Researchers at Duke University in Durham, New York, have developed a new AI model, called AsymMirai, which has proven capable of preventing the onset of breast cancer by reading the differences in breast tissue between one breast and another.
Enter the new Fanpage.it WhatsApp channel
A group of researchers from the Department of Computer Science of Duke University in Durham of New York state have developed a new model of artificial intelligence which could make a difference in the care of breast cancermanaging to prevent one to five years in advance the onset of the tumor starting from the reading of the mammograms.
Medicine e artificial intelligence (IA): as this latest research shows, it is now clear that AI will have a decisive role in planning in the coming years cutting-edge diagnostic tools in various fields of medicine, first and foremost that of cancer prevention.
How AsymMirai works, artificial intelligence to prevent breast cancer
As we read in the just published by Radiological Society of North America (RSNA)the new model is called AsymMirai and it is the updated and most reliable version of Miraian artificial intelligence model on which researchers from the MIT (Massachussets Institute of Technology) I’ve been at work for years. In fact, the latter had already shown that he could prevent cancer cases to the breast using the data provided by mammographybut with important limitations of use: namely the possibility of producing incorrect diagnoses and the fact of function in a way that is still unclear. It is in fact based on a complex neural scheme whose operating mechanisms are still unknown.
Breast cancer, more accurate diagnoses with artificial intelligence
Compared to its predecessor, AsymMirai uses mechanism more linear operation: the researchers who created it have in fact kept Mirai’s deep learning structure but replacing the remaining part – the complex one – with a interpretable form who reads the local bilateral dissimilarityor the structural differences in breast tissue between the left and right breasts.
The data used by artificial intelligence for tumor diagnoses
By studying Mirai’s complex reasoning mechanism, the researchers understood that the artificial intelligence also formulated its diagnoses comparing the two breasts. From this data they then built a “simpler” neural network capable of making the comparisons.
To achieve this result, more than 210,000 mammograms Of 81,824 patients carried out between 2013 and 2020. The researchers had it analyzed by both Mirai and AsymMirai and found that their deep learning model it worked almost as well as Mirai, which currently represents the state of the art in the field, and therefore the best tool currently available. Unlike the latter, however, the new system is more easily interpretable. This quality could make it an important tool in the hands of radiologists in the prevention of breast cancer.