BGR – Endometrial cancer is one of the most common cancers in women—and one of the hardest to catch. But a new breakthrough in AI cancer detection is pushing accuracy levels to an unprecedented 99 percent, giving doctors a major new tool in the fight.
This is a massive leap in accuracy from previous AI models, which could only accurately detect the cancer up to 80 percent of the time. Additionally, the new model uses fewer resources, making it faster and more accessible.
What makes the model—named ECgMLP—so effective is how it processes visual data. It enhances the image and then filters out irrelevant noise. This lets it zero in on the most informative areas of tissue, an important factor for AI cancer detection tools.
Then, using advanced self-attention mechanisms—a kind of digital pattern recognition—the model rapidly evaluates the tissue and delivers a diagnostic prediction with impressive accuracy.
We’ve already seem similar developments in breast cancer detecting AI, so this is yet another win for AI in the medical field.
Current automated systems for detecting endometrial cancer top out around 80 percent accuracy, as I mentioned before. However, ECgMLP surpasses that by nearly 20 percentage points, all while using fewer resources. It’s fast, precise, and built to work across a variety of datasets.
The implications of this AI cancer detection model go beyond just one type of cancer, though. When tested on other datasets, ECgMLP correctly identified colorectal cancer with 98.57 percent accuracy, breast cancer at 98.2 percent, and oral cancer at 97.34 percent.
This versatility opens the door to a broader use of the same technology in diagnostics across medical fields.
Plus, researchers say the model could eventually be integrated into clinical software, supporting doctors in decision-making and improving outcomes through earlier intervention …
