Comparative Analysis of Image Processing Methods using GLCM and LBP Feature Extraction for Lung Cancer Detection

This research was conducted to compare image processing accuracy in detecting lung cancer using two feature extraction methods: grey-level co-occurrence matrix feature extraction and local binary pattern feature extraction. The image dataset used in this research is secondary data taken from The Iraq-Oncology Teaching Hospital/National Center for Cancer Diseases (IQ-OTH/NCCD), with the data source coming from the Kaggle website. The image dataset consists of three labels, namely normal lung images, benign lung cancer images, and malignant lung cancer images, where both types of lung cancer (benign and malignant) are combined into one label called lung cancer images. Two feature extraction methods were used, namely grey level co-occurrence matrix and local binary pattern feature extraction and the use of two machine learning methods in making classification decisions: the SVM and Gaussian Naive Bayes methods. The research results showed that the highest accuracy obtained was the combination of the local binary pattern feature extraction method and the machine learning method SVM with 93% accuracy. In comparison, the lowest accuracy obtained was using the grey level co-occurrence matrix and the machine learning method Gaussian naive Bayes with 50% accuracy.