• Login
    • Join
  • FOLLOW:
  • Subscribe Free
    • Magazine
    • eNewsletter
    Checkout
    • Magazine
    • News
    • Opinions
    • Top 30
    • Research
    • Supply Chain
    • Device Sectors
    • Directory
    • Events
    • Resources
    • Microsites
    • More
  • Magazine
  • News
  • Opinions
  • Top 30
  • Research
  • Supply Chain
  • Device Sectors
  • Directory
  • Events
  • Resources
  • Microsites
  • Current / Back Issues
    Features
    Editorial
    Digital Edition
    eNewsletter Archive
    Our Team
    Editorial Guidelines
    Reprints
    Subscribe Now
    Advertise Now
    Top Features
    Sensing Technology Drives the Future of Medical Care

    New and Developing Diabetes Technologies Offer 'Sweet Relief'

    Discussing Matters of Substance in Medtech Materials

    Medical Device Testers Are Caught Up in a Whirlwind

    6 Ways ERP Systems Help Medical Device Manufacturers Manage Risk and Profitability
    OEM News
    Supplier News
    Service / Press Releases
    Online Exclusives
    Press Releases
    People in the News
    Product & Service Releases
    Supplier News
    Medtech Makers
    Technical Features
    International News
    Videos
    Product & Service Releases
    Live From Shows
    Regulatory
    Financial/Business
    Top News
    David Schnur Associates Adds ERI Group to Its Network of Partners

    BlackHägen Design Promotes Jeff Morang to Director of Human Factors Engineering

    Great Speech Wins First Place at ATA 2023 Innovator’s Challenge

    FDA Clears Magstim Transcranial Magnetic Stimulation for OCD

    3D Virtual Simulation Helps Medical Students Diagnose, Treat Patients
    From the Editor
    Blogs
    Guest Opinions
    Top Opinions
    Sensing Technology Drives the Future of Medical Care

    New and Developing Diabetes Technologies Offer 'Sweet Relief'

    Discussing Matters of Substance in Medtech Materials

    Medical Device Testers Are Caught Up in a Whirlwind

    6 Ways ERP Systems Help Medical Device Manufacturers Manage Risk and Profitability
    Top 30 Medical Device Companies
    Market Data
    White Papers
    Top Research
    CR-SOP Neurotechnology Helps to Achieve ‘Sound’ Sleep

    Inside the Hospital Value Analysis Committee

    Face Time: Why Getting Engineers Out of the Lab Is Good for Business

    Common Paratubing Pitfalls and How to Avoid Them

    During Times of Uncertainty, Create Business Value with an Optimized Workforce
    3D/Additive Manufacturing
    Contract Manufacturing
    Electronics
    Machining & Laser Processing
    Materials
    Molding
    Packaging & Sterilization
    R&D & Design
    Software & IT
    Testing
    Tubing & Extrusion
    Cardiovascular
    Diagnostics
    Digital Health
    Neurological
    Patient Monitoring
    Surgical
    Orthopedics
    All Companies
    Categories
    Company Capabilities
    Add New Company
    Outsourcing Directory
    Cirtec Medical

    Forefront Medical Technology

    NDH Medical Inc.

    JBC Technologies

    LEMO USA Inc.
    MPO Summit
    Industry Events
    Webinars
    Live From Show Event
    Industry Associations
    Videos
    Career Central
    eBook
    Slideshows
    Top Resources
    3D Printing Medical Devices Revolutionized Manufacturing

    5 Current Trends in Medical Device Security

    How Advanced Sensors Improve Design & Functionality of Respiratory Care Equipment

    Sourcing Silicone Parts: Injection Mold or 3D Print?

    Cybersecurity in Healthcare: Getting on the Right Track
    Companies
    News Releases
    Product Releases
    Press Releases
    Product Spec Sheets
    Service Releases
    Case Studies
    White Papers
    Brochures
    Videos
    Outsourcing Directory
    Cirtec Medical

    Forefront Medical Technology

    NDH Medical Inc.

