Hyperspectral Imaging for Predicting Bladder Cancer Grading: A Novel Diagnostic Approach

Bladder cancer is a common malignancy of the urinary system, where accurate grading plays a key role in guiding personalized treatment and improving patient outcomes. Traditional grading methods rely on manual assessment of pathological slides, which are prone to subjective bias. This paper proposes a deep learning-based multimodal fusion model, named RVCK-net, which integrates hyperspectral imaging (HSI) and pathological images to achieve precise bladder cancer grading. By leveraging spatial and spectral information from both modalities and employing an adaptive fusion mechanism, the proposed model achieves robust and reliable classification. Experimental results show that the method reaches an average accuracy of 94.1% under 10-fold cross-validation, significantly outperforming single-modality approaches and demonstrating improved diagnostic consistency. This study highlights the potential of multimodal deep learning for enhancing early diagnosis and accurate grading of bladder cancer.