Artificial intelligence to distinguish cancerous cells from healthy ones

BarcelonaArtificial intelligence at the service of cancer detection. A group of researchers, in the presence of Catalan scientists, have developed a tool that scans cells with very high precision and is able to correctly distinguish between cancer cells and healthy cells, as well as detect the early stages of viral infection within cells.

The results of the research, shared between the Centre for Genomic Regulation (CRG) in Barcelona, ​​the University of the Basque Country (UPV/EHU), the Donostia International Centre for Physics (DIPC) and the Biscay Biophysics Foundation (FBB, based at the Institute of Biophysics), should in the future facilitate the rapid detection of organ and fluid tumors (blood cancers, such as leukemia). The study was published Tuesday in magazine Nature and machine intelligence Therefore, it should allow progress in cancer treatment and detection, through a highly accurate tool.

The precision of this new technique was achieved using a microscopy technique known as STORM, which is much more powerful than conventional microscopes. Through an AI algorithm, it was taught to interpret the captured images with great accuracy. The result is called Artificial Intelligence of Nuclei (AINU), which, to put it graphically, is like a giant smart magnifying glass that can distinguish between tissue that is undergoing changes and tissue that remains healthy.

In a telephone conversation with this newspaper, CRG researcher and one of the authors of the study, Pia Cosma, explained that through artificial intelligence an algorithm was developed that, together with high-resolution images, allows “to recognize certain changes in chromatin inside the cell nucleus,” that is, where the DNA is located. To give us an idea, the images are of such high quality that they were able to detect changes – “rearrangements,” as the specialist puts it – inside cells up to 5,000 times smaller than the width of a hair.

This high level of precision should allow us to deal with the fact that scientists still find that the current mechanism is not accurate enough to capture the beginnings of these abnormalities in cells because they are changes that are not detectable by conventional methods. In addition, as Kozma points out, the new technique has the great advantage of being able to obtain results using only a small sample of material.

Although the research is still in the laboratory stage, the researcher hopes to make progress and that the positive results can be implemented in human medicine. In this way, the cellular changes will provide doctors with valuable information to start treatment when the cancer is still in an early stage. In other words, you can “buy time,” emphasizes Kozma, who points out that this will make it possible to personalize treatments and, therefore, it will be easier for patients to improve their prognosis.

The nanometer resolution of the images also made it easy to detect changes in the cell nucleus just an hour after infection with HSV-1. Doctors usually take some time to detect the infection because they “rely on visible symptoms or larger changes in the body,” Ignacio Arganda Carreras, co-author of the study and associate researcher at the UPV/EHU, told Efe.