Using Convolutional Neural Network for Chest X-ray Image classification
Matija Soric, Danijela Pongrac, Iñaki Inza
Abstract
Chest X-ray is an imaging technique that plays an important role in pneumonia diagnosis. Owing to the high availability of medically-oriented image datasets, great success can be achieved using convolutional neural networks (CNNs) in the recognition and classification of these images. Since previous research has shown CNNs to perform as well as the best clinicians in diagnostic tasks, they caused great excitement among researchers. In this paper, convolutional neural network (CNN) machine learning (ML) model was built using a supervised dataset. The dataset used contained both pneumonia and non-pneumonia images, which the model had to classify correctly. In the end, the model is demonstrated to have achieved satisfactory results, with the high accuracy of 90.38%, 98.21% recall and 87.84% precision.