By Monique Brouillette
Last year, scientists published a study that analyzed century old viruses preserved in lung samples of patients that died during the Spanish flu. The study showed the first ever genetic evidence that the flu virus evolved to become more deadly during the second wave of the pandemic. For virologists, it was an exciting finding, but for pathologists, it was a reminder of just how little their techniques have changed since then.
Despite big advances like digital imaging processing and artificial intelligence, the field of pathology still relies on methods developed one hundred years ago or more. Pathologists today still slice tissues thin, stain them and embed them in paraffin, just like they did a century ago. The methods are time intensive, limit the speed of diagnosis and even hinder the adoption of new advances in AI-driven diagnostics.
In order for pathology to move into the digital century, advances in slide preparation and microscopy will be essential. Pandemic-inspired rollbacks in regulations coupled with new advances in stain-free microscopy, in vivo imaging and even augmented reality could bring the next generation of pathology tools to reality.
Stirrings of change
Although technological advancements have been available, the field of pathology has been slower to adapt than other fields, like radiology, for example, which has rapidly embraced machine learning.
In 2017, the FDA approved the first whole slide imaging system, which allowed pathologists to scan their tissue samples into the computer. For the first time, diagnoses based on histology could be made digitally with machine learning tools. The advance should have catapulted faster digital adoption, but several hurdles stood in the way. Workflow was one. Scanning a slide into the computer adds extra steps and expense into the process. The slide still needs to be created, after all, and many argue in the amount of time it takes to scan a slide a pathologist could make a diagnosis. Simplified and more efficient tissue processing is needed to bring the field forward.
“I think there is a greater opening right now for adaptation of these technologies,” says Yair Riverson, a computer engineering professor from UCLA and CEO of Pictorlabs, a company founded in 2020 to develop stain-free technology. “One of the key benefits is that they will enable us to share slides across multiple platforms and hospitals and other clinical environments.” Riverson said the pandemic really gave the field a boost as pathologists saw the value in working remotely and getting access to digital slides.
In 2020 Riverson co-founded Pictorlabs, a company focused on virtual histological staining of tissue samples. The company is one of many that are trying to streamline the slide preparation by eliminating the need for staining. Unlike radiological techniques in which images are collected digitally and seamlessly integrated into a digital workflow, whole slide imaging adds time and additional steps to the process. It also adds costs, like scanners and fancy monitors. For radiology, digital images save money on film. No such savings exist in pathology yet.
But entrepreneurs like Riverson hope that they can improve the workflow using technologies based on autofluorescence that take advantage of naturally fluorescent cell constituents like proteins and fatty acids. The technology hastens slide processing times and makes digital analysis more practical. His company is developing machine learning algorithms that allow the imaging of the tissue without the use of dyes and stains. In a paper published in Nature Biotechnology in 2019, their technique was shown to be equivalent to common histology stains like H&E, Jones attain and Masson’s trichrome in multiple organ tissues.
These “virtual stains” can be overlaid on the image like an instagram filter. Histological features pop out just as clearly if they had been stained with traditional dyes, according to Riverson and the technique saves hours, especially with more complicated assays.
“Radiologists have multiple screens with different views like MRI and CT to correlate data and make their decisions. We hope our technology will enable that experience for pathologists—giving them the multiple views of the same tissue without having to cut more slides and order more assays,” says Riverson.
Others who are looking to hasten and streamline the pathology workflow want to go a step further even—eliminating the slide entirely. “Histology is fixed in this idea that one can only image tissue if you paraffin-embed it, slice it thin and stick it on a white light microscope,” says Columbia University professor of biomedical engineering, Elizabeth Hillman. “It’s just not compatible with getting data quickly”
Hillman is developing an approach using light-sheet microscopy, a high-powered type of fluorescent microscopy, that can be used on a patient without having to take a tissue biopsy, fix it in formalin or stain it. In March, Hillman published a paper in Nature Biotechnology detailing the technique’s application in mice and humans. In mice, a hand-held scope was able to image pancreatic tissue in a live mouse undergoing surgery and in human liver biopsied tissue. In humans, the scope was able to identify key diagnostic features like signs of arteriosclerosis and detailed views of the tissue. In addition, the scope was used on a human volunteer to image their tongue. Clear images of the tongue tissue layers were seen up to 200 microns in depth without the need for biopsy or chemical dyes.
Other approaches to get rid of the slide are taking advantage of photoacoustic microscopy, that uses light and ultrasound to generate images, and even ultraviolet light. “The dream is that in the future, you won’t need to cut the tissue out at all,” says Hillman.
Bridging the gap
According to Gabe Siegal, CEO of Augmentiqs, an Israeli-based digital pathology start-up, the digital future of the field may still be another 30 years away. In the meantime, his company has found a solution. Augmentiqs builds augmented reality technology that can be plugged into a traditional light microscope. When the pathologist peers through the objective, she will see a glass slide with digital information layered on top, including artificial intelligence-backed tools to measure tumor size, for example. The technology is meant to bridge the gap between the traditional glass slide and next-generation digital processing tools.
“The large-scale transition to digital pathology where everything will be looked at on a screen instead of a microscope is very far from being mainstream,” he says, “this technology offers the seamless integration of the computer with the microscope.”
The augmented reality device, which is added onto a regular light microscope, also connects the user to the full range of machine learning applications. “It’s like the app store for your cell phone,” he says.
Monique Brouillette, is a freelance journalist who covers science and health.