Whole slide imaging

[AZpath]

Is anyone using whole slide imaging? My microscope is tough to beat.

I see some advantages to digital but won't be going there soon.

Thoughts and experiences anyone?

---

Members don't see this ad.

 

[yaah]

It's great for individual cases, taking images, presenting cases at tumor boards, limited thus far in volume. Still takes quite a bit of time to scan one slide (a few minutes, although some technologies are better or more real time) and takes a lot of gigabytes. Will probably be a while before it replaces microscopes (maybe never). Will be most useful likely first in consults and frozen sections. Would suspect there would be some academic places that would develop a way to have you scan in your case for a consult. For frozens it seems close to equivalent to looking at slides.

Problems are reliability (has to be 100% reliable if used for offsite frozens) and time to scan and view. It takes very little time to look at a slide on a microscope. But each slide takes minutes to scan and pull up on your screen and then scan around. Multiply X whatever for a busy slide day. Plus, currently you have to make the slide first anyway, why not just use the scope?

 

[Johnny Sunshine]

It will be very interesting to see how it develops, especially if auto-scanning algorithms (like those used in cyto for paps for over a decade now) become more prevalent. The slides may still need to be made, but they can then be popped into a large loading machine that can organize them by label bar codes and work on scanning them 24/7 while the techs are off at lunch or sleeping at home. The machines can then flag suspicious, complicated, stained, or immuno-positive slides for review by a human, plus a certain percentage of those that the machine deems normal for quality control sake.

The scanning algorithms won't have to be amazing to have an impact. They will mostly exist to swipe away a large chunk of the normal/negative/routine images so a human doesn't have to waste time with them, or at least will have a pseudo second-opinion to consider. While telepathology for frozen sections may need to be 100% reliable, other digital slide applications need not be. They only need to be as good as humans when comparative studies check their agreement rates, which the pap algorithms have been for many years now.

It will be a long time before no pathologists are needed, but even a minor advancement in this sort of technology could dramatically affect the number of pathologists society needs. Fun times!

 

[WEBB PINKERTON]

It will be very interesting to see how it develops, especially if auto-scanning algorithms (like those used in cyto for paps for over a decade now) become more prevalent. The slides may still need to be made, but they can then be popped into a large loading machine that can organize them by label bar codes and work on scanning them 24/7 while the techs are off at lunch or sleeping at home. The machines can then flag suspicious, complicated, stained, or immuno-positive slides for review by a human, plus a certain percentage of those that the machine deems normal for quality control sake.

The scanning algorithms won't have to be amazing to have an impact. They will mostly exist to swipe away a large chunk of the normal/negative/routine images so a human doesn't have to waste time with them, or at least will have a pseudo second-opinion to consider. While telepathology for frozen sections may need to be 100% reliable, other digital slide applications need not be. They only need to be as good as humans when comparative studies check their agreement rates, which the pap algorithms have been for many years now.

It will be a long time before no pathologists are needed, but even a minor advancement in this sort of technology could dramatically affect the number of pathologists society needs. Fun times!

Not much sunshine in that post

 

[deleted314957]

Not much sunshine in that post

I don't know if this is applicable. I've been retired for about three years but I used to frequently e mail a JPG Image of an obviously invasive malignancy to a colleague to the get the obligatory confirmatory "consultation". And I mean obvious.

 

[icpshootyz]

I don't know if this is applicable. I've been retired for about three years but I used to frequently e mail a JPG Image of an obviously invasive malignancy to a colleague to the get the obligatory confirmatory "consultation". And I mean obvious.

Is it kosher to actually put someone's name on the report using just a jpeg as a consult? We don't even use our telepath system for individual consults, only for consensus conference where we can say the case was shown at a conference. To put another pathologist's name on the case we always give them the actual glass slides. I thought it was a CAP rule or something.

 

[WEBB PINKERTON]

I don't know if this is applicable. I've been retired for about three years but I used to frequently e mail a JPG Image of an obviously invasive malignancy to a colleague to the get the obligatory confirmatory "consultation". And I mean obvious.

What did Quest think of this?

 

[path_spouse]

I'm not a pathologist; my husband is--I am a scientist interested in all aspects of computer vision. There are many subtle issues with WSI, including how accurately color, etc. can be reproduced. To some extent the science surrounding how to validate WSI is still evolving. In addition to CAP, which influences what all of you do, the FDA has regulatory authority over how WSI devices can be marketed. I think (I am not certain) that FDA has not actually cleared a WSI device yet for US clinical use. Please correct me if I am wrong.

 

[BU Pathology]

The FDA has not cleared microscopes for diagnostic work, so there may be a delay before they get whole slide imaging.

 

[jupiterianvibe]

The FDA has not cleared microscopes for diagnostic work, so there may be a delay before they get whole slide imaging.

Absolutely true, and quite shocking.

I wonder how the imagers managed to have digital imaging classified as a class 2 device? The principles are similar.

 

[yaah]

It will be very interesting to see how it develops, especially if auto-scanning algorithms (like those used in cyto for paps for over a decade now) become more prevalent. The slides may still need to be made, but they can then be popped into a large loading machine that can organize them by label bar codes and work on scanning them 24/7 while the techs are off at lunch or sleeping at home. The machines can then flag suspicious, complicated, stained, or immuno-positive slides for review by a human, plus a certain percentage of those that the machine deems normal for quality control sake.

The scanning algorithms won't have to be amazing to have an impact. They will mostly exist to swipe away a large chunk of the normal/negative/routine images so a human doesn't have to waste time with them, or at least will have a pseudo second-opinion to consider. While telepathology for frozen sections may need to be 100% reliable, other digital slide applications need not be. They only need to be as good as humans when comparative studies check their agreement rates, which the pap algorithms have been for many years now.

It will be a long time before no pathologists are needed, but even a minor advancement in this sort of technology could dramatically affect the number of pathologists society needs. Fun times!

Eric Glassy at the recent USCAP meeting gave a provocative talk on the likely impact of technology in pathology practice. I do not recall much during his talk that implied doom for pathology or the replacement of pathologists by machines. He described it more as augmenting practice and adding efficiencies. This would imply (to me) that fewer pathologists may be needed, but not that they would be replaced. He talked about scenarios where the computer could help with a complex cancer case by pulling all the pertinent info out of the EMR and displaying a summarized history, operative procedure, and then go through the slides, count the nodes, measure the depth of invasion and such. Then the pathologist would confirm and refine because the computer would invariably make some mistakes.

 

[AZpath]

I think we need a killer app to pay for these systems.