Comparative Analysis of Machine Learning Algorithms for the Effective Detection of Lung Cancer

Abstract

Cancer is becoming the major reason of mortality. Automatic detection of lung cancer leads to early diagnosis and appropriate treatment. This work describes the development of an automated system that detects lung cancer using machine learning. The created system can capture medical images through computerized tomography. The model proposed here is developed using DCT for trait selection and SVM, KNN, Random Forest, Naive Bayes, linear regression and logistic regression classifiers for classification. The proposed system accepts medical images and efficiently detects cancer cells from CT images. Superpixel segmentation is utilized for the purpose of extracting the region of interest from the CT images and Gabor filter is applied for denoising the images. In the cancer detection system, the effectiveness of each of the above-mentioned classifiers was compared based on the parameters such as accuracy, precision, F1 score, MCC and error rate.