anyone using digital pathology for sign out?

[narvanation]

I want to get y'all reviews on digital pathology sign out. what instrument do you use to scan. How feasible is it?
Many thanks

---

Members don't see this ad.

 

[LADoc00]

VERY FEASIBLE. like so feasible I would say there is a 99.98% probability that digi is the future of anatomic path.

 

[Napoleon1801]

It seems feasible, but I think the jury is still out on the amount of value it adds, despite the initial cost. If you could 1) Do remote frozens for distant hospitals and expand your market or in the future 2) Get AI that helped you make a diagnosis easier/quicker, then I could see a potential use case. However, it's still cheaper to push glass and these systems do not replace glass. We've not found a good way to justify the cost or angst of bringing it in yet.

 

[LADoc00]

the value? If the difference is between paying an entire another FTE OR having digital and beaming the cases to beach or the snow, even if it the cost is a 150K set up, your ROI is mere months.

I think digi will KILL demand for new pathologists coming out of training though. Even with this big wave of retirements, if you arent a "made pathologist" by now, you are looking at a drop in new staff pay as a result IMO.

 

[HeyDalaron]

the value? If the difference is between paying an entire another FTE OR having digital and beaming the cases to beach or the snow, even if it the cost is a 150K set up, your ROI is mere months.

I think digi will KILL demand for new pathologists coming out of training though. Even with this big wave of retirements, if you arent a "made pathologist" by now, you are looking at a drop in new staff pay as a result IMO.

Echoing this, programs expanding or maintaining large resident rosters reflects very poor forward planning.

Digital will cause mega consolidation, leading to a significant contraction of the workforce. Lots of these pathologists will be unemployed.

Much of what allows isolated rural groups to persist is the isolation. If you park a scanner there instead, the images can be sent to fewer pathologists elsewhere.

I would love to double my caseload and sign it out from a more comfortable location with a nice view. Digital would give me this option.

 

[LADoc00]

Yah so many pathologists are so poorly utilized now in terms of throughput volume productivity that once digi takes off I dont think the field will need the bottom 20-25% of professionals. Things like locums will simply vanish.

Now do I see telepath corporations being a thing? Jury still out, all the companies Ive spoken with including Teledoc are not currently at all interested.

I think digi is the most significant change in the field in my lifetime. I think it will lead to a 2nd mass force reduction of residency programs similar to what happened in the 80s. May take some time though.

 

[caffeinegirl]

Digital pathology is here to stay. There is significant venture capital funding into several digital pathology companies at the moment - and AI is just around the corner.
For practical aspects of digital pathology, check out the Digital Pathology Association

DPA: Digital Pathology Association

Digital Pathology Association (DPA) - Benefits include access to workshops and seminars, as well as a registration discount for the Pathology Visions conference.

digitalpathologyassociation.org

CAP also has guidelines for validation on their website.

 

[HeyDalaron]

Digital pathology is here to stay. There is significant venture capital funding into several digital pathology companies at the moment - and AI is just around the corner.
For practical aspects of digital pathology, check out the Digital Pathology Association

DPA: Digital Pathology Association

Digital Pathology Association (DPA) - Benefits include access to workshops and seminars, as well as a registration discount for the Pathology Visions conference.

digitalpathologyassociation.org

CAP also has guidelines for validation on their website.

Where can one find the venture capital interest in this?

 

[caffeinegirl]

Where can one find the venture capital interest in this?

Many health care VC funds are investing in AI companies. Check out the investors for the major digital pathology companies - Path AI, Paige, Proscia, PathologyWatch - even Google has a pathology AI branch.

 

[HeyDalaron]

Many health care VC funds are investing in AI companies. Check out the investors for the major digital pathology companies - Path AI, Paige, Proscia, PathologyWatch - even Google has a pathology AI branch.

So the VC focus is on AI and not consolidation of practices?

 

[caffeinegirl]

So the VC focus is on AI and not consolidation of practices?

