Segmentation of Lungs in Chest Radiographs using Optimized Clustering Technique

Lung Cancer is considered the most common type of disease, when compared to that of the other types. It is one of the leading types of cancer that causes majority of the patient death worldwide. It is often detected only at the later stage as they are usually diagnosed at their advanced stages. Survival of patients with lung cancer is almost impossible. They often die within one year after the onset of clinical symptoms. Screening and early detection play an important role in saving the life of a patient. Chest radiography and computerized tomography scans are the techniques mostly used to diagnose and detect tumor in lungs. They require less radiation dose, and is available in most of the diagnostic centers. Their cost is also less when compared to the other techniques used for diagnosis. Nodule detection by using conventional radiographs is still not much effective, so there arises a need for alternative image processing techniques to improve the efficiency of detection. Image segmentation is considered as the first step in processing an image. Further analysis of the image would be made more effective if segmentation is efficient. There exits many segmentation algorithms based on clustering and thresholding approaches. In this paper, a bimodal, optimized and modified k-means algorithm is developed to segment the chest images.