Advertisement

Find answers, ask questions, and connect with our community around the world.

  • What’ s Wrong With AI?

    Posted by jonhanse_770 on September 10, 2020 at 7:58 am

                Nothing is wrong with AI itself but the marketing and education going on (or not going on as the case might be) in this market is abysmal at best.
     
                On September  3[sup]rd[/sup] AI provider Vis.ai got CMS (Medcare/Medicaid) to approve payment for Viz.ai’s LVO stroke software. This was THE VERY FIRST AI software to receive CMS’s New Technology Add-on Payment. (NTAP), that provides  payment of up to $1,040 per use in patients with suspected strokes. Not taking a single thing away from Viz.ai at all whose excellent press release clearly saw this as a groundbreaking ruling but you would have thought that finally seeing reimbursement of AI by CMS would have been shouted from the rooftops by each and every AI company out there. Reimbursement  has been one of the biggest Achilles heels AI has had to deal with and yet no one covered the story from that angle. Sadly,the release about Viz.ais breakthrough was also buried 5 stories deep, even in AI-focused journals!! Why? Because so many AI companies are rift with VC money that they hire the biggest baddest PR agencies VC money can buy but who really havent a clue what AI is and how to promote it . Simply put they didnt see the importance of this. But hey, theAI company just got $XX millions of dollars invested in them so they chose the PR and ad agency with the slickest pitch, so they must be good, right?  The fact they cant find their a$$ with both hands is moot.Their pitch was great and they have worked in medical before.not medical imaging, mind you, but medical something or other so… Sigh. Some, but not all, of the marketing people in AI companies arent much better either. This is an area that so many AI companies skimp on, investing nearly all the VC money in software development so they can have the very best product no one ever heard of or about because there is no money left for marketing..either that or no one knows how to promote the product properly.or both..
     
                Talking about PR and marketing in AI (and even PACS/EIS) could easily be an hour long Webinar in and of itself. Howard Im mad as hell and not going to take it any more!! Beal from the movie Network has nothing on me either. I cant tell you how much this drives me absolutely &*^%# crazy.
     
                 Ill be touching on this issue and more in my online Web presentation. The Challenge of Integrating Enterprise Imaging with AI. given at 10 a.m. on Wednesday 9/23.  It is FREE to attend  and is part of AuntMinnies two-day long FREE Fall Virtual Conference on AI, Enterprise Imaging and Beyond. Register now at [link=https://auntminnie.vfairs.com/en/#register]https://auntminnie.vfairs.com/en/#register[/link]
     
                Hope to see you there.
     
    Mike cannavo
    PACSMan
    [email protected]
     
     

    jonhanse_770 replied 3 years, 9 months ago 21 Members · 48 Replies
  • 48 Replies
  • mpezeshkirad_710

    Member
    September 10, 2020 at 11:35 pm

    They said the same thing about ATMs to bank tellers, kiosks to cashiers and outsourcing to American workers.  They were all supposed to “help” or ease the burden.
     
    How have those turned out?

    • khodadadi_babak89

      Member
      September 11, 2020 at 3:19 am

      Quote from Takayasu

      They said the same thing about ATMs to bank tellers, kiosks to cashiers and outsourcing to American workers.  They were all supposed to “help” or ease the burden.

      How have those turned out?

      also – RA’s, PA’s, and NPs. 
      That isn’t going well.
      I think there is a principle here – the “burden” is why you are paid. No burden, no pay.
       

      • jonhanse_770

        Member
        September 11, 2020 at 5:45 am

        Somehow the point got missed here. It wasn’t about “Will AI replace rads” but just that the hottest story about AI- finally getting reimbursement for its use- wasn’t addressed.
         
        It is and has always been my premise that AI is not a competitive technology to rads but a complementary one- one that can help a rad provide a better diagnosis and second opinion- and finally getting paid for using AI elminates one of the primary barriers that exists to using the technology.
         
        The comments made just reinforce my statement that more education is needed on AI so that we can remove the paranoia that seems to exist to AI’s mere existance let alone its use.
         
        Hopefully practicing  radiology isn’t a burden. If it is it may be time to consider a new profession….
         
