Image breaking method for lung isolation from chest x-rays

Abstract

Most of the techniques that are available for chest x ray image segmentation are not purely based on image processing. Majority of the existing work involves a limited preprocessing part, and the rest is conducted with the leverage of Machine learning and Convolutional neural networks. This study strives to fill this void by introducing a fully image processing-based lung segmentation method. This paper discusses a novel method introduced to segment lungs from chest x ray images purely based on image processing. The method introduced here is named as “Image Breaking method” due to its unique wary of breaking the CXR in to segments to maximize the data extraction. Reason for selecting Xray images is X Ray images are common and reachable in almost all the general hospitals and for most of the lung related diseases CXRs are the most common primary medical imaging technique that is used. Because of its complexity and lesser quality of the image, chest x ray images are hard to segmentate only based on image processing. Here we have taken an attempt to provide guidance for consumers the pathways that can be taken to achieve this task only using image processing. It needs lesser resources; lesser line of code and all the steps are based on experience we gained through experiments. Final evaluation showed that this method provides a fairly good output when it compared with structural similarity to images that are segmented by specialists.