suspected of the disease, medical expert relies on several conventional approaches which are hindered by several limitations which include time consuming, high workload, false positive results, etc. This calls for the need to develop smart and automatic approaches that can address these challenges. Majority of existing studies reported the use of 1 or 2 pretrained models and the use of SoftMax as the classifier. Moreover, majority of the studies trained models using a single type of dataset which are mostly curated from public accessible domains. Thus, this study addressed these challenges by: (1) The use of several pretrained models (2) The use of 2 classifiers which include SVM and KNN and (3) Training and validating pretrained models fused with classifiers on microscopic slide and chest X-ray images. The result achieved in this study highlights the prospect of computer-assisted techniques in triaging and screening of TB. The integration of Internet of Everything (IoE) in medical diagnosis has the potential to increase healthcare outcome, boost productivity, and reduce workload.