Optimum thresholding for nodule segmentation of lung CT images

Abstract

Lung cancer is killing more people throughout the world. This encourages early cancer diagnosis. The first and most important step in detecting cancer using Computer Vision (CV) and Machine Learning (ML) techniques is segmentation of the lung region and, from there, nodules. This research adds to the system of cancer diagnosis based on a CV. This work suggests the use of the optimum thresholding method to improve the initial lung segmentation, and the nodules were then segmented from the segmented lung image using the Active Contour (AC) approach. The Markov Random Field (MRF) approach is used to fine-tune the post-processing following the nodule segmentation procedure. The findings of the experimenting of the recommended lung nodule segmentation technique are shown in the results section.