AI Diagnosis of Sarcoma Cancer Twice as Accurate as Biopsies, Could Improve Patient Outcomes: Lancet Study

AI Algorithm Trains on 170 CT Scans, Achieving 82% Accuracy in Assessing Tumor Aggressiveness and Showing Potential for Global Use and Other Cancer Types, Demonstrating the Power of AI in Oncology

• An AI algorithm developed by a team at the Royal Marsden NHS Foundation Trust and the Institute of Cancer Research achieved an accuracy rate of 82% in grading the aggressiveness of sarcoma, a cancer that develops in the body’s connective tissues. The algorithm's accuracy is almost double the 44% rate of current biopsy examinations.
• The AI tool was trained using CT scans from 170 patients with two types of sarcoma and then tested on 89 patients from centers in Europe and the US. The AI was able to differentiate between two types of sarcoma better than radiologists.
• The team aims to apply this AI technology for the diagnosis and grading of a wider range of cancers. This could allow for faster diagnosis and more personalized treatments, thereby improving patient outcomes.
• There is an urgent need for treatment improvements for retroperitoneal sarcoma—a form of sarcoma developing in the back of the abdomen—as it is rare, hard to diagnose, and challenging to treat. The AI algorithm has shown promising results for this particular type of sarcoma.
• The UK government has recognized the potential of AI in revolutionizing cancer care and is investing over £1.2 billion a year into research for the development of treatments and diagnostic tools. Additionally, Prime Minister Rishi Sunak announced a £100 million fund for AI research into diseases previously deemed incurable.

3 Articles

The Next Web BBC The Telegraph