AI spots cancer and viral infections at nanoscale

Today, getting diagnosed with a viral infection or cancer usually requires waiting until you feel sick, show symptoms, or develop a detectable tumor. A team of researchers wants to catch these threats before your body even realizes they are there.

In a new paper published in Nature Machine Intelligence, scientists from Spain's Centre for Genomic Regulation, the University of the Basque Country, and China's Guangdong Provincial People's Hospital introduced AINU (AI of the NUcleus). It is an AI model capable of spotting cancer cells and detecting viral infections inside cells with nanoscale precision.

Facial recognition at the nanoscale

To understand the scale at play, consider that a single strand of human hair is about 100,000 nanometers wide. AINU can detect internal cellular alterations as tiny as 20 nanometers, roughly 5,000 times thinner than a hair.

Human eyes and conventional microscopes cannot see changes that subtle. To feed the AI, the team used a super-resolution imaging method called STORM, which captures high-definition snapshots of the cell nucleus.

AINU is built on a convolutional neural network, the same visual architecture behind smartphone face unlock and self-driving cars. Trained on nanoscale snapshots, the model learns the spatial arrangement of DNA and nuclear enzymes. Cancer cells, for instance, rearrange their nuclear machinery in distinctive patterns that AINU can immediately flag.

Even more remarkably, the AI detected the presence of herpes simplex virus type-1 just one hour after infection. When a virus takes over, it subtly alters how tightly DNA is packed inside the nucleus. AINU caught those shifts long before any traditional symptoms or bulk cellular damage could appear.

Beyond pathogens, the researchers used AINU to identify pluripotent stem cells with extreme accuracy. Verifying high-quality stem cells typically requires slow animal testing, but the AI achieved the same result from stained imaging samples alone, potentially sparing laboratory animals.

The hospital floor is still years away

Despite the breakthrough, you will not find AINU at your local clinic anytime soon.

The biggest bottleneck is the hardware. STORM microscopes are costly, specialized machines mostly confined to advanced biomedical research facilities. Setting them up requires heavy capital investment and technical talent.

Throughput is another issue. STORM currently captures only a handful of cells at a time. In clinical diagnostics, pathologists need to evaluate thousands of cells in a single workflow to reach a reliable verdict. The researchers noted that while optical hardware is advancing rapidly, clinical deployment remains a long-term goal.

Why it matters

We think AINU represents a conceptual shift in computational pathology, moving AI from spotting visible damage to catching initial molecular rewiring.

For doctors and regular patients, this is not an off-the-shelf test you can take today. But for virologists and pharmaceutical researchers, the utility is immediate. Watching viral impact one hour post-entry allows researchers to evaluate antivirals and vaccines in near real time, while stem cell labs can drastically accelerate quality control without relying on animal models.

For the AI industry, this marks a departure from typical medical imaging models. Most clinical AI tools analyze standard CT scans or tissue slides, acting essentially as faster digital assistants for human radiologists. AINU pairs deep learning with super-resolution physics to analyze biological phenomena that no human observer could ever verify manually.

That said, the model's reliance on exotic hardware is a major vulnerability. Software scales infinitely, but expensive, single-digit-cell microscopes do not. Until imaging manufacturers solve the throughput and hardware cost problems, AINU will remain an extraordinary research microscope rather than a routine bedside screen.