AI in Radiology at Stanford: Rise of the Machines

[@ramchandracheke]

Thanks Matt for sharing your experience!

[@mattlungrenmd]

My pleasure!

[@richardjones1777]

I automatically detected that this talk was awesome, and I’m not even a robot.

[@Guoenyi]

Very good ppt. Loved the animations!

[@mdrafikbk]

Brilliant presentation!

[@joshkim978]

incredible video. thank you so much for this information!

[@BryanSwanSQLServerBI]

This is so incredible.

[@user-to7dp2uc3o]

hello, Brilliant presentation! I am new to Medical Image Processing. Can you tell me about MRI image processing or basically segmentation and features extraction. How can I study about this in python only

[@dibnasinha7254]

But still u gonna need radiologists...radiologists and AI can help in diagnosis,AI can never do this alone but yes it will be of help ,coz there r many grey areas in medicine....

[@RR-vk2tl]

why you think that AI will not replace radiologists one day? Or this is only your wish?

[@n1303144]

RR like autopilot wont totally wiped out man pilots, there are still things other than reading through studies of imaging radiologists have to do. Like correlate clinical information with the data you get, teaching medical student, researching, walking other medical specialties through imaging study like in pre-operative planing. I don’t think it gonna replace the radiologist but rather augment them to become more effective.

[@mandalorian4620]

@@n1303144 You wont need radiologists for that. Other specialists will teach about the imaging in their respective field and can check themselves whether the diagnosis makes sense. However there will always be interventional radiologists and the field of IR will grow bigger and bigger, with more and more procedures being enabled in the future. Probably neuroradiology will also remain, since it is so complex that I personally would neither trust a black box prediction by a bot, nor the neurologist/neurosurgeon. On the other hand, I am convinced that diagnostic radiology wont even be the first specialty to be replaced, since we are probably 2 decades away from those bots to cover all diseases and do so quickly. Personally I think all subjects which are diagnostic in nature can and will be replaced first, be it by apps or bots that perform quick tests and scans etc. For example internal medicine, diagn. Radiology, pathology, primary care med, etc etc. Surgery will be difficult but I am sure that at one point those robots will be able to perform better and more secure. Finally, I think every job will be replaced by a robot at some point and medicine will not be among the first to be replaced.

[@bruhh4676]

@@mandalorian4620 same about software engineers

[@jehansuleik7153]

What is example of AI in radiology?

[@subashchandrapakhrin3537]

Awesome video, I always thought of Interstitial lung disease classification but with meager data couldn't do the same. Dear sir, I am very interested in Myocardial infarction (MI) detection using CNN can you help me on this matter.

[@mandalorian4620]

No function a computer cant do in radiology? How about TIPSS? Any intervention?

[@hassaanhameed]

last year I go to radiologist for ultrasound i have same idea but my damn idea was not much polished I saw that video damn good explain @Matthew Lungren ✌👌 I want to work on that to make a product

[@jozi_mob]

Hi . Great video . If I wanted to use some of material from the slides , how should I cite it? Or are they copyrighted?

[@mattlungrenmd]

Thanks! Depends on which material I suppose, many of the images are from the web our our papers. Fine to cite the link to this video or specific paper in any case.

[@remasteredretropcgames3312]

Again..... as if humans let alone fairly recent adoptees to this process of scanning and appreciating images, and their artifacts, dont make mistakes.

[@linuxgm4015]

this is amazing. can i contact you for a couple questions regarding AI in imaging?

[@CarolusMagnus_College]

How can I participate?

[@sniperacs]

Look like you fear your success on successful Medical AI.....

[@robinwarren6924]

Huh? What are you talking about?