<p>It must have to be a proper and early diagnosis of lung cancer since it remains among the leading causes of cancer related mortality all over the world. Despite the potential in deep learning as well as traditional machine learning in the diagnosis of lung cancer, both tend to fail in extracting features and cross-imaging modalities. The study provides a state-of-the-art structure, Multimodal Biomedical Imaging and Deep Learning Models (MMBI-DLM) of lung cancer segmentation and classification with the help of multimodal biomedical imaging and transformer-based deep learning. We combine computed tomography (CT), positron emission tomography (PET) and histopathological imaging in our method, which enhances the accuracy of classification, representation of features as well as precision in segmentation. In order to derive local features, long-range associations, and global context data, we apply convolutional neural networks (CNNs) and vision transformers (ViTs) to our hybrid architecture. Besides generating high-resolution segmentation maps, this fusion increases the ability of this model to distinguish between benign and malignant tumors. We make sure that we are resilient in different imaging conditions through additional optimization of the performance through attention mechanisms and multi-scale fusion of features. The proposed MMBI-DLM framework offers better performance than state-of-the art deep learning models in terms of computational efficiency, segmentation and classification accuracy as evidenced by experimental studies in benchmark datasets. The findings suggest that multimodal imaging, together with deep learning based on transformers, has a significant and impressive effect on the diagnosis and characterization of lung cancer.</p>