        PACSMan

        • jonhanse_770

          Member
          September 11, 2020 at 7:33 am

                     As an industry AI has shot itself in the foot for way too long. So what is the significance of the industry finally showing reimbursement for studies read by AI? Its a way for hospitals to make money, pure and simple. And while we can talk about improved patient care and all that other gobbledygook, the bottom line still comes down to money.
                     If a hospital can make an additional $X using a stroke algorithm IN ADDITION TO using the radiologists diagnostic interpretation and be paid for it dont you think they are going to use it? Not just yes but hell yes. The same holds true for every other AI algorithm out there. Once payment for using other algorithms is approved you are going to see this market take off.
                    Its nice to think back to 1971 with us all holding hands singing Id like to buy the world a Coke but this isnt 1971 anymore. And whether you like it or not or accept it or not making money is, always has been and always will be the bottom line. All you need to do is look at the most read posts on AMs Discussion Forum and you see they all deal with money so radiologists arent immune to this thought process either.
                    You might say I dont want AI used!!. Guess what? You dont have a say in it. I dont like ATMs taking the place of branch offices but 3 of the 5 local branch offices of my bank close to me have shut down leaving me to deal with ATMs and other technologies. Hell, the branch offices are even closed on Wednesday as well. Do I like it? No. Do I have to deal with it? What choice do I have? Brick and mortar stores being supplanted by on-line purchasing? Like it? No, but I do like the lower prices on-line offers even with shipping costs,.LOL. RAs PAs NPs? You have Orthos and OB-Gyns and others conducting radiology studies already and soon to have ED departments doing and interpreting their own US scans. How is this different?
                    Now what if you pout and decide you just wont use the AI interpretation provided. That leaves the door open to the hospital setting its own ground rules- Use it or else- orMEDNAX sold to RadPartners takes on a new reality. Yes, its very real. And if, G-d forbid, you have to defend yourself in a lawsuit where AI was right (or even wrong) and you didnt look at the AI interpretation because you dont like using AI? Open your checkbook.
                    . Dont like it? Tough. Rick Warren started his book The Purpose Driven Life with four words that define where we are in medicine today-  Its not about you. I dont like it any more than you do but it is what it is. You are no longer in control of your own destiny. Its no different than the day you said I do quickly followed by the words Yes, dear. Thats why I remain singleLOL
                    Want to learn more about how AI works with an EIS (PACS) and what the future holds for rads? Join me on Weds 9/24 at 10 a.m. as I give a presentation on The Challenge of Integrating Enterprise Imaging with AI. I will  discuss this and more. And the Q&A’s are sure to be rocking….The event is FREE. To register: [link=https://auntminnie.vfairs.com/en/#register]https://auntminnie.vfairs.com/en/#register[/link]
           
          PACSMan
           
          We are products of our past, but we don’t have to be prisoners of it. 
              

          • heartmirror_672

            Member
            September 11, 2020 at 11:27 am

            Generally these things turn out to be a zero sum game.  More money for AI means less for radiologists.  The algorithm has less than 90% specificity and sensitivity on a curated data set.  I have never seen a radiologist miss an M1 occlusion.  This sounds like a waste of money…

            • bola09

              Member
              September 24, 2020 at 5:04 pm

              I have seen a radiologist miss an M1 occlusion. Fortunately the radiologist is not a critical link in that chain. The patient was already in transit to the comprehensive stroke center by the time the report was addended.

              • Dr_Cocciolillo

                Member
                September 24, 2020 at 9:39 pm

                A boat ? Ha. Come on. Thats the people out 10 yrs longer than me with great single jobs.

                Lifestyle requirements is a funny word until one realizes how f u some parts of us society are

                • Unknown Member

                  Deleted User
                  September 25, 2020 at 8:24 am

                  Why is there a need to make everyone believe that AI is only a complement to the radiologist?  AI will only initially be a complement, it will eventually replace a lot of routine reading, this is clear as day.   There will be a generation of radiologists who get screwed by AI, won’t be me.  I will be the generation who read alongside AI.  My kids won’t be doing radiology, I am fully aware of corportization/commoditazation of radiology as well as future AI risks.   
                   
                   

                  • julie.young_645

                    Member
                    September 25, 2020 at 9:39 am

                    Here we go again…

                    • afazio.uk_887

                      Member
                      September 25, 2020 at 11:08 am

                      I agree with Striker – if my kids decide to go into medicine I will steer them away from Rads due to AI. While almost all of us in practice have little worry about would you bet your career that AI wont significantly impact Radiology in 20 or 30 years? I wouldnt and there are safer options in medicine. Corp radiology has huge incentives to push AI and hollow out the field over the coming decades. Its just risk vs reward assessment.

                    • enrirad2000

                      Member
                      September 25, 2020 at 6:08 pm

                      AI may not affect a certain segment of radiologists, but it will definitely negatively affect significant proportion of radiologist’s income and leverage. 

