The Robust Computer Aided Diagnostic System for Lung Nodule Diagnosis

Abstract

Lung cancer has been the numerous dangerous among all other variations of cancer. The fast detection of cancer is conjectured to improve the persistence rate of people living with cancer. Our objective is to present an adequate Computer-Aided Diagnosis (CAD) for the identification of lung nodules from the parenchyma area of the lung and yield the nodule into except cancerous or non-cancerous. In this suggestion, A new Hybrid Classifier method has been Prescribed to detect lung nodules based on numerous image processing and machine learning approaches. The construction of this hybrid system is the combination of unsupervised Enhanced Fuzzy C-Mean (EFCM) clustering and Weighted supervised support vector machine (WSVM). The suggested process includes the subsequent operations: i) the image used is magnified originally. Then the area of concern is cropped, where the user can choose the area to be cropped. ii) The morphological process is implemented to overcome the blood vessels and magnify the nodules. iii)Nodules are distinguished by labeling.iv)Those classified nodule's characteristics are obtained.v)Neural networks are performed as the classifiers that work primarily based on the features chosen. And also, this proposed flexible computing system was associated with the various well-known learning Techniques. The WSVM for analysis is exhibited in this paper, where the execution of misclassification for each practice sample is unusual. The proposed work was capable of detecting the lung nodule that appears near the lung wall. The Provisional results intimate that the recommended method defeats the impact of outliers and yields higher classification speed than previous algorithms.