Lung tumor detection and diagnosis in CT scan images

In recent years, the image processing mechanisms are widely used in medical image diagnosis, especially in detection of various tumors. In this paper, we propose a level set-active contour model with minimizer function for lung tumor diagnosis and segmentation. Kernel based non-local neighborhood denoising function is used to get noise free image. Second order histogram based feature extraction is accomplished for classifying the images under normal and abnormal classes. Following tumor detection, exact segmentation of the tumor is effected by level set-active contour modeling with minimized gradient. Experiments demonstrated that our methodology could segment the lung field with pathology of variant forms more precisely.