Correct - the funding is based on practice efficiency due to AI. Consolidation will likely be a secondary consequence due to decrease in workload, and lack of need to be on site for diagnosis.

 

[Allegri]

Meh, I have been doing both digital and traditional for the past year, and what stands out to me is how much slower digital is. Unless the interface improves to the level of a slide on a stage in terms of speed, there is a limit to how much digital pathology can improve efficiency. If you want to double your caseload with digital, you are going to triple or quadruple the amount of time you are working.

 

[WEBB PINKERTON]

You think pathology sucks now, wait until there are national venture capital backed outfits doing remote sign out. Just ask radiology about working for Aris and others.

Lot of obgyns and derms getting bought out by VC in my area. If it isn't a hospital, it is VC buyouts. Most of them have their own laboratories that they are forcing the specimens to.

 

[deleted314957]

I certainly cannot comment on the state of the art but it must be pretty darn good by now. We used it before i retired in 2013 for intra group consults.

 

[LADoc00]

So the VC focus is on AI and not consolidation of practices?

Google is not going to run pathology practices. Tech is interested in SELLING crap, they dont like or want to employ actual humans including patients or pathologists.

AI still is a LONG way off, it will come one day but they have been beating the Path AI drum for now 20+ years with zero real results.

 

[caffeinegirl]

Meh, I have been doing both digital and traditional for the past year, and what stands out to me is how much slower digital is. Unless the interface improves to the level of a slide on a stage in terms of speed, there is a limit to how much digital pathology can improve efficiency. If you want to double your caseload with digital, you are going to triple or quadruple the amount of time you are working.

Digital requires a good tech infrastructure (processing speeds and download speeds). When in an optimized environment - it is much faster than glass slides. Much, much faster. And, in environments that integrate the LIS/patient data - you're golden.
There are two different things we're talking about here - digital pathology infrastructure for pathologist signout (which is up and running now) and AI for assisted diagnoses (similar to Paps in cytology). AI assistance is less than 10 years away - probably 5 years out. And yes, the majority of the AI pathology companies are working on tech/software products, not end-to-end workflow solutions - but there are a few who have their eye on the ball.
As for VC/PE buyouts - that's been happening for a while in all fields of medicine. And they are incentivized to improve efficiency.

 

[Allegri]

hmm, sorry but I'm skeptical that even an optimized environment is much, much faster than the 4 seconds it takes to pick up a slide, pass it under the scope, and pick up another one. What platform are you referring to?

 

[caffeinegirl]

There are many digital viewers on the market - home grown and company based. And yes, in the correct technical conditions, it is faster to scan a slide, zoom in and out and also double check the label, identity and enter the case diagnosis.

Here is a quick overview

10 Open-source Whole-Slide Image Viewers and Analysis Programs; Redefining Digital Pathology

What is Whole-slide image or Virtual Slides? Whole-slide images (Virtual Slides), are a high-resolution image of physical histopathology scanned with special scanners, that produces the images in gigapixels. It allows scientists, researchers, clinical pathologists to use software to zoom for...

medevel.com

Aperio ImageScope - Pathology Slide Viewing Software

Join the thousands who use our freely downloadable ImageScope Slide Viewing Software. Adjust magnification, compare different stains, zoom, annotate and more.

www.leicabiosystems.com

What to Look For in a Digital Pathology Image Viewer | PathologyWatch

Digital pathology image viewer can help dermatology practice with ease of access, quality assurance, and the ability to collaborate.

pathologywatch.com

 

[LADoc00]

Next 5-10 years, which is ALL I care about, after that Im Audi.
Digital pathology--->Happening

AI Pathology--->Not Happening

Reasoning: even if there is huge leap forward in AI, like mega ground breaking leap, there will still not be AI systems in diagnosis because the liability assignment for errors is so important for the legal and therefore the political spheres. Meaning when a slide is purely AI read by like a Google system, the medical-legal liability will be so stratospheric in the few times it is wrong (assuming its near flawless to begin with) that these tech companies wont want to have anything to do with them for a very long time.