                    • julie.young_645

                      Member
                      September 25, 2020 at 6:36 pm

                      I would worry more about the NP’s and such. If you don’t want them taking your job, or AI for that matter, DO SOMETHING! We have to bypass the folks who stand to make a lot of money by foisting these shams on the public. Sadly, most of us are AFRAID to buck people who stand to make a lot of money by foisting these shams on the public. 

                    • enrirad2000

                      Member
                      September 25, 2020 at 9:52 pm

                      I can not believe how fast PP rads are selling out to Corps. There is really no right price which is worth selling our future freedom and dignity of being well educated and trained professionals. Most of us doctors are not even realizing what we are selling, because it will be lost forever.  Sad…

                    • afazio.uk_887

                      Member
                      September 25, 2020 at 10:01 pm

                      AI is unavoidable in Rads as it is actively being targeted and tech progress is not going to stop or reverse. I feel best case scenario is all AI does to Radiology is decrease demand for Rads…. not great, but not worst case scenario either.  There will be a generation of Rads that kinda get squeezed but residency training numbers will eventually adjust and the field will remain viable albeit smaller in size.  The worst case scenario…. well I am sure we all know what that is.  Both of these scenarios are undesirable basically and best to simply avoid the field if one is thinking about med school someday down the road IMO.   Those in the career already (as Dalai states above) should actively get involved the in the AI scene if possible and try to shape it to benefit us – not much to lose by trying. 
                       
                       
                       
                       

                    • al.georgiev_193

                      Member
                      September 26, 2020 at 4:20 am

                      Better tell these kids that:

                      [link=https://www.webmd.com/lung/news/20200922/medical-schools-report-record-application-numbers]https://www.webmd.com/lun…rd-application-numbers[/link]

                      (TLDR: Med school applications are at record levels, up 17% from last year)

                    • afazio.uk_887

                      Member
                      September 26, 2020 at 9:46 am

                      Ah… the idealistic youth… youth is wasted on the young!

                    • Unknown Member

                      Deleted User
                      September 27, 2020 at 4:28 pm

                      One should be really stupid to go into Medicine these days. 

                       

                    • sanad50_506

                      Member
                      September 27, 2020 at 5:15 pm

                      I bet all their Personal statements will read similar that the land wh made them want to be on front lines and its personal cuz they lost a loved one to Covid. Nothing wrong with that but it will all blend in with others.

                      No MCAT ..I would apply if I want really serious but had grades and see what if I got accepted. If you get in then all the time and stress saved from not trying to ace MCAT s.

                      I agree I would t go into medicine if I was In their shoes or if i did definitely not rads even though I love the field. I love learning about new stuff in the field etc which I dont think I could want to pick up a journal or attend a conference on how to treat HTN. I dont have the hands for surgery I knew but I could have done GI.

                    • Unknown Member

                      Deleted User
                      September 27, 2020 at 7:42 pm

                      There is a lot of money being poured into developing AI for radiology by different companies as there is a lot of money being poured into developing AI for fully autonomous cars.  Everyone knows that the payoff for the first one to successfully do it will be tremendous.  That first mover should stay ahead of the game, for cars we know its likely to be Tesla since they started way before everyone else.  Who will it be for imaging?

                    • jonhanse_770

                      Member
                      September 28, 2020 at 8:13 am

                      The more I read these posts the more it seems like all that those responding are only concerned about losing money and control. Neither should happen… but if all you do is stand on the sidelines saying  The sky is falling and not bother to look up to see if it really is than that is indeed your reality.
                       
                      AI is here to stay. There have been billions (with a B) invested in AI in medical imaging alone so to think it is going away is a folly. On the Webinar in  AuntMinnie’s Virtual Fair I commented there are going to be two types of radiologists- those who embrace AI and those choose to not practice it any more.  I stand by that statement.
                       
                      Radiology is used to people encroaching (or impinging) on its turf. It started with cardiologists, then orthos, then ob/gyn and now its NPs and others. Yet every year when I see the radiologists’ salary surveys rad income keeps going up. The same will happen when AI is fully embraced as well,
                       
                      I appreciate the concern for the next generation of rads but they will be fine too. They are smart and will not only survive but  thrive. It is their willingness to adapt to change that makes the difference, I know about this personally as my youngest son is part of that next generation as an R1….
                       
                      I have been asked by a few what it will take to change the radiologists perception of AI.  I have developed a program that can do just that but frankly I am a capitalist and have done more than my fair share for the AI community for free.  If the big AI vendors want to pony up to the bar collectively or venture capitlaists see an ROI on their investment in this lifetime and start to do REAL market education vs PR fluff I am game to help. Sadly instead most continue to spend a small fortune hiring PR and marketing firms that havent a clue what the AI market needs and what the fears of the rads are relative to AI . That is the classic definition of insanity- Doing the same thing and expecting the outcome to be different; it rings very true here,
                       
                      Can one person change the perception of AI? Only if everyone cooperates towards meeting that goal.  It will also take a bit of time having to first reverse nearly everything that was said to date. but it CAN be done.
                       
                      So please stop whining about the inevitable and embrace it. Who knows…it just might make you a better radiologist in the process.
                       
                      Mike Cannavo
                      PACSMan
                      (407) 359-9191 (office)
                       [email protected]
                       

                    • Unknown Member

                      Deleted User
                      September 28, 2020 at 12:32 pm

                      Quote from ThePACSman

                      The more I read these posts the more it seems like all that those responding are only concerned about losing money and control. Neither should happen… but if all you do is stand on the sidelines saying  The sky is falling and not bother to look up to see if it really is that that is indeed your reality.

                      AI is here to stay. There have been billions (with a B) invested in AI in medical imaging alone so to think it is going away is a folly. On the Webinar in  AuntMinnie’s Virtual Fair I commented there are going to be two types of radiologists- those who embrace AI and those choose to not practice it any more.  I stand by that statement.

                      Radiology is used to people encroaching (or impinging)  on its turf. It started with cardiologists, then orthos, then ob/gyn and now its NPs and others. Yet every year when I see the radiologists  salary surveys rads income keeps going up. The same will happen when AI is fully embraced as well,

                      I appreciate the concern for the next generation of rads but they will be fine too. They are smart and will not only survive but  thrive. It is their willingness to adapt to change that makes the difference, I know about this personally as my youngest son is part of that nexzt generation as an R1….

                      I have been asked by a few what it will take to change the radiologists perception of AI.  I have developed a program that can do just that but frankly I am a capitalist and have done more than my fair share for the AI community for free.  If the big AI vendors want to pony up to the bar collectively and  start doing real market education I am game to help. Sadly they continue to spend a small fortune hiring PR and marketing firms that havent a clue what the AI market needs and what the fears of the rads are relative to AI . That is the classic definition of insanity- Doing the same thing and expecting the outcome to be different and rings very true here,

                      Can one person change the perception of AI? Only if everyone cooperates towards meeting that goal.  It will also  take a bit of time having to first reverse nearly everything that was said to date. butcan be done.

                      So please stop whining about the inevitable and embrace it. Who knows…it just might make you a better radiologist in the process.

                      Mike Cannavo
                      PACSMan
                      (407) 359-9191 (office)
                      [[email protected]][email protected][/email]

                      good advice; thanks.

                    • Robbro524_990

                      Member
                      September 28, 2020 at 6:59 pm

                      Yea, it’s true.

                      Change is inevitable.

                      Either embrace it, or get out of the way of it before it runs you over…

                    • bunnie_face_936

                      Member
                      September 28, 2020 at 7:54 pm

                      Quote from ThePACSman

                      The more I read these posts the more it seems like all that those responding are only concerned about losing money and control. Neither should happen… but if all you do is stand on the sidelines saying  The sky is falling and not bother to look up to see if it really is than that is indeed your reality.

                      AI is here to stay. There have been billions (with a B) invested in AI in medical imaging alone so to think it is going away is a folly. On the Webinar in  AuntMinnie’s Virtual Fair I commented there are going to be two types of radiologists- those who embrace AI and those choose to not practice it any more.  I stand by that statement.

                      Radiology is used to people encroaching (or impinging) on its turf. It started with cardiologists, then orthos, then ob/gyn and now its NPs and others. Yet every year when I see the radiologists’ salary surveys rad income keeps going up. The same will happen when AI is fully embraced as well,

                      I appreciate the concern for the next generation of rads but they will be fine too. They are smart and will not only survive but  thrive. It is their willingness to adapt to change that makes the difference, I know about this personally as my youngest son is part of that next generation as an R1….

                      I have been asked by a few what it will take to change the radiologists perception of AI.  I have developed a program that can do just that but frankly I am a capitalist and have done more than my fair share for the AI community for free.  If the big AI vendors want to pony up to the bar collectively or venture capitlaists see an ROI on their investment in this lifetime and start to do REAL market education vs PR fluff I am game to help. Sadly instead most continue to spend a small fortune hiring PR and marketing firms that havent a clue what the AI market needs and what the fears of the rads are relative to AI . That is the classic definition of insanity- Doing the same thing and expecting the outcome to be different; it rings very true here,

                      Can one person change the perception of AI? Only if everyone cooperates towards meeting that goal.  It will also take a bit of time having to first reverse nearly everything that was said to date. but it CAN be done.

                      So please stop whining about the inevitable and embrace it. Who knows…it just might make you a better radiologist in the process.

                      Mike Cannavo
                      PACSMan
                      (407) 359-9191 (office)
                      [[email protected]][email protected][/email]

                       
                      Is there anyway those of us who missed the presentation can have access? I was interested in going but my schedule did not allow!

                    • cytek1

                      Member
                      September 28, 2020 at 8:28 pm

                      I also wouldn’t mind a recording if available. I have seen very few AI algorithms that impressed me, maybe you could show something unique and helpful to the radiologist, rather than something that seems like a lame private equity attempt to displace them. The tiny subdural detector is a handy helper. Automatic volume calculators are helpful. Accurate lesion tracking is helpful. Something that could detect tiny potential GI bleeds on a multiphase would be helpful. An AI that can call pneumonia on a bunch of inpatient films, or “accurately diagnose appendicitis like a radiologist! or detect fractures on plain film, are not helpful. An LVO detector algorithm is not helpful, and its laughable that CMS is going to pay anyone 1000 bucks for it. No radiologists on that committee I guess. 

                    • jonhanse_770

                      Member
                      September 29, 2020 at 2:33 am

                      To view the AM Virtual Fair webinar go to the AM home page, click the upper right hand corner that shows the Webinar (6 photos of the presenters and reads “View on demand now:). VClick on view on demand now and register. This will lead you to the exhibit hall. Click on the entrance to the exhibition hall which will let you inside. Once inside  click on “auditorium”. A list of all the prior presentatons will be shown there. Mine is on day two (you need to click on day two), and is the 1st presentation. There are some other really great presenations as well as well as those from some of the sponsors and now that they are on demand you can watch them at your convenience.
                       
                      ope that helps. If you have any questions after viewing my presentation feel free to reach out to me at (407) 359-0191 or via email at [email protected].
                       
                      Mike Cannavo
                      PACSMan. 

  • sehyj1

    Member
    September 11, 2020 at 11:31 am

    Nothing. We talkin’ about practice

    • jonhanse_770

      Member
      September 11, 2020 at 1:22 pm

      How is making up to $1,040 per case from using an algorithm a “zero sum gain” and a “waste of money”? Once again it shows that those who fear AI just dont understand how this all works. Nowhere does it also say a rad will get less for doing the interpretation either- just that either the hospital or rad will get paid additional for using AI….”The sky is falling the sky is falling!!!” Sell sky…
       
      You may have never seen a rad miss a stroke diagnosis, gasmasspsoas, but someone apparently has. Google “Radiologist Misses Stroke” and up pops this: “Radiologist Misdiagnosis Causes Stroke and $11M Verdict” ([link=https://nationaltriallaw.com/radiologist-misdiagnosis/).]https://nationaltriallaw….logist-misdiagnosis/).[/link] What is your malpractice limit? Or try this “Repercussions of a Missed Stroke Diagnosis- Lawsuits to Patient Paralysis” ([link=https://www.expertinstitute.com/resources/insights/expert-perspective-standard-of-care-in-stroke-cases/). ]https://www.expertinstitu…-stroke-cases/). [/link]
       
      Believe it or not there is life outside radiology too. It’s not the rads you have to worry about- what about the ER docs? There is a reason why strokes are all stat cases requiring a 15 minute turnaround time and you why well too.  “Research has found that every one minute a patients treatment is delayed leads to an additional one week of disability…..Stroke is the No. 1 cause of adult disability in the United States and the No. 5 cause of death,” ([link=https://www.radiologybusiness.com/sponsored/22221/topics/artificial-intelligence/vizai-artificial-intelligence-stroke-software)]https://www.radiologybusi…gence-stroke-software)[/link] (Disclaimer- While this is from Radiology Business magazine it reads like a promo piece..but still contains good data). 
       
      Last time I checked 90% beats 0% every day, especially for ED docs. And if you really want to get technical here are the #’s about Viz.ai’s LVO software (Sensitivity and specificity were 87.8% (95% CI: 81.2% – 92.5% and 89.6% 95% CI: 83.7% – 93.9% respectively *(from their FDA filing) .
       
      I have zero connection to Viz.ai but like to think I take a balanced approach to AI, pro and con, something you hopefully you will see in the presentation in a few weeks. All I ask is that you try and be objective about AI please- thats all.
       
