A Deep Learning Based Approach towards the Automatic Diagnosis of Pneumonia from Chest Radio-Graphs
Anuja Kumar Acharya*and Rajalakshmi SatapathySchool of Computer Engineering, KIIT University, India
Corresponding Author E-mail : anujafcs@kiit.ac.in
Abstract: In the research work, we propose an automatic detection of pneumonia from chest radiography image using the deep Siamese based neural network. Although in the recent past many method were devoted but these methods are either solely depends on the transfer learning approach or the traditional handcrafted techniques towards classifying the pneumonia disease. Viral and bacterial pneumonia infections are distinguished by analyzing the amount of white substance that is spread across the two segment of the chest X ray image. Deep Siamese network use the symmetric structure of the two input image to compute or classify the problem. Each of the chest X-ray image is divided into two segment and then feed it to the network to compare the symmetric structure along with the amount of the infection that is spread across these two region. We use the Kaggle dataset, to train and validate our model towards the automatic detection of the different kind’s pneumonia disease. This proposed model could help the medical practioner towards easily identifying the pneumonia problem from the X-ray imagery. Simulation results The model outperforms the state-of-the-art in all performance metrics and demonstrates reduced bias and improved generalization.
Keywords: Clinical Decision; Chest X-rays(CXR); Computer-Aided Diagnosis; Convolution Neural Networks(CNN); Deep Siamese Network(DSN); Pneumonia; Segmentation Back to TOC