 

[WEBB PINKERTON]

What about "liquid biopsies"? What if there are far less slides to screen?

 

[caffeinegirl]

Agree- independent AI pathology is not happening - you still need a human to actually sign off on the report. However, AI can be used to triage and/or give out preliminary reports - this is already happening in radiology and pathology. At the end of the day, you need less pathologist work-hours, less pathologists "on-site", and can triage cases to specialists appropriately. My hope is that in the long future we can standardize diagnoses - there is too much variability out there.
As for liquid biopsies - as far as I know that technology is still greater than 10+ years away (for primary diagnosis), whereas it is helpful for detecting recurrence.

 

[LADoc00]

Agree- independent AI pathology is not happening - you still need a human to actually sign off on the report. However, AI can be used to triage and/or give out preliminary reports - this is already happening in radiology and pathology. At the end of the day, you need less pathologist work-hours, less pathologists "on-site", and can triage cases to specialists appropriately. My hope is that in the long future we can standardize diagnoses - there is too much variability out there.
As for liquid biopsies - as far as I know that technology is still greater than 10+ years away (for primary diagnosis), whereas it is helpful for detecting recurrence.

Just to play Devil's advocate here: why am I buying what is likely crazy expensive AI software to give prelim when I have to read the slide anyway? Meaning this will be cost with no additional payment attached. Medicare is not going to pay double if you buy some software system to read them.
Where digital path leverages the manpower you already have to improve efficiency and make more money by cost cutting, not cost bloating.

Liquid biopsies are Fugazi. I wont even address that crap.

Decision algorithm:

Digital path: SAVES me money by decreasing all locums needs, recoups time I would otherwise spent doing nothing while I relaxed in Tahoe and amplifies expertise inside groups to improve quality.

AI Path: COSTS me money, does nothing to decrease locums needs. doesn't amplify group expertise, isnt synergistic with current standard processes and has already been done with Pap smears with completely miserable results ie- massive costs with no throughput gain or increased revenue attained.

 

[Allegri]

All of those platforms look like they require a mouse. I have a lot of experience with Aperio and Qpath, which we use to do consults from overseas. Maybe it's just me, but pointing and clicking a mouse accurately takes more time than slapping a slide around. To beat the speed of glass, they're going to have to come up with a fingers on a pad arrangement (not a touch screen -- too big) or an eyeball-tracking system.

AI is ready to go, please look at the Ibex company's prostate biopsy tool. This thing basically acts like a really good fellow who previews, writes up the case, and marks where the cancer is. And won't complain about it. Like most of AI in the near future, it is an assistant rather than a replacement.

 

[LADoc00]

All of those platforms look like they require a mouse. I have a lot of experience with Aperio and Qpath, which we use to do consults from overseas. Maybe it's just me, but pointing and clicking a mouse accurately takes more time than slapping a slide around. To beat the speed of glass, they're going to have to come up with a fingers on a pad arrangement (not a touch screen -- too big) or an eyeball-tracking system.

AI is ready to go, please look at the Ibex company's prostate biopsy tool. This thing basically acts like a really good fellow who previews, writes up the case, and marks where the cancer is. And won't complain about it. Like most of AI in the near future, it is an assistant rather than a replacement.

Even as an adjunct assistant, the cost of the AI would have to be phenomenally cheap for me to use it. My brother works in software, it is all a giant grift. Software wants huge sums of money from you on a subscription or per click model.

Yay, I just added 20 bucks in cost to a case medicare pays me 32 bucks for! Yippeee!!!

 

[Allegri]

yeah, it's probably more interesting to very large hospital departments

 

[HeyDalaron]

As far as I can see, true AI signout in pathology is a pipedream, and implementing it now actually costs money and adds little of value.

Whereas on the other hand, consolidating disparate, distant practices under one roof via digital signout could be done tomorrow. There is a huge opportunity here.

 

[LADoc00]

yeah, it's probably more interesting to very large hospital departments

But still the only way it could possibly help large hospitals is that somehow could replace actual pathologists. That's where the offset is. Otherwise you are just paying more money in costs.