      PACSMan
       

      • heartmirror_672

        Member
        September 11, 2020 at 1:50 pm

        I am open to any AI application that
        1. Makes my job easier
        2. Makes me a better radiologist
        3. Does not unfairly reduce my compensation
        I am a neuroradiologist.  An application that misses 10-15% of LVOs (probably more in real world) and overcalls 10-15% is going to make my job more difficult.  “Wait wait, don’t give the TPA, the patient doesn’t actually have an LVO…”  It may make me a better radiologist but I sincerely doubt it, because I would never trust it over my own abilities.  That extra money is not going to me BTW, it is going to the hospital. Once CMS gets wind of AI getting paid for interpretation, it will definitely reduce the radiologists fee.  Honestly, I would love an AI system that hangs the sequences the way I like, labels spines, and autocorrects words from my crappy dictation system.  I would PAY for that…

        • bunnie_face_936

          Member
          September 11, 2020 at 2:27 pm

          I feel like this will not affect overall compensation and is more to get the technology implemented at hospitals that need to up their times. I’m not sure how the CMS accounting works but for this, it sounds like a pretty separate “fund” that was used for heartflow and CAD long ago (since been removed for both).

        • jonhanse_770

          Member
          September 11, 2020 at 2:32 pm

          No one has unequivically said who gets the added money from the Ai study but its a pretty sure bet you won’t get it. I guess it depends who paid for AI to begin with…I need read up more about it….This is all new so…And I agree once CMS gets wind of how much you are getting paid for a study with AI used as well someone will be impacted financially, most probably the rad. “Do more,  get less”- isn’t that the new radiology mantra?
           
          There are PACS/EIS that can hang the sequences like you like them (using AI no less), do spine labelling, and autocorrect reports. The good thing is the hospital typically owns the PACS so THEY have to pay for that- you just need find a valid reason to get rid of the system you have. Sadly what may be valid to you I and 40K other rads (I am not a rad BTW) it might not be valid to the bean counter. And you know the golden rule always wins- them that has the gold, rules.
           
          Make a great weekend!!
           
          PACSMan

          • heartmirror_672

            Member
            September 12, 2020 at 10:26 am

            Thanks for the comments.  What is the name of the PACS system that does these things”  The hospital system I work for is using a system with early 2000s technology, and is sorely in need of an upgrade.  Unfortunately, the pandemic will likely delay this at least a few more years. 🙁

            • jonhanse_770

              Member
              September 14, 2020 at 4:29 am

              In a few years if/when you get the money to do a full upgrade give me a call and I’ll see if any other vendors have adopted the features you are looking for and if the vendors that have them now still have them. Often when software gets upgraded somethiing else you really like gets taken away so…
               
              Money is a commodity in very short supply these days with so many PACS needing to last years beyond their typical useful life. t the very least many need OS upgrades so they have no choice but to get off their existing platform.
               
              Vendors are getting creative with ways to spend operating and not capital budget so you might talk to your Administrative VP and see if that is something your facility might consider. Just keep in mind that like secnd marriages where 3 out of 4 fail,  using the same PACS selxction process you used 7+ years ago (RFP’s, etc) won’t work now. The process is totally different and systems completely different as well. Selecting an AI system is a bit easier but there are also a lot of questions you need to ask before you buy, That is where people like myself come in- poviding you with informed objective information so you can make a sound decision. (This has been a paid political announcment. I’m the PACSMan and I approved this message…LOL) 
               
              Mike Cannavo
              PACSMan
               
               
               
               
               

              • drmakyuz

                Member
                September 23, 2020 at 6:03 am

                Question:
                Strip away the technical implementation and Gartner hype cycle and AI for diagnosis seems just like Mammo CAD.  Eventually everyone purchased mammo CAD but it just became an add-on step rather than displacing anything.  What make the current round of techno-craze any different
                 

                • ruszja

                  Member
                  September 23, 2020 at 6:09 am

                  I want an AI system that takes calls from noctors and explains to them that the 20mm cyst at the upper pole last week is the same as the 23mm cyst described 4 years ago.

                  • cytek1

                    Member
                    September 23, 2020 at 6:53 am

                    Its erroneous to act like the time delay caused by the radiology interpretation is a big factor in the overall timeline. The delay in reperfusion is overwhelmingly due to transfer time from a satellite hospital, the ER understanding how to deal with a stroke, and the transport service actually moving the patient where they need to be.

                    If someone really wants to shave a few minutes off of the already short time cost for CTA interpretation, thats an easy process improvement job that probably just consists of marking the study as read now. It doesnt need a $1000 AI solution.

                    This is an expensive solution searching for a problem.

                    You want to apply the same process to a regular ct head? Sure, I could see that saving someone with a head bleed whose scan is buried among 50 others on a worklist. Code stroke cta? Not so much.

                    • cindyanne_522

                      Member
                      September 23, 2020 at 6:58 am

                      didnt AI retire a decade ago? 

