{"id":3670,"date":"2015-10-07T12:01:42","date_gmt":"2015-10-07T12:01:42","guid":{"rendered":"http:\/\/ncjolt.org\/?p=3670"},"modified":"2020-10-15T14:29:10","modified_gmt":"2020-10-15T14:29:10","slug":"dr-watson-how-ibm-is-teaching-watson-to-read-your-medical-scans","status":"publish","type":"post","link":"https:\/\/journals.law.unc.edu\/ncjolt\/blogs\/dr-watson-how-ibm-is-teaching-watson-to-read-your-medical-scans\/","title":{"rendered":"Dr. Watson?: How IBM is teaching Watson to Read your Medical Scans"},"content":{"rendered":"

Remember Watson, the super computer that beat the \u201cunbeatable\u201d Ken Jennings<\/a> at Jeopardy? Well, now Watson may be more to you than a computer with limitless trivia knowledge. He may be your Radiologist, or at least a tool to push some of the reforms of the Affordable Care Act, namely to improve quality care and reduce cost.
\nIn August, IBM announced its plans to acquire Merge<\/a>, a leading provider of medical image handling and processing, in an attempt to teach super computer Watson to read medical images. With Merge in its arsenal, IBM will use Merge\u2019s medical imaging platform and the company\u2019s billions of medical images to give Watson experience reading and comparing various medical scans.<\/p>\n

Ultimately IMB aims to teach Watson to read and interpret scans like a radiologist, but better.<\/p><\/blockquote>\n

 
\nAccording to IBM, Watson\u2019s ability to read medical images will fill a need in the medical community. Medical images are the largest and fastest-growing data source in the health care industry, and although these images are helpful in diagnosing and treating diseases there are various challenges to their use:<\/p>\n

    \n
  1. the sheer quantity of scans that Radiologists must read each day is overwhelming;<\/li>\n
  2. few tools exist to assist Radiologists in reading these scans so reading must be done manually;<\/li>\n
  3. there are no data sets for scanned medical images where researchers can find trends or patterns across scans.<\/li>\n<\/ol>\n

    \u201cIBM plans to expand Watson\u2019s capabilities to<\/a>:
    \n– \u00a0Offer researchers insights to aid clinical trial design, monitoring and evaluation;
    \n– \u00a0Help clinicians to efficiently identify options for the diagnosis, treatment\u00a0 and monitoring a
    \nbroad array of health conditions such as cancer, stroke and heart disease;
    \n– \u00a0Enable providers and payers to integrate and optimize patient engagement in alignment
    \nwith meaningful use and value-based care guidelines; and
    \n– \u00a0Support researchers and healthcare professionals as they advance the emerging discipline
    \nof population health, which aims to optimize an individual\u2019s care by identifying trends in large numbers of people with similar health status.\u201d
    \nIBM is already working towards its goal. Watson is currently working with 30 billion medical images from Merge\u2019s platform to learn how to distinguish normal from abnormal images. With the billions of images from Merge, Watson will learn to compare a patient\u2019s medical images to images of similar patients. Watson will learn to identify differences in scans and the corresponding diseases, tumors, and cancers that each variation represents. Watson is expected to compare scans point-by-point to detect small substances and changes that appear on scans that may not be perceptible to the human eye of the Radiologist.
    \nDespite the superpower that is Watson, Radiologists don\u2019t fear losing their jobs to Watson. They are skeptical of Watson\u2019s ability to read medical images better than them because interpreting radiological images is a task that no computer has ever done, and even more importantly, it is tougher than any task any computer has ever done. Specifically, Dr.\u00a0Daniel Sodickson, the vice chair for research in radiology at the New York University School of Medicine     said<\/a>, “The human brain is a remarkable pattern recognition machine. It’s going to be difficult [for a computer] to beat the brain.” Radiologists are skeptical that a computer can interpret patterns as well as a human.
    \nBut, Radiologists are hopeful, that Watson will one day act as their future colleague. Even though they don\u2019t believe that Watson can be as skilled at reading medical images as the human Radiologists, they are hopeful that Watson can help them sort out medical scans, or confirm their diagnoses. Specifically, Dr. Michael Recht, chair of radiology at the NYU School of Medicine     said<\/a>, “As good as we are as radiologists, and as much training as we have, there are still things that we miss on images every day, and the goal would be to have aids that would help us make sure we wouldn’t miss things.” Many Radiologists agree that Watson could serve as a vital \u201cfirst filter\u201d for medical images which would direct doctors to serious problems immediately and allow them to diagnose the major concerns more quickly. Others say that Watson could act as a second-opinion to confirm or contest doctors\u2019 initial diagnoses. Having Watson as a pre-screening tool for medical images or as a second-opinion for doctors will help cut down on duplicative screening, which will save patients time, and money \u2013 both of which are key reform measures in the Affordable Care Act.<\/p>\n","protected":false},"excerpt":{"rendered":"

    Remember Watson, the super computer that beat the \u201cunbeatable\u201d Ken Jennings at Jeopardy? Well, now Watson may be more to you than a computer with limitless trivia knowledge. He may be your Radiologist, or at least a tool to push some of the reforms of the Affordable Care Act, namely to improve quality care and …<\/a><\/p>\n","protected":false},"author":1,"featured_media":3674,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":[],"categories":[51],"tags":[],"_links":{"self":[{"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/posts\/3670"}],"collection":[{"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/comments?post=3670"}],"version-history":[{"count":1,"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/posts\/3670\/revisions"}],"predecessor-version":[{"id":7290,"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/posts\/3670\/revisions\/7290"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/media\/3674"}],"wp:attachment":[{"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/media?parent=3670"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/categories?post=3670"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/tags?post=3670"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}