Spanish researchers create AI that detects cancer and viral infections at the nanoscale

A group of Spanish researchers has developed Artificial intelligence (AI) that can distinguish cancer cells from normal cells, as well as detect the early stages of a viral infection inside cells.. In particular, they belong to the Center for Genomic Regulation (CRG), the University of the Basque Country (UPV/EHU), the Donostia International Physics Center (DIPC) and the Fundación Biofisica Bizkaia (FBB, located at the Institute of Biophysics). These results are published in the journal Nature Machine Intelligence. They open the way to more advanced diagnostic methods and new monitoring strategies.

Tool, AINU (nuclear AI), is responsible for scanning high-resolution images of cells. They are collected through a special microscopy method called STORMwhich creates an image that captures much more detail than conventional microscopes can see. High-resolution images allow structures to be seen with nanometer resolution. “We can recognize cancer cells very early and with greater efficiency. The same thing happens with a cell infected with a virus such as herpes,” the company noted in an interview with GM. ICREA research professor Pia Cosma, co-author of the study and research fellow at CRG Barcelona.

A breakthrough in the fight against pathologies and personalization of treatment

To achieve these results, Cosma argues that They researched it for about three or four years.. In fact, he argues, they are the result of the curiosity inherent in the scientific profession. “We have been using this type of microscopy for over 10 years, it is very powerful, and we have made many discoveries in the lab thanks to it,” he confirms. “We thought it would be interesting to try an artificial intelligence algorithm because no one had trained it on this type of image.” Thanks to this technology, he notes, they have achieved discoveries such as the identification of a new DNA structurechromatin. “We used it to study the three-dimensional structure of chromatin to understand how it is organized in the nucleus, what type of structure it can form, and how that relates to gene function.”

“Someday this information will allow doctors to buy valuable time to monitor diseases and personalize treatments.”

Pia Cosmaco-author of the study and research fellow at CRG Barcelona.

Among some of the advantages of AINU, Cosma highlights: high resolution, a small number of images required and the ability to diagnose a liquid tumor directly from blood“The resolution of these images is powerful enough that our AI can recognize specific patterns and differences with amazing accuracy, including changes in the way DNA is organized inside cellswhich helps detect changes very soon after they occur,” says ICREA research professor. “We believe that one day such information will allow doctors to gain valuable time to manage diseases, personalize treatments and improve patient outcomes“.

“Face recognition” at the molecular level

AINU is a convolutional neural network, a type of artificial intelligence designed specifically to analyze visual data such as images. Examples of such networks include AI tools that allow users to unlock their smartphones with their faces, or those used in self-driving cars to understand and navigate their environments by recognizing objects on the road.

In medicine, they are used to analyze mammograms or CT scans and detect signs of cancer that the human eye might miss. They also help doctors detect abnormalities in MRIs or X-rays, allowing for faster, more accurate diagnoses.

AINU thus detects and analyzes tiny structures inside cells at the molecular level. The researchers trained the model by feeding it nanoscale images of the nuclei of many different cell types in different states. Thus, The model learned to recognize certain patterns in cells by analyzing how nuclear components are distributed and organized in three-dimensional space.

Sorting cells and detecting changes

Cancer cells exhibit characteristic changes in their nuclear structure compared to normal cells, such as changes in the organization of their DNA or in the distribution of enzymes within the nucleus. After training, AINU could analyze new images of cell nuclei and classify them as cancerous or normal based solely on these characteristics.

“After an hour, we can see a slight change in the center of the chromatin thanks to the algorithm”

Pia Cosmaco-author of the study and research fellow at CRG Barcelona.

Nanometer-level imaging resolution allows AI detect changes in the cell nucleus just one hour after infection with herpes simplex virus type 1. “After infection, within an hour, we can see a slight change in the center of the chromatin thanks to the algorithm,” Cosma guarantees. The model was able to detect the presence of the virus by detecting small differences in the density of DNA packing, which occurs when the virus begins to change the structure of the cell nucleus.

Laying the foundation for clinical training

Among some of the important ones restrictions that researchers must overcome before the technology is ready for testing or implementation in clinical settings, some of which involve STORM images. Are They can only be taken using specialized equipment, which is usually only available in biomedical research laboratories.Moreover, both the installation and maintenance of the visualization systems required for AI are challenging. investments significant both in teams How in technical experience.

Another obstacle associated with these images is that Typically they only analyze a few cells at a time.For diagnostic purposes, especially in clinical settings where speed and efficiency are critical, doctors will have to capture many more cells in a single image to be able to detect or track disease.

“In a few years, these types of microscopes could be used in any laboratory or hospital center, and from there the use of this method could be expanded to other areas.”

Pia Cosmaco-author of the study and research fellow at CRG Barcelona.

While not all labs and hospitals have one at the moment, Cosma says companies are interested in acquiring others that are easier to use. “In a few years, microscopes of this type will be available in any lab or hospital center, and from there, the use of this technology can be expanded to other areas,” he says. “Sample preparation is very simple, imaging is more difficult, but technology is advancing very quickly,” he says.

“There are a lot of rapid advances in the STORM visualization space, which means that Microscopes may soon appear in smaller or less specialized laboratories. and eventually even in a clinical setting. The availability and performance limitations are more manageable problems than we previously thought, and we hope to conduct preclinical experiments soon,” says Cosma.

High accuracy of stem cell identification

Although clinical benefits may take years to emerge, AINU is expected to accelerate scientific research in the short term.. The researchers found that technology can identify stem cells with very high accuracywhich is especially relevant given that they are being studied for their ability to repair or replace damaged tissue.

Thus, AINU is able to speed up and make the process of detecting pluripotent cells more accurate, which will help make stem cell therapy safer and more effective“Current methods for detecting high-quality stem cells rely on animal testing,” he says. Davide Carnevali, first author of the study and CRG research fellow. However, “all that is needed for our AI model to work is a sample stained with specific markers that highlight key features of the nucleus,” he says. In addition to being simpler and faster, he points out that it also “could speed up stem cell research while helping to reduce the use of animals in science.”

Explore its use in other viruses.

As for what he thinks will affect both tumors and this type of viral infection, Cosma says they hope that in the future, since they haven’t evaluated it yet, they will be able to detect both situations very early because it has a very high sensitivity. In the second case, he says they only tested the herpes virus, “we haven’t tested the other one, but there is already proof of concept.” The reason for this confirms that the next steps are in to confirm whether it can be applied to other types of viruses to block possible viral infection..

Delving deeper into whether this is intended extrapolate these results to other pathologiesThe researcher says they are focusing on liquid-type cancers “at the moment.” “Anything that can be identified by small changes in chromatin, such as drug treatment“You can try to identify yourself using this system,” he confirms.

Identifying the effect of drug treatment through chromatin among problems

Exploring the future of AI in early diagnosis of pathologies and AINU, the CRG researcher argues that the latter has the advantage of going further than current technologies as they can observe extremely small things and train an algorithm on a very small cell.

---

You may also be interested in…