Identification of Lung Lesions in Computed Tomography Scan Images using Machine Learning Techniques

• الحاوية / القاعدة: 14th IEEE International Conference on Control System, Computing and Engineering, ICCSCE 2024 - Proceedings
• المؤلف الرئيسي: 2-s2.0-85207057349
• التنسيق: Conference paper
• اللغة: English
• منشور في: Institute of Electrical and Electronics Engineers Inc. 2024
• الوصول للمادة أونلاين: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85207057349&doi=10.1109%2fICCSCE61582.2024.10696663&partnerID=40&md5=2ace7900a684dbaf59489f3bb74d770a

Lung cancer is a common cause of death among people throughout the world. Other methods for detecting lung cancer include computed tomography (CT), magnetic resonance imaging (MRI), and radiography. This method suggests CT is useful for detecting lung cancer since it is widely available, has a shorter imaging time, and is less expensive. For clinical analysis and effective preventive planning by medical authorities to lower the number of fatalities, early diagnosis of lung lesions is crucial. Visual score is a typical method used by skilled radiologists to manually identify lesions on CT images. Nevertheless, the manual approach is labor-intensive, time-consuming, tedious, and visible. This research proposes a method for automatically detecting soft lung tissue lesions for CT images. To detect lung lesions from CT scans, the system employs image processing and machine learning techniques. The proposed automatic identification system divided into three stages. The first stage is image acquisition and data collection. The second stage proposed method involved designing a procedure for lung region segmentation using an image processing technique specifically for lung region for detection of lesions. The final stage explains an automated lesion identification for further classifying the lesion and non-lesion from CT scan lung images based on geometrical properties that have been calculated. Finally, a system that utilized the best of above-mentioned methods is proposed to perform the automated lesion detection. The method achieved high capability for automatically identifying lung lesions near the manually delineated lesion by radiologists with 98% accuracy. These findings show the possible used of this approach as an assisted tool for the radiologist in detecting lung lesions. © 2024 IEEE.