Single-cell RNA insights and Densely Scaled Vision Transformer-Based MRI Classification for Precision Brain Tumors.
Authors
Abstract
There are more than 1,600 evolutionarily conserved RNA-binding proteins (RBPs) in the human genome. Many multi-omics studies have demonstrated that these proteins are often not working properly in malignancies like glioblastoma and melanoma. These RBPs are very important for the complex regulatory networks that govern the activities that are typical of cancer. RBPs' intricate control of RNA activity at many levels and their post-translational modifications, which make them more functional, make things even more convoluted. Additionally, other RBP-based therapies have emerged, each underpinned by distinct molecular mechanisms, including genomic analysis and the inhibition of RBP functionality. This paper reports findings from patients with brain tumours undergoing experimental RNA interference treatment. We also suggest a Densely Scaled Vision Transformer (DSViT) made to find and locate brain tumors of different types. The model is evaluated on the FigShare Brain Tumor Dataset comprising 3064 MRI images categorized into Glioma, Meningioma, and Pituitary tumors, with final testing conducted on 614 samples. Experimental results show that DSViT achieves an accuracy of 96.09%, precision of 96.57%, recall of 95.97%, and an F1-score of 96.27%, significantly outperforming the ViT-Baseline and ablation variants. Future directions include extending DSViT into multimodal pipelines that fuse imaging with molecular profiles, thereby enhancing precision neuro-oncology. Its modular structure also enables integration into radiological reporting systems for automated annotation and clinician-guided decision support. This innovative RNA interference (iRNAi) based therapeutic intervention has significant therapeutic potential and is, as far as we are aware, the first time RNA interference has been used to treat human disease.