+

IS AI-ASSISTED DIGITAL PATHOLOGY THE FUTURE OF DIAGNOSTICS?

AI makes the pathologist more efficient and allows handling of a much larger sample load than what she or he can do today without the help of AI.

Digital pathology—the ability to scan a glass slide containing a physical-biological specimen with an automated digital microscope and view the sample on a computer screen rather than under a microscope—has been around for several decades now. The first algorithm and implementation of a digital slide scanner were published in the 1970s! Early players in this field building digital slide scanners include Leica, Roche, Hamamatsu, Philips, etc. Digital pathology helps streamline the microscopy work in a clinical laboratory, by allowing remote review, and also takes away the stress and strain of looking through a microscope all day long for the pathologists.

While the technology has been around for some time, its widescale adoption has started to pick up only in the recent past. One of the primary reasons for lukewarm adoption is the cost. These are precision machines, and their cost is unaffordable for all but the largest clinical laboratories. Further, the regulatory clearance for such devices has been slow in coming. The first digital slide scanner to receive USFDA clearance was only in 2017. This was a model from Philips. It was followed by a model from Leica in 2019.

AI for digital pathology builds on top of the automated slide scanners. It takes the digital images from the scanners as input, and automates, at least partially, the detection of different types of cells, tissues, normal and abnormal cells and cell formations, etc. This helps the pathologist by quickly drawing her attention to the areas of interest, and aids in quicker and often better diagnosis of diseases. AI makes the pathologist more efficient and allows handling of a much larger sample load than what she or he can do today without the help of AI. Prominent players in the field of digital pathology include Paige AI, Path AI, Proscia and others.

Cancer is on the rise worldwide. Cases are expected to rise by 70% all over the world in the next two decades. This will translate to the diagnosis of 27.5 million new cases annually if the current trend continues. Pathology is a vital tool for diagnosing cancer. Close to 70-80% of treatment decisions depend on the findings of a pathologist. This, in turn, leads to multiplying workload for the pathologists. Yet, there is a severe shortage of pathologists, not only in India but even in the developed world. For example, a third of pathologists in the UK are expected to retire in the next four to five years. In the US, there will be an estimated shortage of 5,700 pathologists by 2030.

Thus, making the diagnosis process more efficient and accurate through technology is the only way to battle both the shortage of pathologists and the rising incidence of diseases. This is where AI-assisted digital pathology has a very important role to play. Remote review through digital slides helps in making pathological diagnoses more accessible—by removing the need for the pathologist to be co-located with the sample. The use of AI to assist the pathologist helps bring down the diagnosis time and has the potential, at least theoretically, to increase the accuracy of diagnosis as well. This potential for the future is what is driving the quicker adoption of AI-assisted digital pathology worldwide in the recent future.

We, at SigTuple, are at the forefront of AI for pathology. Unlike many other slide scanner manufacturers (who build only the hardware) and AI for pathology companies (who build only the AI), we offer an end-to-end solution consisting of both the hardware—the digital slide scanner—and the AI for the analysis of the sample.

Our solution is much more affordable compared to the existing devices, and hence it is witnessing a very quick adoption even in the most cost-sensitive markets like India, where other providers have hardly any presence.

The writer is founder & CEO, SigTuple.

AI for digital pathology builds on top of the automated slide scanners. It takes the digital images from the scanners as input, and automates, at least partially, the detection of different types of cells, tissues, normal and abnormal cells and cell formations, etc. This helps the pathologist by quickly drawing her attention to the areas of interest, and aids in quicker and often better diagnosis of diseases.

Tags: