Lung cancer is a common and fatal disease worldwide. Early identification is essential to effectively treating the condition and enhancing patient outcomes. One way to improve lung cancer diagnosis is to use advances in deep learning. The current research discusses three, ResNet-50, Inception V3, and MobileNet, to detect lung cancer from computed tomography (CT) images. Based on the assessments, MobileNet demonstrated superior performance to the other analyzed networks, achieving an accuracy rate of 99\%. This indicates that MobileNet has an exemplary architecture for tasks related to lung cancer diagnosis compared to Restnet50 and Inception v3. The impact of data augmentation and normalization techniques on model performance was also investigated. These results determine the suitability of deep learning models for comparable medical imaging tasks.