Lung Cancer Detection Using Modified Fuzzy C-Means Clustering and Adaptive Neuro-Fuzzy Network

Abstract

Air Pollution is responsible for many diseases and death happening all around the world. The Particulate Matter (PM) present in the air is responsible for lung diseases. The PM size is tiny, and when the human inhales air, it directly reaches the lungs and produces many diseases. PM creates many diseases and Lung Cancer is one among them, and it is the third most deadly disease states the American Cancer Society in its 2019 summit report. Early prediction of lung cancer can be treated quickly, and there is an increase in the patient’s lifetime. On implementing CAD systems in hospitals, it assists radiologist in identifying smaller cancer nodules and also it helps the radiologist to validate his observations. The proposed system consists of five steps: They are image acquisition to acquire input images from the user, image pre-processing to enhance the image so that accuracy will be introduced, image segmentation to find out the critical region and eliminating other regions, feature extraction to extract the features of the critical region and finally to classify the nodules as benign and malignant classification is performed. In the proposed system, we have used the wiener filter as a pre-processing technique. To segment, the pre-processed image Modified Fuzzy C-Means Clustering (FCM) Clustering is performed and the region of interest is segmented. Finally, Random Forest Adaptive Neuro-Fuzzy Classifier is applied to classify the nodules. The accuracy, sensitivity and specificity of the proposed system were found to be 98.70%, 95.80% and 90.00% respectively.