Abstarct: Lung cancer is one of the most common and deadliest types of cancer worldwide. Early detection of this disease can significantly increase survival rates and improve patient outcomes. However, the use of deep learning models for lung cancer detection from CT scan images faces challenges such as limited training data and concerns related to patient privacy. In this research, a federated learning–baxsed approach using a CNN (Convolutional Neural Network) baxse model is proposed. This method enables model training across multiple clients without the need to directly share patients’ data. In addition, a data augmentation technique has been used to increase training samples and improve model generalization. The proposed method was evaluated on the IQ-OTH/NCCD dataset and achieved an average accuracy of 99.61%. The evaluation results show that this approach provides strong performance in lung cancer detection and, compared with conventional methods, creates an appropriate balance between diagnostic accuracy and preservation of patient privacy. This study represents an effective step toward leveraging advanced artificial intelligence technologies to improve the diagnosis of complex diseases such as lung cancer.