Meanwhile digital pathology is directly winning right of the gate: cuts locum costs, could remote work to lower cost of living locations, cuts down on staff turnover in undesirable places, allows superior QC and re-review processes, allows clinicians to join in a remote re-review etc. etc.

But the biggest problem with AI is human psychology and cognitive biases: What makes the news: the 1000 times self-driving software drove someone home or the 1 time self-driving software drove someone under a big rig to their death??

ONE HEADLINE: AI pathology software failure diagnoses otherwise healthy 9-year old with terminal brain cancer leading to unnecessary surgery and treatment for NO ONE to ever trust it again.

 

[WEBB PINKERTON]

When AI hit cytopathology it had missed the party by like 10 years. By that time HPV testing was here. I am still not convinced that history won't be repeating itself.

 

[deleted314957]

Folks, my Wife and I bought ( with in-law help because it was all $5000) a new, first IBM PC in 1981 and COBOL compiler. The computer had a 360 K 5 1/4 inch floppy disc so she could study COBOL ( common business oriented language). The microprocessor was an IBM 8088.
Shortly, we got a dial up modem at 300bps.


Things have gone exponential (except mesothelioma) and I am sure AI will be a big deal sooner than we think.

 

[Allegri]

Even as an adjunct assistant, the cost of the AI would have to be phenomenally cheap for me to use it. My brother works in software, it is all a giant grift. Software wants huge sums of money from you on a subscription or per click model.

Yay, I just added 20 bucks in cost to a case medicare pays me 32 bucks for! Yippeee!!!

It's also easy to imagine CMS looking for where to cut more reimbursement and seeing the automation as a justification to cut even more. We pay you 32 bucks when a computer does most of the work? We're knocking that down to 15$.

 

[KeratinPearls]

Folks, my Wife and I bought ( with in-law help because it was all $5000) a new, first IBM PC in 1981 and COBOL compiler. The computer had a 360 K 5 1/4 inch floppy disc so she could study COBOL ( common business oriented language). The microprocessor was an IBM 8088.
Shortly, we got a dial up modem at 300bps.


Things have gone exponential (except mesothelioma) and I am sure AI will be a big deal sooner than we think.

Get in on those AI, EV (NIO, Tesla), solar, autonomous driving stocks now if you haven’t.

 

[WEBB PINKERTON]

Get in on those AI, EV (NIO, Tesla), solar, autonomous driving stocks now if you haven’t.

Will all those hit in 10 years? I noticed that seems to be the prediction for EVERYTHING. "It's 10 years away".

CMS reimburses more for AI use in cytopathology doesn't it? Different CPT code for imaging vs manual.

 

[KeratinPearls]

Chinese EV stocks have been on a tear for the past two months. They have been hitting. Not 10 years....uh like in the next few years Chinese EV market will continue to grow. Nio is competing with Tesla.

 

[WEBB PINKERTON]

I'd be curious to know what role lithium batteries played in the F1 fire yesterday. That was pretty frightening. I have zero interest in EV right now. Love my gas guzzling Dodge.

 

[Thr0mbus]

I'd be curious to know what role lithium batteries played in the F1 fire yesterday. That was pretty frightening. I have zero interest in EV right now. Love my gas guzzling Dodge.

They can’t hardly even keep laptops from exploding. Can you imagine what a cheap Chinese vehicle will do?

FLEE EV NOW!!! (FEVN)!!!

 

[Guest8]

I'd be curious to know what role lithium batteries played in the F1 fire yesterday. That was pretty frightening. I have zero interest in EV right now. Love my gas guzzling Dodge.

Doesn't really make much sense to hate on lithium batteries for causing fires while driving something that literally uses explosions/fire for locomotion.

 

[WEBB PINKERTON]

I ain't hatin, just ain't interested. Maybe in 10 years when those are the only cars on the road.

I still can't believe that formula one driver survived that crash. Amazing.