With the invasion of artificial intelligence (AI) into every pore of our everyday life, it is sometimes shocking how quickly it is becoming smarter and more capable – including in cancer detection, where a new AI model has demonstrated a near-perfect success rate.
Indeed, an international team of scientists, including those from Australia’s Charles Darwin University (CDU), has developed a novel AI model called ECgMPL, which can analyze microscopic images of cells and tissue to identify endometrial cancer, per a New Atlas report published on March 20.
AI outperforms in cancer detection
As it happens, the AI model’s accuracy rate has reached an impressive 99.26% accuracy, and researchers believe it is possible to adapt it to identify a wide array of diseases, including colorectal and oral cancer. Commenting on the study’s results, its co-author, Dr. Asif Karim from CDU, explained:
“The proposed ECgMLP model outperforms existing methods by achieving 99.26% accuracy, surpassing transfer learning and custom models discussed in the research while being computationally efficient.”
Furthermore, ECgMLP makes for a “robust and clinically applicable solution for endometrial cancer diagnosis” due to its optimization “through ablation studies, self-attention mechanisms, and efficient training,” where it “generalized well across multiple histopathology datasets.”
In other words, the AI model can look at these microscopic scans – histopathology images – and boost image quality to allow the discovery of early stages of cancer, narrowing down to specific areas of the scans to find problematic growth that may not be easily detectable by the naked eye.
Currently, human-led diagnostic methods have an accuracy of around 78.91% to 80.93%, and endometrial cancer, one of the most common forms of reproductive tumors (with over 600,000 Americans having battled it), is treatable if identified in time, giving patients a good five-year outcome.
However, if not caught in time before it spreads outside the uterus, it becomes increasingly difficult to get rid of, therefore making timely diagnosis, such as the one enabled by ECgMLP, critical in saving lives.
And it’s not just endometrial cancer that this innovative AI model can identify. Study co-author and associate professor at ACU, Niusha Shafiabady, said:
“The same methodology can be applied for fast and accurate early detection and diagnosis of other diseases, which ultimately leads to better patient outcomes. (…) We evaluated the model on several histopathology image datasets. It diagnosed colorectal cancer with 98.57% accuracy, breast cancer with 98.2% accuracy, and oral cancer with 97.34% accuracy.”
So… can doctors compete?
According to the researchers, this tool is not intended to replace medical professionals. In fact, its primary purpose is to work alongside cancer specialists so they can accurately, quickly, and affordably identify the disease and then observe the success of the treatment.
As Shafiabady pointed out:
“The core AI model developed through this research can be adopted as the brain of a software system to be used to assist the doctors for decision-making in cancer diagnosis.”
All things considered, the scientists concluded that the AI model, which relies on deep learning algorithms to analyze histopathological images, has managed to reach outstanding performance, higher accuracy, and shorter processing time in classifying endometrial cancer, with major implications for cancer diagnosis and treatment.
ReadNext
