Curie Brief
Turn on cookies to sign in
Signing in saves your progress to your Curie account. We can only do that with cookies on — turn them on to continue.
News
A new machine learning system uses dark-field microscopy and light scattering to tell cancerous cells apart from normal ones — even when they look identical under a standard microscope. Researchers from Japan's NAIST achieved ~91% accuracy distinguishing mesothelioma from reactive mesothelial cells. The tool could soon serve as a powerful diagnostic aid for pathologists.
Standard cytology tests rely heavily on a pathologist's trained eye — but some cancer cells are virtually indistinguishable from healthy ones under a conventional microscope. A team of Japanese researchers may have found a way around that limitation, using light itself as a diagnostic tool.
Scientists from the Nara Institute of Science and Technology (NAIST) and Kindai University Faculty of Medicine developed a system that combines dark-field microscopy — which captures how cells scatter light rather than absorb it — with a machine learning pipeline. The approach picks up on nanoscale structural differences in cells (think actin filaments and microtubules) that standard visual inspection simply can't detect. Their findings were published in Scientific Reports on August 3, 2026.
The system was tested on cytology specimens and showed strong promise across multiple cancer types.
By the Numbers:
Why it matters: Misidentifying cancer cells during cytology can delay diagnosis or lead to unnecessary treatment. A spectroscopic AI tool integrated into existing microscopes could meaningfully boost diagnostic accuracy — especially in tricky cases where cells look deceptively normal.
