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

Area of Science:

• Oncology
• Medical Imaging
• Artificial Intelligence

Background:

• Bladder cancer grading is crucial for treatment but traditionally subjective.
• Manual pathological slide assessment leads to diagnostic inconsistencies.

Purpose of the Study:

• To develop a precise bladder cancer grading method using deep learning.
• To integrate hyperspectral imaging (HSI) and pathological images for multimodal analysis.

Main Methods:

• Proposed a deep learning multimodal fusion model named RVCK-net.
• Leveraged spatial and spectral information from HSI and pathological images.
• Employed an adaptive fusion mechanism for robust classification.

Main Results:

• Achieved an average accuracy of 94.1% via 10-fold cross-validation.
• Significantly outperformed single-modality grading approaches.
• Demonstrated improved diagnostic consistency in bladder cancer grading.

Conclusions:

• Multimodal deep learning shows significant potential for bladder cancer diagnosis.
• RVCK-net offers a more objective and accurate method for bladder cancer grading.
• This approach can enhance early diagnosis and guide personalized treatment strategies.