Developing PathAssist Derm: How PathAI is Poised to Disrupt Dermatopathology Research

Interview with Director of AI Products, Santhosh Balasubramanian

This week, PathAI introduced the company’s newest tool on the AISight? Image Management System (IMS): PathAssist Derm. This announcement is especially exciting, as it marks PathAI’s expansion into dermatopathlogy research, a new area of focus for the company.?

Driven by factors such as a decrease in the number of pathologists specializing in dermatology and the unique complexity of skin conditions, there is a dire need of AI tools to support dermatopathology research.

We sat down with Santhosh Balasubramanian, PathAI’s Director of AI Products, to learn more about the new PathAssist Derm tool, the challenges and opportunities for dermatopathlogy research in general, and how AI has the potential to benefit this scientific field.

<<Learn more about PathAssist Derm>>

PathAI has created multiple leading products in oncology, inflammatory bowel disease (IBD), and other disease areas. What inspired the company to take the leap into dermatology research?

Santhosh Balasubramanian: Dermatopathology is a rapidly growing field with some unique challenges for pathologists. Because many skin conditions share similar features, dermatopathology samples require a certain level of experience and specialization to interpret and triage in research settings and beyond.?

Unfortunately, the number of pathologists specializing in dermatopathology isn’t growing: some estimates put the number at around 1,100 in the U.S. This means that there aren’t many new people doing dermatopathology research, and so there’s more pressure on existing researchers, especially as the U.S. population ages and rates of skin cancer—and other skin conditions—grow globally.?

Due to all these factors, the pressure is on for researching these conditions, testing hypotheses, and bringing new insights to the community about these diseases. For all these reasons and more, we saw a real need to build tools that can help increase efficiency for dermatopathology researchers.

It sounds like dermatopathology research faces some unique challenges that are different from other areas of science. What are the shortcomings of the field today??

Santhosh Balasubramanian: As mentioned, the rate of skin conditions is on the rise globally. According to a 2017 paper in Current Dermatology Reports, skin diseases take the number four spot for non-fatal conditions around the world. As demand for more treatments and innovation grows, we believe that there will be more pressure on dermatopathology researchers, too.?

However, due to the nature of skin conditions, dermatopathology research requires lots of experts: researchers who can review cases and understand the nuances of what they’re seeing on the slide.?

The reality is that there’s just not enough technology developed for the unique needs of dermatopathology.?

What are the primary features of PathAI’s new PathAssist Derm offering, and how are these features specialized for these challenges and quirks of dermatopathology?

Santhosh Balasubramanian: PathAssist Derm is an AI-Assisted tool designed for dermatopathology research. PathAssist Derm accelerates workflows by automating specimen orientation, prioritization, and measurement.

• Orient: Skin is a multilayered part of the body. Ordinarily, a pathologist has to manually rotate a slide into position to see the epidermis—the top layer of skin—so they can understand the context of the specimen. This might only take a few seconds, but when you're reviewing hundreds of slides reviews per day, that really adds up. PathAI’s new PathAssist Derm tool automatically identifies the proper orientation and orients the slides, so that when a pathologist logs into the system, the slide is ready to review.

• Prioritization: PathAssist Derm identifies the top three likely entities within each specimen, and presents those to the user along with a confidence rating. We also include visual evidence like Fields of Interest and AI Overlays, which help explain those results by highlighting the specific regions of the slide that are most representative of each entity. This can help researchers decide how to prioritize specimens or inform their decision on how to further review that specimen. PathAssist Derm’s prioritization model can currently detect 17 inflammatory and neoplastic entities, including both common entities, such as actinic keratoses or basal cell carcinomas, and more rare entities, such as lichenoid keratoses or melanomas. Collectively, these 17 conditions represent more than 95 percent of standard dermatopathology specimen volumes. ??

• Measure: For specimens with possible malignant entities, PathAssist Derm provides a more detailed assessment of the lesion. This enables the researchers to move faster and avoid fully manual measurement.

PathAssist Derm is currently for research use only, so who can use PathAssist Derm today?

Santhosh Balasubramanian: We’ve designed this “research use” iteration of PathAssist Derm for diagnostic researchers specializing in dermatopathology, we’re still developing and testing the tool to one day realize its full clinical potential. We hope to work with early partners on pilot studies to assess the accuracy and impact of the product, with the goal of informing our plans to pursue FDA clearance.

Where are most dermatopathology researchers on their digital pathology or AI journey?

Santhosh Balasubramanian: As mentioned earlier, skin cancer cases globally are rising, which showcases the need for innovation to improve research workflows to address the growing workload. As a result, we’ve found that many of the researchers we spoke with in developing PathAssist Derm were very early in their AI journey. Even if they’ve seen other applications of AI in other organ types or other specialties, there just aren't that many products optimized or designed for skin pathology researchers. PathAssist Derm is one of the first applications of AI for this use case.

Looking at the broader landscape, what can AI tools do for dermatopathology research as a whole?

Santhosh Balasubramanian: Like many of PathAI’s other research products, there’s so much potential to improve efficiencies and reduce the tedious, manual work researchers have to do. With more time back in their day, they can focus on elements of their research that move the needle in terms of discoveries, insights, and innovation. Even just a few seconds spent orienting a slide, for example, can add up very quickly for researchers working with high volumes of samples.

AI tools can also improve the accuracy and consistency across pathologists as they review samples, take measurements, and assess downstream staining.

PathAI has added a lot of product algorithms to AISight over time. What are the aspirations for the platform as a whole?

Santhosh Balasubramanian: Currently, maybe 10 percent of labs have fully implemented digital pathology. But in the next decade, that number will rise to close to 100 percent, as most labs—both research based and clinical—are either actively implementing or developing their strategy for “going digital.”?

While PathAssist Derm is still for research use only, we hope that products like this—and future PathAssist products for other specialties—can help bolster the business case for why that transition is worth it.

Beyond PathAssist, PathAI continues to invest in a few other categories of products for pathology labs, each with a distinct value proposition. The first is a set of non-diagnostic products that deliver workflow optimization and improve efficiency, such as tools that detect artifacts, prioritize cases, assess specimen sufficiency for downstream testing, and more.

A second category of offering is tools for biomarker measurements, such as our AIM-IHC Breast Panel. For research products like this, the real goal is to increase the precision of histopathology features and provide scores that increase consistency between researchers in terms of how they assess these different biomarkers.

And lastly, we’re excited about the potential for truly transformational prognostic AI tools which are able to provide risk assessments for patients that ultimately inform downstream treatment decisions in ways that may not be possible with manual pathology review alone.

PathAI’s ambition with our AISight? platform is to be the hub that facilitates labs’ digital pathology experience. We want to integrate all possible manner of AI—not only AI that we've developed, but also AI developed by other vendors. Our goal is to make both our AI and others’ seamlessly available to pathologists, with the goal of improving their experiences and efficiency, which ultimately improves patient outcomes.

Learn more about PathAssist Derm and reach out to us for a free demo: [email protected]

PathAssist Derm is for research use only. Not for use in diagnostic procedures.

AISight? is for Research Use Only in the US; AISight? Dx is CE-IVD marked in the EEA, UK, and Switzerland.