Researchers developed a new computer system designed to help analyze images of gastrointestinal cancer tissue. This system uses advanced artificial intelligence to look at patterns in microscope slides, aiming to assist in diagnosing and studying cancer. The team tested this new tool using about 99,000 digital images of stained tissue samples from two different medical databases.
The computer model successfully identified cancer cells and measured specific areas within the tissue. It achieved a high score of 0.938 for correctly classifying cancer types and a score of 0.839 for accurately outlining tumor regions. These results were compared against older, standard computer tools, and the new model performed better in these specific tasks.
This study was conducted in a controlled setting using carefully selected data, meaning it is currently a preclinical validation system rather than a tool ready for immediate use in hospitals. While the results are promising, the researchers note that further testing with larger groups of patients and real-world hospital workflows is needed before this technology can be used routinely in clinical practice.