Deep Nanometry (DNM) is an innovative technique combining high-speed optical detection with AI-driven noise reduction, allowing researchers to find rare nanoparticles like extracellular vesicles (EVs).

Since EVs play a role in disease detection, DNM could revolutionize early cancer diagnosis. Its applications stretch beyond healthcare, promising advances in vaccine research, and environmental science.

A Breakthrough in Nanoparticle Detection

Researchers from the University of Tokyo and beyond have developed Deep Nanometry, a cutting-edge technique that combines advanced optical technology with an AI-driven noise removal algorithm. This approach, powered by unsupervised deep learning, allows for the rapid and highly accurate detection of nanoparticles in medical samples. By identifying even trace amounts of rare particles, Deep Nanometry has demonstrated its potential for detecting extracellular vesicles — tiny biological markers that may signal early signs of colon cancer. Researchers hope this breakthrough will extend to other medical and industrial applications.

Extracellular Vesicles: Tiny Clues to Big Diseases

Your body is filled with microscopic particles even smaller than cells, including extracellular vesicles (EVs). These tiny particles play a crucial role in early disease detection and drug delivery. However, because EVs are so rare, identifying them among millions of other particles has traditionally required costly and time-consuming pre-enrichment processes. To overcome this challenge, Yuichiro Iwamoto, a postdoctoral researcher at the Research Center for Advanced Science and Technology, and his team have developed a faster, more reliable way to detect EVs — bringing us one step closer to more efficient and accessible disease diagnostics.

Optofluidic Apparatus Schematic
Schematic of the optofluidic apparatus. A stream of nanoparticles is tightly focused through an equally tightly focused light beam. The particles emit photons which pass through a filter to remove noise and are then detected using photomultiplier tubes. Credit: ©2025 Iwamoto et al. CC-BY-ND

The Challenge of Detecting Rare Particles

“Conventional measurement techniques often have limited throughput, making it difficult to reliably detect rare particles in a short space of time,” said Iwamoto. “To address this, we developed Deep Nanometry (DNM), a new nanoparticle detection device and an unsupervised deep learning noise-reduction method to boost its sensitivity. This allows for high throughput, making it possible to detect rare particles such as EVs.”

At the heart of DNM is its ability to detect particles as small as 30 nanometers (billionths of a meter) in size, while also being able to detect more than 100,000 particles per second. With conventional high-speed detection tools, strong signals are detected but weak signals may be missed, while DNM is capable of catching them. This might be analogous to searching for a small boat on a turbulent ocean amidst crashing waves — it becomes much easier if the waves would dissipate leaving a calm ocean to scout for the boat. The artificial intelligence (AI) component helps in this regard, by learning the characteristics of, and thus helping filter out, the behavior of the waves.

Future Applications Beyond Medicine

This technology can be expanded to a wide range of clinical diagnoses that rely on particle detection, and it also has potential in fields such as vaccine development and environmental monitoring. Additionally, the AI-based signal denoising could be applied to electrical signals, amongst others.

“The development of DNM has been a very personal journey for me,” said Iwamoto. “It is not only a scientific advancement, but also a tribute to my late mother, who inspired me to research the early detection of cancer. Our dream is to make life-saving diagnostics faster and more accessible to everyone.”

Reference: “High throughput analysis of rare nanoparticles with deep-enhanced sensitivity via unsupervised denoising” by Yuichiro Iwamoto, Benjamin Salmon, Yusuke Yoshioka, Ryosuke Kojima, Alexander Krull and Sadao Ota, 20 February 2025, Nature Communications.
DOI: 10.1038/s41467-025-56812-y

News