Commonly, lung cancer is one of the leading causes of cancer-related deaths due to the lack of early symptoms, which hinders immediate diagnosis. This study aims to develop an advanced computer-aided diagnosis (CAD) system that enhances the early detection of lung nodules using computed tomography (CT) scan images. The proposed methodology involves three main stages: (1) pre-processing to enhance image quality through contrast adjustment and resizing, (2) feature extraction using a pre-trained Vision Transformer (ViT) model via transfer learning to obtain robust feature representations, and (3) classification of lung nodules into malignant, benign, and normal categories. Four machine learning algorithms—Random Forest (RF), Support Vector Machines (SVM), Decision Tree (DT), and eXtreme Gradient Boosting (XGBoost)—were used for classification. The results demonstrate that SVM achieved the highest accuracy of 98%, outperforming other models. This work highlights the effectiveness of combining transfer learning for feature extraction with powerful machine learning classifiers to develop a reliable system for early lung cancer detection, offering promising advancements for medical imaging diagnostics.