The effectiveness of therapy for lung cancer is greatly increased by early identification. Deep learning (DL) techniques are efficient approaches to the early detection of malignant nodules in the lungs. In order to examine CT scans, identify anomalies, and categorize tumors as either benign or malignant, this research presents the Deep Nodule-Detect, a computer-aided detection (CAD) method that uses a group of convolutional neural networks (CNNs). The IQ_OTH/NCCD lung cancer dataset, which includes 1,097 CT scan pictures of both healthy and malignant cases, was used to create and instruct 3 CNN models. Our model achieved 95.43% accuracy, 93.40% sensitivity, and 97.09% specificity for a three-class classification (normal, benign, malignant). For binary classification (cancerous vs. noncancerous), it achieved 98.17% accuracy, 98.21% sensitivity, and 98.13% specificity. Comparative analysis showed that Deep Nodule-Detect outperformed the fine-tuned AlexNet and previous models on the same dataset. Additionally, we investigated the role of region-of-interest (ROI) segmentation and data augmentation in improving medical image analysis. These techniques enhanced model performance, demonstrating their importance in deep learning-based lung cancer detection.