Open AccessClassification of Lung Cancer Stages from CT Scan Images Using Image Processing and k-Nearest Neighbours
Author(s) -
Mohd Firdaus Abdullah,
Siti Noraini Sulaiman,
Muhammad Khusairi Usman,
Nor Khairiah A. Karim,
Ibrahim Lutfi Shuaib,
Muhamad Daniyal Irfan Alhamdu,
Adi Irfan Che Ani
Publication year - 2022
Publication title -
journal of human centered technology
Language(s) - English
Resource type - Journals
ISSN - 2821-3467
DOI - 10.11113/humentech.v1n1.6
Subject(s) - lung cancer , image processing , artificial intelligence , computer science , pattern recognition (psychology) , magnetic resonance imaging , data set , cancer , medicine , computer vision , radiology , image (mathematics) , pathology
Lung cancer is the prevalent cause of death among people around the world. The detection of the existence of lung cancer can be performed in a variety of ways, such as magnetic resonance imaging (MRI), radiography, and computed tomography (CT). Such techniques take up a lot of time and financial resources. Nevertheless, for the detection of lung cancer, CT provides a lower cost, fast imaging time, and increased availability. Early diagnosis of lung cancer may help physicians treat patients to minimize the number of deaths. This paper revolves around the categorization of lung cancer stages from CT scan images using image processing and k-Nearest Neighbor. The central objective of this study is therefore to establish an image processing technique for extracting features of lung cancer from CT scan images. Extracting the features from the segmented image can help to detect cancer inside the lung. The purposed method comprises the following steps by using image processing techniques: data collection, data pre-processing, features selection, and lung cancer classification. The pre-processing was done using a median filter to remove noise contained in the images. Three features need to be extracted which are area, perimeter, and centroid. Finally, the set of data with these features were used as inputs for lung cancer classification. By analysis results, the kNN method has a high accuracy of 98.15%.