Abstract
AbstractDetection and classification of lung disease detection using recent methodologies have become an important research problem for smart computer-aided diagnosis (CAD) tools. The emergence of deep learning brings automation across the different domains to address the concerns related to manual techniques. The chest X-ray image remains one of the effective tools for lung disease detection such as pneumonia. This paper presents a framework for pneumonia disease detection and classification from the raw X-ray images. The proposed framework consists of image preprocessing, adaptive segmentation, features extraction, and automatic disease detection. Raw X-ray images are preprocessed by applying the lightweight and effective filtering algorithm. The region of interest from the preprocessed image has been located by using the adaptive segmentation algorithm. We propose a dynamic threshold mechanism followed by morphological operations for adaptive segmentation. The hybrid feature vector has been implemented using visual, texture, shape, and intensity features. For disease detection and classification, the hybrid features are normalized using robust normalization and then automatic deep learning classifier model recurrent neural network (RNN) with long short-term memory (LSTM) designed. The simulation results show that the proposed model outperformed state-of-the-art similar methods.KeywordsChest X-ray imageLung diseasePneumoniaComputer visionDeep learningSegmentation
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have