A novel model of feature extraction for lung cysts detection in CT image using Minutiae based Mumford and Shah functional model

摘要

Lung cancer is one of the commonly occurring and most hazardous diseases to cure which increases the death rate day by day. In order to reduce the death rate, it is necessary to detect the lung cancer in its initial stages and thereby to assist the surgeons to clear away the portion of lung for the treatment of lung cancer, and tumours. This paper concentrates at developing a Computer-Aided Diagnosis (CAD) system for detecting lung cancer by analysing the Computed Tomography (CT) images of lungs. And it is carried out with filtering, binariza-tion, image segmentation based on the Mumford and Shah model, Image enhancement including binarization and thinning, Minutiae Extraction using Termination and Bifurcation, Removal of False Minutiae Points and finally feature Extraction using grey level co-occurrence matrix (GLCM). Also, the system removes 98% of false minutiae and thus, is more efficient than other algorithms which achieve 80% to 90%. In addition, the image quality is analysed using various assessment Metrics. Finally, a comparative and robustness analyses are carried out with the existing self-learning approach in terms extracted values. The results of the proposed system show that there is significant improvement in PSNR value compared to the existing method.