Pixel machine learning with clonal selection algorithm for lung nodules visualization

• Main Authors: Pang, Yuen Yuen, Hang, See Pheng, Yan, Soon Weei
• Format: Article
• Language: English
• Published: FAZ Publishing 2019
• Subjects: QA Mathematics
• Online Access: http://eprints.utm.my/id/eprint/87364/1/HangSeePheng2019_PixelMachineLearningwithClonal.pdf; http://eprints.utm.my/id/eprint/87364/; https://fazpublishing.com/ccam/index.php/ccam/article/download/4/2/

The early detection of lung nodules is critical to provide a better chance of survival from lung cancer. Since benign/malignant lung cancer may be caused by the growth of lung nodules, the diagnosis of an early detection of lung nodules is important. With rapidly development of advanced technology, detection of lung nodules becomes efficient by utilizing computer-aided detection (CAD) systems that can automatically detect and localize the nodules from computed tomography (CT) scans. CAD is fundamentally based on pattern recognition by extensive use of machine learning approaches which is highly interrelated to mathematical algorithms. In this study, a pixel machine learning algorithm which is developed by artificial immune system (AIS) based algorithm – Clonal Section Algorithm (CSA) is proposed for lung nodules visualization. By using pixel machine learning algorithm, several pre-processing procedures can be avoided to prevent the loss of information from image intensities. It is found that the proposed classification algorithm using original intensity values from CT scans is able to provide reasonable visualization results for lung nodules detection.