Innovative AI model for bladder cancer diagnosis

Area of Science:

• Medical Imaging
• Artificial Intelligence in Oncology
• Urologic Oncology

Background:

• Bladder cancer is a common urinary system malignancy.
• Early diagnosis and treatment are vital for improving patient prognosis.
• Current diagnostic methods can be improved with advanced technologies.

Purpose of the Study:

• To develop and validate an artificial intelligence (AI) model for accurate bladder cancer diagnosis.
• To assess the diagnostic performance of the AI model using CT imaging data.
• To ensure transparency in AI-driven tumor detection.

Main Methods:

• Developed a deep learning-based AI model for bladder cancer detection.
• Trained and validated the model using retrospective CT imaging data from a hospital and The Cancer Imaging Archive (TCIA).
• Evaluated model performance using various metrics and utilized Grad-CAM for visualization and radiologist review.

Main Results:

• The AI model achieved a test-set accuracy exceeding 0.90.
• Grad-CAM successfully highlighted tumors, confirming the AI model's ability to localize cancerous regions.
• The model demonstrated accurate and transparent bladder cancer detection capabilities.

Conclusions:

• The developed AI model provides accurate and transparent bladder cancer detection on routine CT scans.
• The AI model shows promise for enhancing early diagnosis and treatment planning.
• Prospective validation of the AI model is warranted to confirm its clinical utility.