    JBC Technologies

    LEMO USA Inc.
    • Magazine
      • Current/Back Issues
      • Features
      • Editorial
      • Columns
      • Digital Editions
      • Subscribe Now
      • Advertise Now
    • News
    • Directory
      • All Companies
      • ALL CATEGORIES
      • Industry Associations
      • Company Capabilities
      • Add Your Company
    • Supply Chain
      • 3D/Additive Manufacturing
      • Contract Manufacturing
      • Electronics
      • Machining & Laser Processing
      • Materials
      • Molding
      • Packaging & Sterilization
      • R&D & Design
      • Software & IT
      • Testing
      • Tubing & Extrusion
    • Device Sectors
      • Cardiovascular
      • Diagnostics
      • Digital Health
      • Neurological
      • Patient Monitoring
      • Surgical
      • Orthopedics
    • Top 30 Company Report
    • Expert Insights
    • Slideshows
    • Videos
    • eBook
    • Resources
    • Podcasts
    • Infographics
    • Whitepapers
    • Research
      • White Papers
      • Case Studies
      • Product Spec Sheets
      • Market Data
    • MPO Summit
    • Events
      • Industry Events
      • Live From Show Events
      • Webinars
    • Microsite
      • Companies
      • Product Releases
      • Product Spec Sheets
      • Services
      • White Papers / Tech Papers
      • Press Releases
      • Videos
      • Literature / Brochures
      • Case Studies
    • About Us
      • About Us
      • Contact Us
      • Advertise with Us
      • eNewsletter Archive
      • Privacy Policy
      • Terms of Use
    Breaking News

    From One Brain Scan, More Info for Medical AI

    System helps machine-learning models glean training information for diagnosing and treating brain conditions.

    From One Brain Scan, More Info for Medical AI
    MIT researchers have developed a system that gleans far more labeled training data from unlabeled data, which could help machine-learning models better detect structural patterns in brain scans associated with neurological diseases. The system learns structural and appearance variations in unlabeled scans, and uses that information to shape and mold one labeled scan into thousands of new, distinct labeled scans. Image courtesy of the researchers.
    Rob Matheson, MIT News Office06.20.19
    MIT researchers have devised a novel method to glean more information from images used to train machine-learning models, including those that can analyze medical scans to help diagnose and treat brain conditions.
     
    An active new area in medicine involves training deep-learning models to detect structural patterns in brain scans associated with neurological diseases and disorders, such as Alzheimer’s disease and multiple sclerosis. But collecting the training data is laborious: All anatomical structures in each scan must be separately outlined or hand-labeled by neurological experts. And, in some cases, such as for rare brain conditions in children, only a few scans may be available in the first place.
     
    In a paper presented at the recent Conference on Computer Vision and Pattern Recognition, the MIT researchers describe a system that uses a single labeled scan, along with unlabeled scans, to automatically synthesize a massive dataset of distinct training examples. The dataset can be used to better train machine-learning models to find anatomical structures in new scans—the more training data, the better those predictions.
     
    The crux of the work is automatically generating data for the “image segmentation” process, which partitions an image into regions of pixels that are more meaningful and easier to analyze. To do so, the system uses a convolutional neural network (CNN), a machine-learning model that’s become a powerhouse for image-processing tasks. The network analyzes a lot of unlabeled scans from different patients and different equipment to “learn” anatomical, brightness, and contrast variations. Then, it applies a random combination of those learned variations to a single labeled scan to synthesize new scans that are both realistic and accurately labeled. These newly synthesized scans are then fed into a different CNN that learns how to segment new images.
     
    “We’re hoping this will make image segmentation more accessible in realistic situations where you don’t have a lot of training data,” said first author Amy Zhao, a graduate student in the Department of Electrical Engineering and Computer Science (EECS) and Computer Science and Artificial Intelligence Laboratory (CSAIL). “In our approach, you can learn to mimic the variations in unlabeled scans to intelligently synthesize a large dataset to train your network.”
     
    There’s interest in using the system, for instance, to help train predictive-analytics models at Massachusetts General Hospital, Zhao said, where only one or two labeled scans may exist of particularly uncommon brain conditions among child patients.
     
    Joining Zhao on the paper are: Guha Balakrishnan, a postdoc in EECS and CSAIL; EECS professors Fredo Durand and John Guttag, and senior author Adrian Dalca, who is also a faculty member in radiology at Harvard Medical School.
     
    The “Magic” Behind the System
    Although now applied to medical imaging, the system actually started as a means to synthesize training data for a smartphone app that could identify and retrieve information about cards from the popular collectable card game, “Magic: The Gathering.” Released in the early 1990s, “Magic” has more than 20,000 unique cards—with more released every few months—that players can use to build custom playing decks.
     
    Zhao, an avid “Magic” player, wanted to develop a CNN-powered app that took a photo of any card with a smartphone camera and automatically pulled information such as price and rating from online card databases. “When I was picking out cards from a game store, I got tired of entering all their names into my phone and looking up ratings and combos,” Zhao said. “Wouldn’t it be awesome if I could scan them with my phone and pull up that information?”
     
