This paper explores the use of pre-trained convolutional neural network (CNN) models for lung cancer detection from CT scan images. The primary objectives are to study state-of-the-art research, adopt pre-trained CNN models, evaluate their performance, and modify these models with additional layers and parameters. The pre-trained models, including ResNet-50, VGG-16, Xception, and MobileNet, achieve an accuracy range of 78% to 86%. The best performance is achieved by the pre-trained VGG-16 model with added fully connected layers, a batch size of 16, and the Adam optimizer, achieving 86.71% accuracy. The research also discusses the importance of early lung cancer detection and the limitations of chest radiography, emphasizing the need for computer-aided detection tools. The proposed method involves data preprocessing, feature extraction, and model training using Python and PyImage. The study uses the LIDC-IDRI dataset, which contains 7374 images after preprocessing. The results show that the pre-trained VGG-16 model outperforms other models, while machine learning models like KNN regression and SVM achieve 100% accuracy. The paper concludes by suggesting future improvements, such as using the DenseNet model and combining metadata with pre-processed imagery data.This paper explores the use of pre-trained convolutional neural network (CNN) models for lung cancer detection from CT scan images. The primary objectives are to study state-of-the-art research, adopt pre-trained CNN models, evaluate their performance, and modify these models with additional layers and parameters. The pre-trained models, including ResNet-50, VGG-16, Xception, and MobileNet, achieve an accuracy range of 78% to 86%. The best performance is achieved by the pre-trained VGG-16 model with added fully connected layers, a batch size of 16, and the Adam optimizer, achieving 86.71% accuracy. The research also discusses the importance of early lung cancer detection and the limitations of chest radiography, emphasizing the need for computer-aided detection tools. The proposed method involves data preprocessing, feature extraction, and model training using Python and PyImage. The study uses the LIDC-IDRI dataset, which contains 7374 images after preprocessing. The results show that the pre-trained VGG-16 model outperforms other models, while machine learning models like KNN regression and SVM achieve 100% accuracy. The paper concludes by suggesting future improvements, such as using the DenseNet model and combining metadata with pre-processed imagery data.