Optimized feature selection for enhancing lung cancer prediction using machine learning techniques

Abstract

One of the primary causes of deaths related to cancer worldwide is Lung cancer, and this condition is widespread in India. History of the patient and their histological classification in terms of lung cancer provides critical information regarding the characteristics of tissues and anatomical locations. The cancer symptoms normally appear only in the advanced stages, so it is very hard to detect, resulting in a high mortality rate among the other types of cancers. Histopathology images of lung scans are used for the classification of lung cancer using image processing methods. Radiomics has been recognized as an effective technique using a quantitative image feature of high throughput for diagnosis; it uses the images as data and performs data mining for predicting. Many different studies have depicted the radiomic features and their power of prediction in detecting lung cancer. The features are extracted from its large corpus belonging to the training data in which an object of interest is described. The Algorithm can extract all its quantitative features. But, its quantitative size in terms of data is large and leads to major challenges during classification. To overcome this, a symbolic approach to data analysis is proposed. For symbolic data state, histology of lungs was obtained over a maximum of three different time intervals. The work examines various feature selection methods to predict lung cancer histologic subtypes employing radiomics and symbolic procedures. The features have been extracted using a Grey Level Co-Occurrence Matrix (GLCM) and the Gabor filter, which are fused using Z-Score Normalization. newline