Blog

For hundreds of years, the clarity and magnification of microscopes were ultimately limited by the physical properties of their optical lenses. Microscope manufacturers have pushed these limits by making increasingly complex and expensive stacks of lens elements. Still, scientists had to decide between high resolution and a small field of view on the one hand, or low resolution and a large field of view on the other.

In 2013, a team of Caltech engineers introduced a microscopy technique called FPM (for Fourier Ptychographic Microscopy). This technology marks the advent of computational microscopy, the use of techniques that combine the feel of conventional microscopes with computer algorithms that process the detected information in new ways to create deeper, sharper images covering larger areas. Since then, FPM has been widely adopted for its ability to acquire high-resolution images of samples while maintaining a large field of view using relatively inexpensive equipment.

Now, the same lab has developed a new method that can outperform FPM in terms of its ability to obtain images without blurring or distortion, even when fewer measurements are taken. The new technique described in an article that appeared in the journal Nature Communicationscould lead to advances in areas such as biomedical imaging, digital pathology and drug screening.

The new method, called APIC (for Angular Ptychographic Imaging with Closed-form method), has all the advantages of FPM without what can be described as its biggest weakness – namely that to arrive at a final image, the FPM algorithm it relies on starting with one or more best guesses and then adjusting bit by bit to arrive at its “optimal” solution, which may not always be true to the original image.

Led by Changhuei Yang, the Thomas G. Myers Professor of Electrical Engineering, Bioengineering, and Medical Engineering, and a researcher at the Heritage Medical Research Institute, the Caltech team realized that it was possible to eliminate this iterative nature of the algorithm.

Instead of relying on trial and error to try to find a solution, APIC solves a linear equation, detailing the aberrations or distortions introduced by the microscope’s optical system. Once the aberrations are known, the system can correct them, generally performing as if it were ideal, and producing clear images covering large fields of view.

“We’re getting to a closed-form, high-resolution complex field solution because we now have a deeper understanding of what the microscope is capturing, what we already know, and what we really need to understand, so no ‘There’s no need for any iteration,'” says Ruizhi Cao, a co-author of the paper, is a former student in Yang’s lab and now a postdoctoral fellow at UC Berkeley. “That way, we can basically ensure that we’re seeing the true final details of a sample.”

As with FPM, the new method measures not only the intensity of the light seen through the microscope, but also an important property of light called “phase,” which is related to the distance the light travels. This property goes unnoticed by the human eye, but it contains information that is very useful in terms of correcting aberrations.

It was in solving this phase information that FPM relied on a trial-and-error method, explains Cheng Shen, co-author of the APIC report, who also completed the work while in Yang’s lab and is now a computer vision algorithm engineer at Apple.

“We’ve proven that our method gives you an analytical solution, and in a much simpler way. It’s faster, more accurate, and uses some deep insights into the optical system,” says Shen.

Besides eliminating the iterative nature of the phase-resolving algorithm, the new technique also allows researchers to collect clear images over a large field of view without repeatedly refocusing the microscope. With FPM, if the height of the sample varies by even a few tens of microns from one section to the next, the person using the microscope will have to refocus for the algorithm to work.

As these computed microscopy techniques often involve stitching together more than 100 lower-resolution images to assemble the larger field of view, this means APIC can make the process much faster and prevent possible introduction of human error in many steps.

“We developed a framework to correct aberrations and also improve resolution,” Cao says. “These two capabilities could be potentially beneficial for a wider range of imaging systems.”

Yang says the APIC development is vital to the broader scope of work his lab is currently working on to optimize image data input for artificial intelligence (AI) applications.

“My lab recently showed that artificial intelligence can outperform expert pathologists in predicting metastatic progression from simple histopathology slides from lung cancer patients,” says Yang. “This predictive ability is highly dependent on obtaining uniformly focused and high-quality microscopic images, something APIC is well suited for.”