6 ms·

This shows clear ignorance of what radiologists do, it's like a fish telling someone how to improve land transportation. For some reason CS people fixate on ra

This shows clear ignorance of what radiologists do, it's like a fish telling someone how to improve land transportation. For some reason CS people fixate on radiology for automation just because it is imaging, but there's a lot of context behind it. There's a reason a radiologist/pathologist is called a doctor's doctor, and no one in the field is worried about automation. For perspective: Training to be a radiologist is 5+1 years, your family doc trains for 3.

908B64B197 5y ago
> For some reason CS people fixate on radiology for automation just because it is imaging Precisely because it's imaging. Training data is abundant, and has the potential to be well labelled. And one of the most active field of AI research is... computer vision. So it's no wonder the low hanging fruit would be medical imaging. > There's a reason a radiologist/pathologist is called a doctor's doctor, and no one in the field is worried about automation. Spoken like Garry Kasparov. > For perspective: Training to be a radiologist is 5+1 years, your family doc trains for 3. What I love is how the immediate knee jerk reaction isn't to explain why the ML approaches won't work but to immediately retreat behind gatekeeping, in this case the tittle and number of years of schooling.
ipaddr 5y ago
They are a specialist. A family doctor would order tests that a radiologist would read. Replacing them with AI would seems like a straightforward process. What behind the scenes context would make it impossible?
- 908B64B197 5y ago
  The years of schooling, of course!
- kspacewalk2 5y ago
  > What behind the scenes context would make it impossible? The MD context. One first needs to train AI to perform the job of a general MD before you can get into stuff like radiology (that is, their real job, not what some novice CS grads, or not so novice AI experts like Hinton imagine it to be - I.e. not segmenting things into funky shapes or running some funky black box magic that spits out "tumor/not tumor" with no context whatsoever, no. Actually diagnosing real people, where a life is on the line, and if you fuck up enough times, your career).
  - 908B64B197 5y ago
    > The MD context. One first needs to train AI to perform the job of a general MD before you can get into stuff like radiology That's true. Waymo had to train their AI to perform the job of a firefighter for them to recognize a fire hydrant and a firetruck. > I.e. not segmenting things into funky shapes or running some funky black box magic that spits out "tumor/not tumor" with no context whatsoever, no. ... But that's the goal right? There's a tumor right there at these coordinates or not.
    - mattkrause 5y ago
      > But that's the goal right? Part of it. Human doctors will not only tell you if there's something funky in the image, but will also interpret it in light of a patient's medical history, symptoms, possible diagnoses, etc. Subtle shading near some structure involved in one of two possible conditions might be very important, but an obvious cyst in an unrelated organ likely means nothing. People are weird close-up!
      - Saptarishi 5y ago
        This is an excellent point. An AI may be able to do a diagnosis to you, but a doc could do diagnosis+post diagnosis, with the foresight of medical history. Theoretically an AI could possibly do this as well, but we're nowhere close.
    - jmhmd 5y ago
      No, identifying the tumor is only step 1, and is the easiest step. Most non-radiologists can identify whether a tumor is present. The harder part (and the true value of radiologist reads) is everything that comes after finding the tumor: what structures are the tumor invading? Is there spread to lymph nodes? Are there secondary findings that might affect the diagnosis or treatment? These questions and their relevance changes for every individual case, and while each question by itself may be approachable with AI, getting a detailed and relevant report without meaningless noise from an AI ensemble is a very very hard problem.
- mcguire 5y ago
  "Replacing them with AI would seems like a straightforward process." And yet....
  - 908B64B197 5y ago
    "A machine will never beat a human at Go, the problem space is just too vast."

908B64B197 5y ago

> For some reason CS people fixate on radiology for automation just because it is imaging Precisely because it's imaging. Training data is abundant, and has the potential to be well labelled. And one of the most active field of AI research is... computer vision. So it's no wonder the low hanging fruit would be medical imaging. > There's a reason a radiologist/pathologist is called a doctor's doctor, and no one in the field is worried about automation. Spoken like Garry Kasparov. > For perspective: Training to be a radiologist is 5+1 years, your family doc trains for 3. What I love is how the immediate knee jerk reaction isn't to explain why the ML approaches won't work but to immediately retreat behind gatekeeping, in this case the tittle and number of years of schooling.

ipaddr 5y ago

They are a specialist. A family doctor would order tests that a radiologist would read. Replacing them with AI would seems like a straightforward process. What behind the scenes context would make it impossible?

908B64B197 5y ago
The years of schooling, of course!
kspacewalk2 5y ago
> What behind the scenes context would make it impossible? The MD context. One first needs to train AI to perform the job of a general MD before you can get into stuff like radiology (that is, their real job, not what some novice CS grads, or not so novice AI experts like Hinton imagine it to be - I.e. not segmenting things into funky shapes or running some funky black box magic that spits out "tumor/not tumor" with no context whatsoever, no. Actually diagnosing real people, where a life is on the line, and if you fuck up enough times, your career).
- 908B64B197 5y ago
  > The MD context. One first needs to train AI to perform the job of a general MD before you can get into stuff like radiology That's true. Waymo had to train their AI to perform the job of a firefighter for them to recognize a fire hydrant and a firetruck. > I.e. not segmenting things into funky shapes or running some funky black box magic that spits out "tumor/not tumor" with no context whatsoever, no. ... But that's the goal right? There's a tumor right there at these coordinates or not.
  - mattkrause 5y ago
    > But that's the goal right? Part of it. Human doctors will not only tell you if there's something funky in the image, but will also interpret it in light of a patient's medical history, symptoms, possible diagnoses, etc. Subtle shading near some structure involved in one of two possible conditions might be very important, but an obvious cyst in an unrelated organ likely means nothing. People are weird close-up!
    - Saptarishi 5y ago
      This is an excellent point. An AI may be able to do a diagnosis to you, but a doc could do diagnosis+post diagnosis, with the foresight of medical history. Theoretically an AI could possibly do this as well, but we're nowhere close.
  - jmhmd 5y ago
    No, identifying the tumor is only step 1, and is the easiest step. Most non-radiologists can identify whether a tumor is present. The harder part (and the true value of radiologist reads) is everything that comes after finding the tumor: what structures are the tumor invading? Is there spread to lymph nodes? Are there secondary findings that might affect the diagnosis or treatment? These questions and their relevance changes for every individual case, and while each question by itself may be approachable with AI, getting a detailed and relevant report without meaningless noise from an AI ensemble is a very very hard problem.
mcguire 5y ago
"Replacing them with AI would seems like a straightforward process." And yet....
- 908B64B197 5y ago
  "A machine will never beat a human at Go, the problem space is just too vast."