A research team led by Professor Kevin Tsia from the University of Hong Kong (HKU) has developed an AI-driven imaging tool for rapid and precise cancer diagnosis, significantly improving treatment effectiveness. Collaborating with HKUMed and Queen Mary Hospital, the team introduced Cyto-Morphology Adversarial Distillation (CytoMAD), a generative AI method that enhances lung cancer diagnosis and drug testing.
AI-Enhanced Cell Imaging
CytoMAD, combined with microfluidic technology, enables label-free imaging of human cells, allowing high-precision tumor assessment and metastasis risk identification. The AI system automatically corrects imaging inconsistencies, enhances images, and extracts previously undetectable cellular information, ensuring accurate diagnosis and analysis. Unlike traditional staining and labeling methods, CytoMAD simplifies sample preparation, reducing time and costs while improving efficiency in diagnosis and drug discovery.
Label-Free Contrast & Consistency
A key feature of CytoMAD is its ability to convert bright-field images into high-contrast, information-rich images, revealing cellular properties without the need for fluorescence markers. The AI model also overcomes batch effect inconsistencies, providing unbiased and generalizable analysis for diverse applications.
High-Speed Imaging & Future Outlook
Supported by ultrafast optical imaging technology, CytoMAD can generate millions of cell images daily, accelerating AI-based medical research. Beyond lung cancer, it holds potential for drug screening and early disease prediction. The team plans to further train the AI model to predict cancer risks and study lung cancer patients over the next three years.
Reference :
(News) https://medicalxpress.com/news/2025-02-ai-driven-tool-cancer-diagnosis.html
(Research) https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202307591