Enhancing Skin Cancer Detection Across Diverse Skin Types

BBSRC funded with Check4Cancer, UK (2024)

  • Funding Amount: £15,100.51
  • Funder: BBSRC IAA
  • Role: Principal Investigator (PI)
  • Academic Team: Dr Haider Raza (PI) and Professor John Q Gan.
  • Associate: Mohsin Ali and Rahmat Ibrahim
  • Industry Lead: Professor Gordan Wishart
  • Year: 2024

The AI model’s performance, using a combination of image data and metadata (7C4C risk factors and C4C risk score) for overall skin cancer detection, achieved an accuracy of 99.66%. The individual results were 99.37% for melanoma, 100% for squamous cell carcinoma, 99.70% for basal cell carcinoma, and 74.45% for benign (non-malignant) cases. Check4Cancer has initiated the processes to file a patent application and register the AI model as a Type IIa medical device. However, the current AI model can only be commercialised in countries with skin tones similar to those in the UK, including Western Europe, USA, Canada, Australia, and New Zealand. We want to enhance the performance of our AI model on darker skin tones.

References