Lung Cancer knobs Classification Model Using Fusion Approach

Abstract

Introduction: Cancer disorders are caused when certain body cells enlarge uncontrollably and spread throughout the body. The second most prevalent malignancy that causes cancer-related fatalities is lung cancer. Early detection makes it simpler to lower the mortality rate from lung malignant lesions. Radiologists can identify lung malignant lesion tumours utilising medical imaging methods such as chest X-rays, magnetic resonance imaging (MRI), and computed tomography (CT) scan images. In this article, we suggest a fusion technique based on a 3D Convolutional Neural Network (CNN) that can increase the precision in identifying lung cancer knobs. Based on the 1018 patients' CT scans from LIDC, a dataset comprising 2000 CT image nodule samples was created to train and assess the CNN model. To evaluate the output Accuracy, sensitivity, and specificity these parameters are used. A 95.8% accuracy, 95.3% sensitivity and 96.4% specificity for testing of nodule classification is attained after applying 3D CNN fusion. The benefit of knowledge obtained from the classifiers is that it helps to increase accuracy while also decreasing the rate of false positives