Innovation
/
/

How the Moichor AI Works

August 19, 2021

A common experience for companies onboarding and starting to use Moichor involves running a side-by-side study to see how our results stack up against those they are used to getting either from a reference lab or by running them in-house.

Some of the feedback we regularly hear is that the results clients receive from Moichor are similar to the ones they are used to seeing.

In this article, Moichor Director of Pathology Kyle Webb, DVM, DACVP and Chief Technology Officer Thanh Le share how the technology works and what makes Moichor’s results different from those captured via a manual CBC read.

Thanh Le explained that one of the things that makes Moichor’s results unprecedented is the standardization of sample processing, post processing, and imaging. This standardization enables a deeper level of accuracy and reproducibility of lab results.

“Moichor is training the computation tool to be accurate in a way that will be way more repeatable than a human could ever be, starting with looking at every cell on the slide,” Le said.

Dr. Webb explained that when pathologists do manual reviews, there can be differences between two pathologists readings on the same sample because manual review is somewhat subjective. 

They don't have the time to fully evaluate  every single cell on a slide — instead they review a subset of 100 cells and the specific area reviewed within an individual slide can vary by pathologist.

“This computational tool makes blood smear review a lot faster and more in-depth because it’s looking at more than I can,” Dr. Webb said.

“The ability of the tool to image hunDr.eds of cells quickly and the pathologist’s ability to then hone in on what’s abnormal and not miss things that may have previously gone unnoticed if they were only differentiating 100 cells manually — that’s a huge advantage that veterinarians will like to see, “Dr. Webb said.  

Thanh Le explained that precision, when it comes to CBC is about lowering the result variability. “Our level of variability is 10 times less than with manual CBC,” Le said.

“One of the things that we do really well is that we provide aggregate information to the pathologist and clinical veterinarians,” Le said. “That’s the difference between showing a lot of images to a pathologist and aggregating them in a way that makes it easy to interpret,” he said. 

Dr. Webb said this translates into a lot of time-intensive tasks that a pathologist would no longer have to handle. “From a pathologist’s point of view, we no longer have to go hunting for things,” she said.

“Having the images already aggregated  makes it lightning fast for me to look through and identify what’s wrong — Or alternatively, see that there’s nothing wrong,” she said.

The manual process for a white blood cell count involves counting cells and using a mathematical formula to estimate a number. Moichor has automated this process and counts every cell to provide a precise number.

After completing a manual CBC estimate, the pathologist must go through enough fields to categorize 100 cells. 

“For a healthy avian or reptile, this takes at least 15 minutes to do well, and for a sick patient it can take from 30 to 60 minutes,” Dr. Webb said.

Moichor enables this process to be done in one minute for a healthy patient and five minutes for a sick patient.

Dr. Webb was excited by the future opportunities that will be possible with Moichor AI. 

“As the company builds out the data it has aggregated, that will allow the software to create reference intervals for species that don’t have them, and hone the reference intervals for species that do,” Dr. Webb said. 

“A big problem out there is that labs that may have access to samples from a large numbers of healthy birds don’t have the time, money, or energy to create these reference intervals,” she said.

“Moichor will be able to do it really fast because its model and program can do all the imaging and classify the cells so quickly that it’s going to be able aggregate this data and publish new reference intervals.” 

“I’m excited because this is going to help provide better clinicopathologic insights for these species in areas where we’re just kind of flying by the seat of our pants in a way, and have been for decades.”

Innovation
/
/

How the Moichor AI Works

A common experience for companies onboarding and starting to use Moichor involves running a side-by-side study to see how our results stack up against those they are used to getting either from a reference lab or by running them in-house.

August 19, 2021
No items found.
Close icon "X"

SIGN UP TO OUR NEWSLETTER

Monthly updates on science and technology shaping our future.