A Simple Method to Detection the Lung Cancer Tumor using CT images on Deep Learning

Abstract

This study proposes a method for lung cancer tumor detection by first pre-processing DICOM files, then separating the lung and soft tissue areas of these files via a linear combination, and finally analyzing the CT images more clearly. The image processing method used in this study achieved a sensitivity of 97.96%, precision of 99.23% of, and F1-score of 98.56%. In addition, because of its higher performance in comparison with the reference dataset, it can be concluded that this image processing method has an effect on the learning result of the neural network. © 2021 IEEE.