                    • ruszja

                      Member
                      September 23, 2020 at 7:01 am

                      Quote from Valerian

                      Its erroneous to act like the time delay caused by the radiology interpretation is a big factor in the overall timeline. The delay in reperfusion is overwhelmingly due to transfer time from a satellite hospital, the ER understanding how to deal with a stroke, and the transport service actually moving the patient where they need to be.

                       
                      Often enough I get a stroke-code read while the nursing and tech staff is still getting the patient off the table yet the attending of record doesn’t even know who the patient is or where they are currently located. ‘Time is brain’ only works if everyone takes this seriously.
                       

                      You want to apply the same process to a regular ct head? Sure, I could see that saving someone with a head bleed whose scan is buried among 50 others on a worklist. Code stroke cta? Not so much.

                       
                      If your CT techs are not of the ‘just making the donuts’ variety, you’ll get a heads-up about any acute bleed before the patient is off the table. So that shaves a solid 10 minutes off the 2 hrs it takes until the neurosurg PA shows up for the consult.

                    • ruszja

                      Member
                      September 23, 2020 at 8:17 am

                      I need an AI to create the following prose automatically, rather than someone having to call up a macro:
                       
                      [i]….suspicious for a mass of the cecum. Recommend correlation with age-appropriate colonoscopy screening when clinically appropriate.[/i]
                       
                       

                    • charnie

                      Member
                      September 23, 2020 at 10:03 am

                      Quote from fw

                      If your CT techs are not of the ‘just making the donuts’ variety, you’ll get a heads-up about any acute bleed before the patient is off the table. So that shaves a solid 10 minutes off the 2 hrs it takes until the neurosurg PA shows up for the consult.

                       
                      True. I assume there are a lot of dummy techs out there though, or  other urgent things like strokes or herniation othey may not notice (although you’d think the primary team would already suspect it and your more rapid read would be irrelevant) . I am in agreement with you, I was just trying to think of a somewhat justifiable alternative compared to what I think is an entirely useless application in a code stroke setting. Someone should design an AI that can replace EMS in  assessing patients and send them straight to the primary stroke center, if they want to really improve reperfusion times. This is chasing single digit percentage points here.

                    • ruszja

                      Member
                      September 23, 2020 at 11:15 am

                      Quote from jimmysworld

                      Someone should design an AI that can replace EMS in  assessing patients and send them straight to the primary stroke center, if they want to really improve reperfusion times. This is chasing single digit percentage points here.

                       
                      Not an AI issue, but EMS standing orders already call for patients with hand involvement to go straight to a comprehensive stroke center.

                • jonhanse_770

                  Member
                  September 23, 2020 at 11:07 am

                  Question:Strip away the technical implementation and Gartner hype cycle and AI for diagnosis seems just like Mammo CAD.  Eventually everyone purchased mammo CAD but it just became an add-on step rather than displacing anything.  What make the current round of techno-craze any different

                   
                  If you sat in on today’s presentation you would know there is light years worth of different between basic CAD and advanced AI (deep learning and Neural networks.) That “technical presentation” is where the rubber meets the road. Comparing one to another is like saying in the dark all things are equal…
                   
                  AI will be an add on as well but instead of just providing additional revenue for a low dollar mamm study and here and there used to  point out something that a rad might have missed AI will actually assist in the diagnostic interpretation process. No one should be also be displaced either, with the interpretation process made better with AI. 
                   
                  PACSMan

                  • jennycullmann

                    Member
                    September 23, 2020 at 6:46 pm

                    Quote from ThePACSman

                    Question:Strip away the technical implementation and Gartner hype cycle and AI for diagnosis seems just like Mammo CAD.  Eventually everyone purchased mammo CAD but it just became an add-on step rather than displacing anything.  What make the current round of techno-craze any different

                    If you sat in on today’s presentation you would know there is light years worth of different between basic CAD and advanced AI (deep learning and Neural networks.) That “technical presentation” is where the rubber meets the road. Comparing one to another is like saying in the dark all things are equal…

                    AI will be an add on as well but instead of just providing additional revenue for a low dollar mamm study and here and there used to  point out something that a rad might have missed AI will actually assist in the diagnostic interpretation process. No one should be also be displaced either, with the interpretation process made better with AI. 

                    PACSMan

                     
                    PACSMan, are you a practicing radiologist?

                    • jonhanse_770

                      Member
                      September 24, 2020 at 6:09 am

                      If you value your life or those of your loved ones you don’t want me reading your X-ray. My report would be something like “There is a gray area over in the corner by the black area next to the mottled gray and black area that looks like clouds before a rainstorm that could be something very significant….or could be bowel gas too. An MRi with contrast is recommended along with some PPG, overseen by the TSA carrying RPG’s.”.
                       
