AI-powered microscope has no lens yet ‘sees’ nanoparticles for pathologists

Artificial intelligence is driving change into pathology as well as radiology. In Canada, for example, researchers have developed a lens-less microscope that uses algorithms based on mathematical models of light to produce large-scale slide images in 3D.

The device can be built for a few hundred dollars, and its inventors hope it can bring pathology to underserved parts of the world, according to a news release sent by the University of Waterloo in Ontario.

Led by Waterloo engineer and chair of medical imaging research Alexander Wong and PhD candidate Farnoud Kazemzadeh, the researchers say the current generation of their microscope—the first was described last year in Nature Scientific Reports—captures light fields and allows for 3D images that are around 100 times larger than the 2D images captured by more traditional microscopes.

Up to now, the only way to do get such wide-field slide visualization has been to stitch together multiple images from traditional microscopes, a much fussier and costlier proposition.

“Currently, the technology required to operate a pathology lab is quite expensive and is largely restricted to places such as Europe and North America, which can afford them,” says Kazemzadeh, an adjunct professor of systems design engineering. “It would be interesting to see what a more affordable, mobile pathology lab could achieve.”

“We introduced a wide-field lensfree on-chip phase contrast microscopy instrument capable of detecting particles at the nanometer resolution,” Wong and Kazemzadeh wrote in last year’s journal article introducing the device, which the university calls a spectral light-fusion microscope. “The instrument does not require hologram magnification, specialized sample preparation, or the use of synthetic aperture- or lateral shift-based techniques to accomplish detection of nanoparticles.”

They underscored that, despite its use of sophisticated algorithmic processing tools, their proposed instrument is “extremely simple and economical to implement, allowing for democratization and proliferation of such systems at every level of healthcare, industry, education or research.”