Non-small-cell lung cancer prediction using radiomic features and machine learning methods. Issue 12 (2nd December 2022)

• Record Type: Journal Article
• Authors: Shanthi, S.; Rajkumar, N.
• Is Part Of: International journal of computers and applications. Volume 44:Issue 12(2022)
• Page Start: 1161
• Page End: 1169
• Publication Date: 2022-12-02
• Subjects: Lung cancer -- non-small-cell lung cancer (NSCLA) -- gray-level co-occurrence matrix (GLCM) -- minimum redundancy maximum relevance (MRMR) -- correlation-based feature selection (CFS) -- Naïve Bayes and decision tree; Computers -- Periodicals; Computer software -- Periodicals; Computer networks -- Periodicals; Multimedia systems -- Periodicals; Internet -- Periodicals; World Wide Web -- Periodicals; Minicomputers -- Periodicals; Microcomputers -- Periodicals; 004.05
• DOI: 10.1080/1206212X.2019.1693723
• Languages: English
• ISSNs: 1206-212X

Abstract

Abstract : One of the primary causes of deaths related to cancer all over the world is Lung cancer. The history of the patient and his histological classification in terms of lung cancer has provided critical information regarding the characteristics of tissues and anatomical locations. There are many different studies that have depicted the radiomic features and their power of prediction in the detection of lung cancer. But its quantitative size in terms of data is large and has been resulting in major challenges in the algorithms of classification. In order to overcome this, symbolic approach to data analysis which employs many different quantitative data is proposed. The work further investigates different techniques of feature selection in order to predict the histologic subtypes of lung cancer by using either symbolic data or the radiomic features. These features have been extracted by using a gray-level co-occurrence matrix (GLCM), the Gabor filter and the fusion that was achieved by making use of concatenation once there is a normalization of the Z score. The results of the experiment have proved that the proposed method had a better performance compared to the other methods.