Lung cancer continues to be among the top causes of mortality globally, and early and precise detection is crucial in order to enhance survival rates. This paper discusses a mixed deep learning approach model proposed for lung cancer detection and classification from Computed Tomography (CT) scans. model uses VGG19 and ResNet50 separately, and a combined model that combines the two architectures to enhance feature extraction and classification performance. Extensive experiments were carried out on the publicly shared IQ-OTH/NCCD dataset, with 93% accuracy with ResNet50, 96% with Vgg19, and 98% with the proposed model (VGG19 + ResNet50), showing its effectiveness to distinguish between benign and cancerous lung nodules. The system can be incorporated with computer-aided diagnosis (CAD) software, assisting radiologists to identify early lung cancer and potentially improve patient outcomes.