A Domain-Adaptive Deep Contrastive Learning Network for Lung Cancer CT Image Diagnosis in Internet of Medical Things

Lung cancer is one of the most prevalent and deadly malignancies worldwide, making early and accurate diagnosis critically important. Although computed tomography (CT) imaging is widely used for lung cancer detection, conventional diagnosis still relies heavily on manual interpretation, which is prone to human error and limits diagnostic efficiency and accuracy. Moreover, existing deep learning models often suffer from poor domain adaptability and insufficient feature discriminability in real-world multi-center scenarios. To address these challenges, this paper proposes a Domain-Adaptive Contrastive Learning Network (DACLNet) for lung cancer CT image classification within the Internet of Medical Things (IoMT) framework. The proposed method integrates domain adversarial adaptation with supervised contrastive learning to explicitly alleviate inter-domain data distribution shifts while enhancing intra-class compactness and inter-class separability in the learned feature space. In addition, DACLNet is designed to support flexible deployment on both edge and cloud platforms, enabling efficient remote diagnosis and multi-center collaboration. Extensive experiments conducted on real-world lung cancer CT datasets demonstrate that DACLNet achieves an Accuracy of 0.930, an F1-score of 0.936, and an AUC of 0.986 on the test set, outperforming all baseline methods across all evaluation metrics. These results validate the effectiveness, robustness, and cross-domain generalization capability of the proposed approach. Overall, DACLNet provides a reliable and scalable solution for intelligent lung cancer diagnosis in IoMT-enabled healthcare systems.