Klasifikasi Multi-Kelas Kanker Paru Berbasis Ekstraksi Fitur Hibrida GLCM, LBP dan Gabor Filter Menggunakan Algoritma Random Forest dan KNN pada Citra CT-Scan

Iqbal, Muhammad
Pandji Triadyaksa
Qidir Maulana Binu Soesanto

Jurnal Fisika Unand
Vol 15 No 1 (2026)

This study aims to improve the multi-class classification performance of Non-Small Cell Lung Cancer (NSCLC) based on CT-Scan images through a hybrid feature extraction approach. The method combines Gray Level Co-occurrence Matrix (GLCM), Local Binary Pattern (LBP), and Gabor Filter features, classified using Random Forest and K-Nearest Neighbor (KNN) algorithms. The dataset includes four image classes: adenocarcinoma, large cell carcinoma, squamous cell carcinoma, and normal. Model performance was evaluated using accuracy, sensitivity, specificity, and area under curve (AUC) metrics. The results show that the GLCM+LBP feature combination with the Random Forest algorithm achieved the best performance with an accuracy of 99.22%. The model effectively recognizes both global and local texture variations of lung tissue and remains stable in distinguishing all image classes. It can be concluded that the hybrid texture feature combination and ensemble algorithm produce an accurate, efficient, and potentially applicable classification model for computer-aided diagnosis in medical physics