Tech news site Calcalist recently spoke with Scopio’s co-founder about how Scopio streamlines the hematologic diagnostic process while rendering blood lab microscopes obsolete. Here are some highlights of the convo.
Where did it all begin?
Around eight years ago there was a lot of hype around new possibilities surrounding deep learning and computer vision and how it could be used to analyze image data. We were excited about the potential of harnessing these capabilities to create a great product and we knew we wanted to focus on the world of health care as we saw this as an important opportunity.
The turning point came when we toured a hematology lab where we saw lab techs using microscopes and counting cells manually.
Manually counting cells is something I strongly believe professionals in the 21st century should no longer have to do. When you manually count cells or anything else for that matter, mistakes can be made. Plus it’s extremely time-consuming and tedious. This is not just an issue in hematology labs in Israel, but also worldwide.
So the next step was to purchase a scanner that could digitize the glass slides, turn them into digital files, and apply the tools of deep learning and computer vision to analyze them. But we soon discovered that there was no platform and no technology that could scan and digitize the slides quickly and in high enough resolution, for us to apply these innovative tools on top of it.
So, you shifted gears?
I think the serendipity here is that solving this problem was very deep within our professional background: optics, computational photography, etc. So, yes, we ended up solving a different problem than what we had set out to originally, just to enable ourselves to solve what we really wanted to.
Sometimes in life in order to solve a problem you have to build the solution yourself.
You mentioned ‘computational photography’, what role does that play?
We looked into the problem and tried to understand why there were no scanners that could digitize these microscopy samples. In hematology, you need the highest resolution in light microscopy possible. For example, if for tissues one would use 20X subject magnification, for blood you need a much higher resolution because you want to look into the cells to see the small details. It happens that 100X is the largest magnification you can reach with visible light, and that’s what you use in hematology labs.
Hematology labs are the most challenging types of labs in terms of the magnification needed to review samples. The problem is that when you want to get very high resolution in microscopy, you need to use a lens with a small field of view. It gives you high resolution, but in a very small portion of the sample.
We discovered that the way to solve the problem was by using a completely different collection method for the data. This is very computation-intensive, but luckily advances in GPU technology and processing units by Nvidia and others enabled us to crunch the data and create the image quickly and in beautiful quality and in a cost-effective way. It also enables us to work at 100X resolution.
So, we have managed to break the trade-off between field of view (area) and resolution. This is technology you simply couldn’t have developed 10 years ago.
What are the main advantages of Scopio’s solution?
It’s important to stress that our product is not just a scanner; we provide a whole workflow solution. Digitalization by itself does not hold intrinsic value. If one is using our platform and looking at the images on a computer screen but still insists on counting cells manually via a microscope with a clicker, we haven’t really done anything for the lab techs at all. Our solution not only scans the image, but it’s comprehensive. It solves a problem for the user in two main ways: First, it enables them to do their work remotely, so the expert and the blood sample need not be physically in the same location – which has major benefits both in the level of expertise, the level of care, not to mention efficiency. It’s what I like to call a ‘democratization of expertise’, it levels the playing field so everyone can get the best experts analyzing samples and thereby receive the best possible care anywhere in the world.
The second advantage is that we are actually using AI and computer vision, just as we set out to do initially, to count the cells and classify them and serve as a decision support system, which provides more consistent results than before.
Will Scopio make use of microscopes in hematology labs obsolete?
Our ultimate goal, through our applications, some of which are still in the pipeline, including a bone marrow application, is to completely replace all the microscopy work in hematology labs with fully digital AI-powered work. That’s the future.
We are not replacing the human in the loop completely, we are simply augmenting their capabilities as experts to do their work much more efficiently (including remotely), consistently, and effectively.
Read the full article here: https://www.calcalistech.com/ctechnews/article/gr6j2q984