                      The short answer is no but I am one of those applicants who has what HR calls “equivilent experience”, just not expereonce reading X-rays. I have spent the past 30 years working with and alongside rads, techs, hospital admin and others and as such feel fairly confident I know most of their needs and how technology can both help and hinder them. 
                       
                      Just curious- why do you ask?
                      PACSMan
                       
                       

                    • Unknown Member

                      Deleted User
                      September 24, 2020 at 6:23 am

                      AI is “complementary” right now…we are helping develop the technology that will someday completely replace what we currently do.   Once AI gets really good, it will do reads on its own…you may still be alive to tell your grandkids how you used to scroll through the images making findings and dictating reports.  
                       
                      Tesla’s autopilot is complementary right now…At the final stages of AI development in cars, the cars will be completely self driving and we will have robo taxis.  Taxi car drivers will be replaced by robo taxis far before radiologists get replaced by AI.  There will be a day though that radioloigst reads are replaced by AI, don’t be so naive.  This is how capitalism works.  

                    • cytek1

                      Member
                      September 24, 2020 at 6:57 am

                      There is also a day that robots will build all of our buildings for us, pilot our planes and spaceships, and grow all of our food.

                      I remain unconvinced that there will ever be an AI that can cover all pathologies and modalities to the point where it replaces a radiologist, at least not before every other job is also replaced by automation.

                    • jonhanse_770

                      Member
                      September 24, 2020 at 8:18 am

                      Google “History of Self Driving Cars” and read the Wikipedia entry. Fascinating reading and convinces me even more that ist gonna be a while before we see self driving cars as a reality. Its the same with AI vs rads.
                       
                      I know how capitalism works. Now lets talk about life in a litiginous society. Ai misses a call. Who do you sue? The software developer? Vendor who supplied it? Hospital who required it? Others? All of the above? Now prove it that AI actually missed it. The case drags out for years if not a decade.No one wins everyone loses.
                       
                      Rad misses a call. You sue the rad and the hospital because a lawsuit always ends at the deepest pocket. Even though you may not be guilty the insurance company settles because that is the path of least resistance that allows everyone to move on… 
                       
                      AI and the rad together= better diagnosis. Compliments to each other providing better patient care, not AI as a replacement for the rad. Better diagnosis= fewer lawsuits as well. That is one of many goals…and if you can make some $$ off it as well that makes it even better. After all, capitalism reigns,,,
                       
                      PACSMan
                       
                       

                    • Dr_Cocciolillo

                      Member
                      September 24, 2020 at 8:37 am

                      The big issue with the does it drive down demand for radiologists in this hybrid model.
                      Would increase productivity be at expense of decreasing pay per unit volume ? Will this increase productivity result in additional strain and mental fatigue?

                    • jonhanse_770

                      Member
                      September 24, 2020 at 1:37 pm

                      If you use AI as a compliment to improve patient care by providing a second opinion then I don’t see how it will drive down demand since a rad still has to read it and dictate the final report. Now AI might add a small amount of time to the interpretation process reviewing what the AI findings are before dictating but that’s a small price to pay for increased diagnostic accuracy and a while lot more.
                       
                      I have heard for years that there simply are not enough hours in a day to do all the work a rad has but I also never ever have seen a rad turn down the chance to read that extra study being presented to them either…
                       
                      No one has discussed it yet but you may stand the chance that the rad may become overreliant on AI and perhap not read the study as thoroughly as they would without it but that is merely conjecture at this stage in the game. Repeat after me- AI is a compliment to the rad not a replacement.
                       
                      If AI is used as a first pass for screening mammo studies somewhere way down the road then yes it MAY impact volume but we are years away from that happening. If/when that happens then we’ll look at it. Looking at what you get for a screening mamm vs the liability associated with it you might be glad to see it go away. Here in Orlando a screening mamm costs $30 technical and professional combined. You can’t even buy dinner at Popeyes for that…
                       
                      Increasing productivity at the expense of decreasing PPU volume? Talk to CMS and the insurers about that. With or without AI you are going to take it in the shorts just as you have so many times in the past so you have expereince in dealing with it already. 
                       
                      Increased productivity causing strain and mental fatigue? Only if you feel you have to read X studies per day to meet your current lifestyle requirements. Make a slight lifestyle adjustment where you choose the 35′ boat instead of the 40’er or the Lexus instead of the BMW and you should be fine,,,
                       
                      PACSMan