Please create an account to participate in the Slashdot moderation system

 



Forgot your password?
typodupeerror
Google Medicine Software

Google's AR Microscope Quickly Highlights Cancer Cells (uploadvr.com) 42

An anonymous reader quotes a report from UploadVR: Google Research this week revealed an AR microscope (ARM) capable of detecting cancerous cells in real-time with the help of machine learning. Locating cancer with a standard microscope is a difficult and time-consuming process, with a raft of information for doctors to study and investigate. With this new solution, though, the microscope is able to quickly locate cancerous cells and then highlight them as a doctor peers inside. The platform uses a modified light microscope integrated with image analysis and machine learning algorithms into its field of view. An AR display sits above a camera that communicates with the algorithm to display data as soon as it locates an issue. In order words, the microscope immediately begins looking for cancerous cells as soon as you place a sample beneath it. It's effectively doing the same job as a doctor just, according to Google, a lot faster. Google posted a video about the AR microscope on YouTube.
This discussion has been archived. No new comments can be posted.

Google's AR Microscope Quickly Highlights Cancer Cells

Comments Filter:
  • by Anonymous Coward on Sunday April 22, 2018 @04:18PM (#56485053)
    honestly, much of what doctors do can be automated with machine learning. It's a broad subject with shallow depth. Machines are pretty good at taking care of living things, it turns out. This is a welcome solution to the "doctor shortage" in the USA.
    • honestly, much of what doctors do can be automated with machine learning.

      Much of what doctors do can be automated with a sheet of paper. A nurse with a printed checklist or flowchart can diagnose patients about as accurately as a doctor.

      Does the patient have a fever? -> Yes. -> Does the patient have a runny nose? -> Yes. -> Is the nasal mucus clear? -> Yes. -> Is the flu going around? -> Yes. -> The patient has influenza. Run a RIDT to be sure.

      You don't need a doctorate for that.

      • Like the legal profession, medicine suffers from a patriarchal hierarchy from centuries ago. Certainly the professions require training, but the work done by astutely trained individuals working strictly within guidance protocols could stanch the stark worldwide need for physicians, and yes, lawyers.

        However, when they find themselves as replaceable as fast food workers, there will be hell to pay, and lobbying money to be spent, thus perpetuating the problem.

        • by ShanghaiBill ( 739463 ) on Sunday April 22, 2018 @06:19PM (#56485517)

          the stark worldwide need for physicians

          Actually, it is mostly an American problem. In Europe and Asia, nurses and pharmacists are much more empowered, and doctors are considered normal mortals rather than infallible demigods like they are in America.

          I once got sick while attending the Hannover Messe [wikipedia.org] in Germany, and went to a clinic. I walked in the door, was greeted by a nurse, who measured my temperature, took a look at my throat, asked me a few questions, handed me a bottle of medicated syrup, and sent me back to my hotel. Total elapsed time: 5 minutes. Total out-of-pocket cost: $0. That would be unimaginable in America, and is why America's healthcare costs are so high.

          They system in China is similar. Every hospital or clinic has a screening/triage nurse at a desk by the front door. Most routine cases are handled there, and never even make it to the admissions desk. I once took my daughter to a clinic for an impacted tooth (dentistry and medicine are not separated in China like they are in America). Her baby tooth had not come out, and the adult tooth was erupting through the side of her gum. The screening nurse took a quick look, pulled some pliers out of a drawer and yanked out the baby tooth. She then gave my daughter some gauze to bite down on, and gave me a quick lecture about not letting her drink so much milk, which in her opinion caused my daughter's teeth to grow too quickly. The problem was solved. Total cost: $0.

          ... and yes, lawyers.

          This is a mostly American problem as well.

          • by AvitarX ( 172628 )

            You just described my last medical visit in the US.

            We're catching on, quick clinics with nurse practitioners are the first line for sick people here. Doctors are more about general health and check ups, monitoring ongoing conditions etc.

          • I’m fortunate to be really healthy, but I had a similar experience in Japan. In the US we don’t know what we’re missing.

      • by epine ( 68316 )


        for (i = 0; i < n; ++i) {

        Do stop me precisely the moment I need a doctorate.

      • by mikael ( 484 )

        They tried that in the past. AI Researchers sat down with doctors and thought it was a yes/no branch tree all the way down. Then they realized that the doctors were doing something a bit more complex when they started saying things like "well, there's a good chance it might be .... but then the patient has been out in this region where there is this parasite ..... so it could be .... " So they factored in probabilities and got fuzzy logic which was a bit more accurate. But then they also have to factor in

        • Quite a few branches on the flowchart end with "Refer to a doctor", so no one is expecting the nurse to handle everything.

          But for some problems, the nurse is more often correct than the doctor. Nurses are more likely to make a diagnosis that is mundane and ordinary, while a doctor is biased toward looking for an esoteric ailment that he learned about in medical school. As they say "When you hear hoofbeats, think of horses not zebras."

          In the Checklist Manifesto [amazon.com], the author (a medical doctor) describes how

    • "honestly, much of what doctors do can be automated with machine learning"

      Indeed, while in other news, the burger flipping robot was retired after a few shifts.

  • by Anonymous Coward

    One would use GOogle microscope for cancer screenings and then they would send you ads for cancer drugs and cancer centers?

    Ads for hats when you lose your hair to cancer?

    Is facebook gonna add a "like" button to it?

    Hey, I'm just a bit cynical about these online advertising companies creating any of this tech.

    • "Hey, I'm just a bit cynical about these online advertising companies creating any of this tech."

      Those are just byproducts of their face-recognition AIs, they just make some money on the side.

      It not only recognizes unclie Bob, it also sees, that he has Kaposi's sarcoma.

  • by Provocateur ( 133110 ) <shedied&gmail,com> on Sunday April 22, 2018 @07:19PM (#56485743) Homepage

    This is fiction.

    Technician bursts into room, announcing: The tests were successful!
    (Much cheering and applause)
    Patient: Successful how?
    Technician: The prototype can highlight cancer cells!
    (More cheering, but louder this time)
    Patient: Does this mean I have cancer?
    Technician: (still ecstatic) YES! Do you want to see the video?

  • This sounds like it would make a really interesting project, scaled down. Anyone know of a good hobbyist microscope with a movable stage that a USB camera would play nicely with? It would be fun to step the stage over X and Y for a slide of, say, pond water and create a classifier to determine what's in it.

  • This could make diagnosis actually worse.

    Research from John Hopkins (I think?) has shown that performance of human doctor's DROPS when they are provided with automated results FIRST without a chance to view the un-annotated images. If they are provided with automated results in advance, they simply verify those results without doing much more looking themselves - the result being they wind up missing outliers that humans are good at seeing but machines aren't. It blinds them to outliers they would normally

Always leave room to add an explanation if it doesn't work out.

Working...