Performance Comparison of Deep Learning Methods for Lung Nodule Benign and Malignancy Classification

Today, lung tumors is the major cause of malignancy death. Doctors use the CT scan, x-ray, and manual method for lung nodule detection and diagnosis. However, such methods fail to correctly classify the malignancy of nodules. For early, automatic, and accurate detection of lung nodules, good lung nodule classification methods are required for benign and malignancy classification. To improve accuracy and lung nodule classification performance, many scholars have proposed some deep learning algorithms to solve this problem. In this paper, various deep-learning methods are compared while classifying lung nodule malignancy. Researchers have used deep learning algorithms such as CNN, 2DCNN, 3DCNN, VGG16, ResNet50, and RNN algorithms for lung nodule malignancy classification. We compared all methods performance and accuracy using a confusion matrix. After comparing experimental results, we found that the Deep Residual Network (ResNet50) gives a good result with 98.91% accuracy.