Researchers at the University of Southampton have developed an artificial intelligence tool that reveals previously unseen patterns inside breast cancers, a breakthrough that could help doctors identify high-risk patients and develop more targeted treatments.
The AI platform, called CenSegNet, was used to analyse more than 330,000 cells from 127 patients, uncovering two distinct types of changes inside cancer cells that had previously been treated as the same thing. One type involved cells developing too many centrosomes, which help cells divide properly, while the other involved centrosomes becoming unusually large.
Researchers discovered the two changes could appear in different parts of the same tumour and may play different roles in how cancers develop.
Centrosomes act as the cell’s organising centres, helping cells divide and maintain their structure. When they become abnormal, cells can build up genetic mistakes, which are commonly seen in cancer. The team created CenSegNet to examine tumour samples with “unprecedented speed and precision.”
The findings could help doctors better understand which cancers are more likely to grow, spread or resist treatment. The study found that tumours with higher numbers of enlarged centrosomes were more likely to have features linked to more aggressive disease, including higher tumour grade, cancer that had spread to nearby lymph nodes and certain genetic changes.
Conversely, the study also found that patients with fewer enlarged centrosomes in the centre of their tumours tended to have better overall survival.
Leading the research, the university’s Dr Salah Elias said the technology opens the door to developing new biomarkers and, ultimately, more personalised treatment strategies. The tool could help scientists identify tumours with specific weaknesses that could be targeted with new drugs.
“For more than a century, centrosome abnormalities have been recognised as a hallmark of cancer, but studying them in patient tissues has been extremely challenging,” Elias said. “CenSegNet allows us to analyse these defects at single-cell resolution across entire tumours and uncover patterns that were previously impossible to see.”
While the technology is not yet ready for routine use in hospitals, the team has also shown it can be used on tissue samples from other parts of the body, including the kidney, colon and appendix. CenSegNet is available free of charge as open-source software, allowing researchers around the world to use and study the technology.
The research represents a significant step forward in understanding the complexities of cancer and using AI to unlock new insights that could lead to more effective and personalised treatments for patients.
