Lung cancer is the most common disease in the world, and early detection is critical to diminish mortality rate. Radiological images, such as computed tomography (CT) scans and chest X-rays, are usually analyzed by experienced radiologists and oncologists as part of the standard way of diagnosing cancer. Key traits like the size, shape, and density of the nodule are checked by hand to see whether they are cancerous. A computer-aided diagnosis (CAD) system is built in order to make the diagnostic process go more quickly. CAD uses machine learning (ML), a type of artificial intelligence (AI), to find abnormalities in images and help doctors figure out what is wrong. CNN use convolutional layers to record feature hierarchies in space. This work looks into how convolutional neural networks (CNN) with VGG19 and ResNet50 designs can be used to sort radiological images of lung cancer into different groups. The IQ-OTH/NCCD Lung CT image dataset used in this study has the best accuracy of 0.9952 for CNN rather than the transfer learning models. So, it can be asserted that CNN models consistently deliver excellent performance, exhibiting higher accuracy rates in both training and validation phases. CNN shows how useful they are as a powerful tool for image classification tasks by getting great results and exhibiting new uses in many areas.