Segmentation Effect on Lungs X-Ray Image Classification Using Convolution Neural Network

Abstract The effect of segmentation on lung X-ray image classification has been analyzed in this study. 150 x-ray images study were separated into 78 as training data, 30 validation and 42 testing three categories: normal lungs, effusion cancer lungs. In pre-processing, the modified by adaptive histogram equalization to improve quality increase contrast. aims mark contouring area obtained from thresholding some morphological manipulation processes such filling holes, openings, labelling. Image uses Convolutional Neural Network (CNN) with five convolution layers, an Adam optimizer, epochs. is comparing performance segmented unsegmented images. study, dataset reached overall accuracy 59.52% network process. greater accuracy, 73.81%. It indicated that process could because input pattern easier classify. Furthermore, technique can be one alternatives developing technologies, especially for medical diagnosis. Segmentation Effect Lungs X-Ray Classification Using Convolution Network.