AI Model Identifies Patients with High-Risk Form of Endometrial Cancer

Endometrial cancer is the most common gynecological cancer and varies widely in aggressiveness, with some forms more likely to return than others. This variability underscores the need to identify patients with high-risk endometrial cancer to tailor interventions and prevent recurrence. Researchers are now harnessing artificial intelligence (AI) to develop precision diagnostic tools for endometrial cancer, thereby enhancing patient care.

Researchers at the University of British Columbia (Vancouver, BC, Canada) utilized AI to analyze thousands of cancer cell images and identify a specific subset of endometrial cancer associated with a higher risk of recurrence and death, which might not be detectable through standard pathology and molecular diagnostics. This innovation is set to aid clinicians in identifying patients who require more aggressive treatment strategies. Building on their foundational research from 2013, which categorized endometrial cancer into four molecular subtypes, each with distinct risk levels, the team developed a molecular diagnostic tool called ProMiSE that effectively differentiates these subtypes. However, the most common molecular subtype, which accounts for about half of all cases, serves as a broad category for cancers that lack specific molecular characteristics.

To further segment the category using advanced AI methods, the team created a deep-learning AI model that examines patient tissue sample images. This model was trained to distinguish between subtypes, and after evaluating over 2,300 cancer tissue images, it identified a new subgroup with significantly lower survival rates. The researchers are considering how this AI tool could be incorporated into regular clinical practice alongside traditional diagnostics. An advantage of this AI approach is its cost-effectiveness and the ease with which it can be implemented widely. The AI reviews images typically collected and examined by pathologists, making it accessible for use in smaller medical facilities in rural and remote areas, often involved when seeking second opinions. By integrating molecular and AI-based analyses, many patients might continue receiving care in their local communities, reserving more complex treatments for those who need the resources of larger cancer centers.

“The power of AI is that it can objectively look at large sets of images and identify patterns that elude human pathologists,” said Dr. Ali Bashashati, a machine learning expert and assistant professor of biomedical engineering and pathology and laboratory medicine at UBC. “It’s finding the needle in the haystack. It tells us this group of cancers with these characteristics are the worst offenders and represent a higher risk for patients.” The results of the team's study were published in Nature Communications on June 26, 2024.

Related Links:
University of British Columbia
Gynecologic Cancer Initiative

Visit expo >

Gold Member

Fibrinolysis Assay

HemosIL Fibrinolysis Assay Panel

POC Helicobacter Pylori Test Kit
Hepy Urease Test

Automated Chemiluminescence Immunoassay Analyzer

MS-i3080

Automatic CLIA Analyzer

Shine i9000