A novel hybrid approach for multi stage kidney cancer diagnosis using RCC ProbNet

Area of Science:

• Oncology
• Medical Imaging
• Artificial Intelligence

Background:

• Kidney renal cell carcinoma (RCC) is a significant health concern, with early diagnosis crucial for effective treatment and reduced mortality.
• Current diagnostic methods may lack the precision needed for granular classification of RCC stages.

Purpose of the Study:

• To develop and validate a novel hybrid deep learning model, RCC-ProbNet, for enhanced precision in identifying and classifying kidney renal cell carcinoma (RCC) across various stages.
• To improve diagnostic accuracy and differentiate between RCC stages more effectively.

Main Methods:

• Introduction of RCC-ProbNet, a hybrid deep learning model combining feature extraction from medical imaging with a probabilistic feature model.
• Integration of a Logistic Regression (LR) classifier for final stage prediction.
• Performance validation using k-fold cross-validation.

Main Results:

• The RCC-ProbNet + LR model achieved a diagnostic accuracy of 99.93%, surpassing current state-of-the-art techniques.
• The model demonstrated strong stability across different cross-validation folds.
• Comparative analysis showed consistent outperformance against traditional and other deep learning methods for RCC classification.

Conclusions:

• RCC-ProbNet + LR is a highly accurate and stable classifier for early RCC screening and staging.
• The model's performance suggests potential for assisting in timely, personalized treatments and clinical decision support systems.
• This approach holds promise for optimizing patient outcomes in renal oncology.