Lung cancer remains a leading cause of cancer-related deaths, emphasizing the need for early and accurate diagnosis. Advances in AI-driven medical imaging, particularly CT scans, have significantly improved detection and classification while reducing the workload of radiologists. By automating repetitive tasks and prioritizing high-risk cases, deep learning systems enable faster, more consistent analyses, allowing clinicians to focus on complex decision-making. For lung cancer detection, traditional machine learning methods like Support Vector Machines (SVM) and K-Nearest Neighbor (KNN) have pros and cons [13], while deep learning models such as CNNs demonstrate superior performance, reaching 94.11% accuracy. However, performance varies with dataset size and model tuning, as seen in a CNN with 86.06% accuracy when trained on larger datasets. In lung cancer classification, deep learning models excel in distinguishing between benign, malignant, and subtypes like Adenocarcinoma (AC) and Squamous Cell Carcinoma (SCC). ResNet50 achieves the highest accuracy (98%), followed by VGG19 (97%) and EfficientNetB7 (96%). Traditional methods like the Johnson Reducer (JR) and Repeated Incremental Pruning to Produce Error Reduction (RIPPER) algorithms also perform well, with 95% accuracy for SCC classification. Comparative studies confirm that CT scans outperform PET scans in classification tasks, reinforcing their diagnostic superiority. These findings highlight the transformative potential of AI in lung cancer diagnostics, not only in improving accuracy but also in streamlining clinical workflows. However, the importance of model optimization and integration into real-world practice remains critical.