Lung cancer remains one of the deadliest cancers worldwide, necessitating early detection for improved patient outcomes. This study proposes a novel image processing methodology for detecting and classifying lung tumors from CT scan images, differentiating between malignant, benign, and normal cases. The method involves a multi-step approach including channel separation, thresholding, grayscale conversion, mask creation, and diaphragm removal. The unique aspect of this approach is the emphasis on the red channel for thresholding, based on histogram equalization, and the subsequent removal of the diaphragm to eliminate obstructions in the lung window. Post-processing steps involve binarization, complementation, hole filling, and border smoothing to enhance tumor detection. The methodology was evaluated using a comprehensive dataset, i.e., IQ-OTH/NCCD. Experimental results demonstrate high accuracy in tumor detection and classification, i.e., approximately 95% of images in each class are successfully recognised.  This research contributes to the advancement of computer-aided detection systems, offering a practical and efficient solution for improving diagnostic accuracy in lung cancer screening.
 

lung cancer,machine learning,Image Processing