Automatic Interpretation of Pneumonia Using Deep Neural Networks

V. Geethanjali, V. Yeshwanth and N.L. Anbarivan

An automatic detection of histopathological images has become the most challenging task in the fields of computer vision. Earlier methods used high pass filters that are sensitive to image noise which affected the accuracy and performance. Hence, we are in need of an edge detection algorithm that copes up with image noise as well as combines morphology to enhance the results. This paper describes a deep learning architecture approach based on training convolutional neural networks (CNN) for detecting the presence of pneumonia clouds in chest X-rays (CXR). In addition, it presents the preliminary classification task of applying CNN to learn features and classify images task. The results showed a better training rate and 93 % accuracy.

Volume 11 | 03-Special Issue

Pages: 1977-1983