Researchers at the University of Jyväskylä in Finland say they have developed an artificial intelligence model that can help analyze colorectal cancer tissue samples more quickly and predict how a key DNA repair mechanism is functioning. The team says the approach could shorten diagnosis times, lower costs and improve the accuracy of pathology work.
The study was carried out by the university’s Faculty of Information Technology in collaboration with the Central Finland Welfare Region. It was also supported by the European Union.
At the center of the work is mismatch repair, or MMR, a cellular system that helps fix small mistakes that happen when DNA is copied. When this mechanism fails, it can influence both how cancer develops and how doctors choose treatment. The researchers say assessing whether MMR is working is already part of routine pathology, but the process is slow and labor intensive.
Liisa Petäinen, who led the study, said the difference in turnaround time can be dramatic. In a pathology lab, analysis related to the MMR mechanism can take days, while the AI model can reduce that to minutes. The researchers argue that faster results could help patients receive diagnoses and treatment sooner, while also freeing pathologists to focus on other tasks.
Current analysis usually relies on a pathologist examining the tumor area at high magnification. In the new study, the researchers also tested whether AI could work effectively with a broader view of the tissue at lower magnification. According to the team, the model performed reasonably well even at that scale.
That finding matters because Petäinen says future systems might be able to analyze an entire tissue sample in one step. If the model can evaluate the full slide rather than first locating the tumor region separately, screening could become faster and less dependent on manual pre-selection.
The researchers also reported that tissue around the tumor may contain useful clues about the state of the repair mechanism. That could make whole-sample analysis not only quicker, but also more informative.
The AI system was trained on data from about 1,300 colorectal cancer patients in Central Finland. The researchers worked with pathologists and colorectal cancer specialists from the Central Finland Biobank and the Wellbeing Services County of Central Finland. The model was also tested using data from Oulu University Hospital and from the United States.
Tiina Jokela, another researcher involved in the project, said Finland’s biobanks, health registries and unified healthcare structure make it easier to conduct this kind of translational research. She described Central Finland as a useful pilot environment where clinical work and research can move together more flexibly.
The team said new methods still need validation in larger datasets before they can be more widely adopted, and this study included that kind of testing. The work is part of the AI-Hub II project in Central Finland, which is co-funded by the European Union and supported by regional partners.
For now, the researchers present the model as a tool that could make colorectal cancer diagnostics more efficient rather than a replacement for human expertise. But they suggest that AI-assisted analysis may eventually become a faster route from tissue sample to clinical decision.