Deep pre-trained networks as a feature extractor with XGBoost to detect tuberculosis from chest X-ray. (July 2021)

• Record Type: Journal Article
• Authors: Rahman, Muhammad; Cao, Yongzhong; Sun, Xiaobing; Li, Bin; Hao, Yameng
• Is Part Of: Computers & electrical engineering. Volume 93(2021)
• Publication Date: 2021-07
• Subjects: Tuberculosis diagnosis -- Features extraction -- Deep learning -- Medical image processing -- X-ray screening; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2021.107252
• Languages: English
• ISSNs: 0045-7906

Abstract

Abstract: Pulmonary Tuberculosis is a plague caused by Mycobacterium tuberculosis or Tubercle bacillus, which kills 1.8 million people worldwide. Tuberculosis is among the top 10 deadly diseases. It can be life-threatening if it does not diagnose at the initial stage. This study identifies Tuberculosis from chest X-ray images, utilizing image preprocessing techniques, deep learning methods and a publicly accessible dataset of 7000 (3500 normal and 3500 tuberculosis infected) chest X-ray images. We used three pre-trained networks (ResNet101, VGG19, and DenseNet201) to extract features from chest X-ray images. The eXtreme Gradient Boosting (XGBoost) model is used to classify tuberculosis and normal cases. The highest results (Area Under Curve of 99.93 ± 0.13%, accuracy 99.92 ± 0.14%, precision 99.85 ± 0.20%, sensitivity 100 ± 0.1%, F1-score 99.92 ± 0.14% and specificity 99.85 ± 0.20%) achieved with DenseNet201-XGBoost architecture, for tuberculosis chest X-ray images classification, as compared to the ResNet101-XGBoost and VGG19-XGBoost architectures. The proposed method bestows hope to radiologists and medical facilities in developing countries to tackle Tuberculosis's early diagnosis problem.