Women globally may receive enhanced care with a revolutionary AI technology that improves the detection of damaged cells and more accurately forecasts the likelihood of developing breast cancer, as per new findings from the University of Copenhagen.
Women globally may benefit from improved care thanks to innovative AI technology that enhances the identification of damaged cells and predicts breast cancer risk more accurately, according to research from the University of Copenhagen.
Breast cancer ranks among the most prevalent cancer types, leading to approximately 670,000 fatalities globally in 2022. Recent research from the University of Copenhagen indicates that AI can significantly aid in women’s treatment by identifying abnormal cells, thereby refining risk assessments.
This study, featured in The Lancet Digital Health, discovered that the AI system outperformed conventional clinical methods for evaluating breast cancer risk.
Utilizing deep learning AI technology crafted at the University of Copenhagen, researchers examined tissue biopsies to identify indications of damaged cells, which serve as risk markers for cancer.
“This algorithm represents a significant enhancement in our capability to pinpoint these cells. With millions of biopsies conducted annually, this technology can facilitate more accurate risk identification, leading to improved treatment for women,” states Associate Professor Morten Scheibye-Knudsen from the Department of Cellular and Molecular Medicine and the study’s senior author.
Identifying cases with fivefold breast cancer risk
Critical to evaluating cancer risk is the detection of dying cells, a phenomenon linked to cellular senescence. Senescent cells, while still metabolically active, have ceased to divide. Previous studies have demonstrated that this senescent condition can hinder cancer progression, though these cells may also trigger inflammation that promotes tumor growth.
Employing deep learning AI to find senescent cells in tissue biopsies allowed researchers to more accurately predict the risk of breast cancer compared to the existing Gail model, the current standard for evaluating breast cancer risk.
“We discovered that merging two of our models or integrating one model with the Gail score yielded significantly improved predictions of cancer risk. One particular model pairing resulted in an odds ratio of 4.70, which is substantial. This means we can examine cells from a healthy biopsy and predict that the donor has nearly five times the chance of developing cancer in the coming years,” notes Indra Heckenbach, the study’s lead author.
Algorithm trained on ‘zombie cells’ enhances treatment
The researchers educated the AI using intentionally damaged cells cultivated in the lab to induce senescence. The AI was then applied to donor biopsies to locate senescent cells.
“We often refer to these as zombie cells because they have lost some functionality but are not entirely dead. They are linked to cancer progression, so we developed and fine-tuned the algorithm to identify cell senescence. Specifically, our algorithm examines the shapes of cell nuclei, which become more irregular in senescent cells,” explains Indra Heckenbach.
While it may take several years for this technology to be implemented in clinics, its application could be global, as it solely requires standard tissue sample images for analysis. This advancement could empower women worldwide to access better treatment options, adds Morten Scheibye-Knudsen:
“We can leverage this information to categorize patients by risk levels and enhance treatment and screening strategies. Medical professionals can closely monitor high-risk individuals, administering more frequent mammograms and biopsies, which might enable earlier cancer detection. Simultaneously, we can alleviate the frequency of biopsies for low-risk individuals.”
