The Model Development for Early Lung Cancer Analysis by Using Image Processing and Neural Network

Abstract

Lung cancer is the leading mortality disease for men and women compared to other organ cancers. The leading cause is smoking until the patient begins to show symptoms. Therefore, they see a doctor determine that the disease has spread in the last stage, causing cumbersome and complex treatment. Therefore, early screening of patients is essential to allow patients to receive treatment promptly and have a chance to recover from the disease. This research has developed a model for early lung cancer analysis by using a CXR image that can screen many patients when they are asymptomatic. Improve the image enhancement to reduce the noise with median filter and then go into image processing by image segmentation with Active contour algorithm, image edge detection with Laplacian of Gaussian (LoG) algorithm, and image extraction with Shape and GLCM in combine with data classification with a neural network using MLP compared against SVM classifiers. Training and testing the model's performance by the result of MLP provides a better time and up to 99% accuracy.