MACHINE LEARNING APPROACHES FOR THE DETECTION OF LUNG CANCER USING MRI IMAGES

Abstract

Lung cancer ranks as the second most prevalent form of cancer worldwide, resulting in thousands of deaths annually. Nevertheless, the mortality rate can be mitigated by enhancing early detection and successful treatment, thereby bolstering the survival prospects of patients. There are different types of electronic modalities, e.g., CT/PET Scan, MRI, X-Ray etc. for lung diagnosis. With the advancement of technologies MRI is being used widely for lung cancer detection. But, the interpretation of MRI image is totally expert dependent and time consuming. An automated computerized approach can make lung cancer identification easier and more reli able. This study describes a fully automated technique for lung cancer detection using lung MRI and following two different approaches, i.e., conventional image processing approach and machine learning approach. The proposed conventional image processing method pro- vided an accuracy of 96.28%. However, CNN and SVM were used in machine learning approach and the classification accuracy were 96.55% and 90.5% respectively.