If you became concerned about a mole on your back — perhaps it had become painful or looked unsightly — your doctor might decide to remove it and have it evaluated. She’d send it to a pathology lab, where a sample of the tissue would be prepared in the form of a slide. It would then be sent to a pathologist, who would examine the slide to determine whether the tissue had any problematic elements, like cancer. After taking a look, the pathologist might ship it to a specialist at another lab for a second opinion. Each time the slide is moved, it is packed up and shipped to a different address. All the while, you’re waiting days or even weeks to hear whether or not you have cancer.

For decades, there have been limited ways to share medical slides, with information management lagging behind even as medical science advanced. But in the past few years, a new industry of digital pathology has emerged that could finally offer a more efficient and large-scale way of distributing and analyzing these kinds of samples.

Despite being a relatively new field, digital pathology and whole slide imaging were approved by the Food and Drug Administration (FDA) in 2017 following eight years of hearings and trials. Whole slide imaging is the practice of scanning and digitizing slide samples so they can be managed, shared, and interpreted faster and, hopefully, with more accuracy. Once digitized, each high-resolution image can then be accessed by computer, to be sent electronically or analyzed by an algorithm trained to identify cancerous cells — and all much faster than humans.

A handful of well-funded startups are now pushing into this space, including Paige.AI, PathAI and Proscia, which have raised $25 million, $15.2 million, and $12 million in VC money, respectively. Proscia’s founders were just named in the Forbes 30 Under 30 list.

David West, CEO and co-founder of Proscia, says cancer runs in his family; his mom is a cancer survivor. Seeing her go through chemotherapy led him to pursue computational biology — the practice of using biological data to develop algorithms that interpret biological systems—and critically examine the intersection of software and cancer.

West wanted to understand why there was such a big gap between what we’re able to do with computers and what was actually being practiced in the medical field. What he found were numerous challenges to the implementation of digital pathology, including expensive scanners, complex IT onboarding, and terrible software.

“These are solvable problems,” West says. “This is coming at a time in very early days of deep learning at that scale, or what we’ve seen as an A.I. wave with tremendous amounts of data and tremendous amounts of computing power, and these new algorithms that are spawning new possibilities. So we started building this stuff to take it to the next level and eliminate subjectivity from pathology.”

Studies indicate that using algorithms to analyze digital slides delivers diagnoses as accurate as an expert pathologist—and in some instances, the algorithms are more accurate. A 2018 study by Massachusetts General Hospital and MIT found that an automated system they developed diagnosed dense breast tissue (an indicator of breast cancer) in mammograms at the same level as an expert radiologist. Another algorithm developed by Google was found to “help detect metastatic breast cancers with significant accuracy and improve pathologist performance,” according to studies in the Archives of Pathology and Laboratory Medicine and the American Journal of Surgical Pathology.

Pathology is the gold standard in diagnosis for good reason, West says. Pathologists are looking at cell structures and have been trained to recognize specific cancerous patterns, so they get a feel for various histological structures (also known as microanatomy, the branch of biology that studies tissue).

Until now, slides have been stored in floor-to-ceiling cabinets in warehouse-sized labs that hold hundreds of thousands of slides. These vast troves of information remain untapped for research until they are digitized, West says. In contrast, a whole slide imaging scanner usually processes 250 or 400 slides at a time and will generate the digital images as it scans. These are very large files — about a gigabyte per image at 40 times zoom, resulting in about a billion pixels per image and a huge amount of information.

“On glass, it degrades very slowly over the course of 100 years, but usually after 10 or 15 years, we throw the glass out,” West says. “Now that labs are going digital, we have billions of pixels, so it’s an immense amount of information breaking down what healthy tissue looks like, what cancerous tissue looks like, and all the various subtypes of cancer.”

Gathering vast amounts of data is critical for all artificial intelligence systems, because the more examples they can analyze, the more they “learn,” adapt their results, and become more accurate. Proscia, for example, is training its algorithm to recognize the fingerprints of specific subsets of both common and rare tumors in a digital form. The analysis of these massive amounts of digital data can discover things invisible to the naked eye, according to Marilyn Bui, MD, PhD, president of the Digital Pathology Association and a pathologist, lab director, and program director at the Moffitt Cancer Center.

Bui envisions a point where artificial intelligence will be able to run complex analyses on images, identifying seven or eight distinct biomarkers and risk factors in tissue. Researchers are also exploring how deep learning and neural networks could predict mutations in some cancers, including lung cancer, and even give a more accurate prognosis than a human pathologist could, according to a presentation Bui gave in December 2018 at the Fifth Digital Pathology and A.I. Congress in London.

“That’s the holy grail,” Bui says. “That’s the real potential of digital pathology — the image analysis, computational pathology, and artificial intelligence component of it.”

Bui also points out that these kinds of services may become more essential, because the number of pathologists in the United States continues to decrease. A study in the Archives of Pathology and Laboratory Medicine warned that America’s pathologists are retiring faster than their replacements, and that population growth and increasing disease in an aging population will lead to a deficit of pathologists. This, the authors say, will have a negative impact on the ability of laboratory services and health care providers to deliver effective care to their patients.

Digital pathology still has a number of challenges to address. In February 2018, Paige.AI partnered with New York City’s Memorial Sloan Kettering Cancer Center, gaining access to its collection of 25 million patient tissue slides. Subsequent reports from ProPublica and the New York Times highlighted the controversy around issues of corporate governance, potential illegality, and other areas over high-ranking members of Sloan Kettering failing to disclose financial stakes in the startup and a noncompetitive bidding process. Other concerns included the use of patient data in a for-profit venture.

“With images specifically, it has proven to be quite effective and actually more accurate than humans alone,” says Pam Dixon, executive director of the World Privacy Forum, which is focused on conducting in-depth research, analysis, and consumer education in the area of data privacy, with a focus on emerging issues. “I am in favor of proceeding very slowly, on a case-by-case basis, and I think we definitely need to be testing the science, especially in regards to images, which is something that the A.I. algorithm does very, very well.”

So, while the field is still developing, and the dynamics between profit and health care are always going to have a rightly scrutinized relationship, digital pathology offers a new field to build out as tech is applied to health care in more and more ways.

“I believe digital pathology is the cornerstone of digital health and within this ecosystem that we’re moving into,” Bui says. “How we do our job really determines how the other physicians in the health care system will do their job and will shape the field in the coming years.”