Tuberculosis (TB) is one of the leading causes of death worldwide and human immunodeficiency virus (HIV) co-infection poses a great challenge. The control and it is estimated in 2015, approximately 1.8 million people infected by tuberculosis and died, even in the developing countries. Mostly the death of the people is not prevented in early stages, though the interpretation methods are more cost for mass adoption. The very most popular interpretation methods are scrutiny of frontal thoracic radiographs, but this method is abated by the need of properly trained radiologists. Researches on automatic interpretation by computational techniques of medical images eliminates the individual analysis and reduces the overall costs. Recently deep learning polished an excellent result in classifying the images on distinct domains but its utilization on tuberculosis is remains limited. Thus, the focus on this work is to overcome the limitation by advance research of the application with convolutional neural network as a feature extractor to detect the disease, where other techniques would have a limitation in medical image processing. The proposed model is check with convolutional neural network without pretrained neural networks. From the experimental evaluation, it is found that the proposed model with Convolution Neural Network as feature extractors works better than the previous works in the diverse research areas.
Volume 11 | 07-Special Issue
Pages: 611-617