Robust Cough Analysis System for Diagnosis of Tuberculosis Using Artificial Neural Network

This research proposed a robust and easily applied method for tuberculosis (TB) screening system based on the analysis of patients' cough sounds. There are various existing diagnostic tests for TB, but they are expensive and require highly skilled physicians and laboratory facilities. Therefore, there is a need for a low-cost, quick-to-diagnose, and easily accessible solution for diagnosing TB in developing countries using a patient's cough sound. The coughing sound of patients with TB have distinct mathematical features or information that can indicate a disease. The use of patients' cough sounds to diagnose pulmonary diseases is an active research field with promising results; however, a robust system for diagnosing tuberculosis using cough sounds is currently unavailable commercially.
For this research, a dataset of 6476 cough and non-cough sound events was collected from patients with various respiratory diseases from Bahir Dar Felege Hiwot compressive specialized hospital using three different recorders. An automatic cough detection and classification system were implemented using an artificial neural network (ANN) and a support vector machine. The