    But she realized that’s a very tough computer-vision training task. “You’d need many photos of all 20,000 cards, under all different lighting conditions and angles. No one is going to collect that dataset,” Zhao said.
     
    Instead, Zhao trained a CNN on smaller dataset of around 200 cards, with 10 distinct photos of each card, to learn how to warp a card into various positions. It computed different lighting, angles, and reflections—for when cards are placed in plastic sleeves—to synthesized realistic warped versions of any card in the dataset. It was an exciting passion project, Zhao said: “But we realized this approach was really well-suited for medical images, because this type of warping fits really well with MRIs.”
     
    Mind Warp
    Magnetic resonance images (MRIs) are composed of three-dimensional pixels, called voxels. When segmenting MRIs, experts separate and label voxel regions based on the anatomical structure containing them. The diversity of scans, caused by variations in individual brains and equipment used, poses a challenge to using machine learning to automate this process.
     
    Some existing methods can synthesize training examples from labeled scans using “data augmentation,” which warps labeled voxels into different positions. But these methods require experts to hand-write various augmentation guidelines, and some synthesized scans look nothing like a realistic human brain, which may be detrimental to the learning process.
     
    Instead, the researchers’ system automatically learns how to synthesize realistic scans. The researchers trained their system on 100 unlabeled scans from real patients to compute spatial transformations—anatomical correspondences from scan to scan. This generated as many “flow fields,” which model how voxels move from one scan to another. Simultaneously, it computes intensity transformations, which capture appearance variations caused by image contrast, noise, and other factors.
     
    In generating a new scan, the system applies a random flow field to the original labeled scan, which shifts around voxels until it structurally matches a real, unlabeled scan. Then, it overlays a random intensity transformation. Finally, the system maps the labels to the new structures, by following how the voxels moved in the flow field. In the end, the synthesized scans closely resemble the real, unlabeled scans—but with accurate labels.
     
    To test their automated segmentation accuracy, the researchers used Dice scores, which measure how well one 3D shape fits over another, on a scale of 0 to 1. They compared their system to traditional segmentation methods—manual and automated—on 30 different brain structures across 100 held-out test scans. Large structures were comparably accurate among all the methods. But the researchers’ system outperformed all other approaches on smaller structures, such as the hippocampus, which occupies only about 0.6 percent of a brain, by volume.
     
    “That shows that our method improves over other methods, especially as you get into the smaller structures, which can be very important in understanding disease,” Zhao said. “And we did that while only needing a single hand-labeled scan.”
     
    In a nod to the work’s “Magic” roots, the code is publicly available on Github under the name of one of the game’s cards, “Brainstorm.”
    Related Searches
    • it
    • imaging
    • medical imaging
    • paper
    Related Knowledge Center
    • Neurological
    • Patient Monitoring
    • Software & IT
    Suggested For You
    FDA Clears Siemens Healthineers FDA Clears Siemens Healthineers' Aidan AI
    Glytec Receives Another Patent Allowance for Therapy Advisor Glytec Receives Another Patent Allowance for Therapy Advisor
    Smartphone App Can Help Detect Huntington Smartphone App Can Help Detect Huntington's Early
    Using AI to Predict Breast Cancer and Personalize Care Using AI to Predict Breast Cancer and Personalize Care
    FDA Approves Boston Scientific FDA Approves Boston Scientific's VICI Venous Stent System
    Imaging System Helps Surgeons Remove Tiny Ovarian Tumors Imaging System Helps Surgeons Remove Tiny Ovarian Tumors
    Diagnosing the Cancer in Medical Imagery Security Diagnosing the Cancer in Medical Imagery Security
    Physician and iTether Create App for Pregnant Women & New Moms with Opioid Addiction Physician and iTether Create App for Pregnant Women & New Moms with Opioid Addiction
    FDA OKs New Device to Help Treat Carbon Monoxide Poisoning FDA OKs New Device to Help Treat Carbon Monoxide Poisoning
    Oxygen-Tracking Method Could Improve Diabetes Treatment Oxygen-Tracking Method Could Improve Diabetes Treatment
    New MRI Sensor Can Image Activity Deep within the Brain New MRI Sensor Can Image Activity Deep within the Brain
    ExThera Medical Elects Two New Board Members ExThera Medical Elects Two New Board Members
    Pill Can Deliver Insulin via Stomach Injection Pill Can Deliver Insulin via Stomach Injection
    Electrifying Healthcare: A Look at Electronics Manufacturing Services for Medtech Electrifying Healthcare: A Look at Electronics Manufacturing Services for Medtech
    Ingestible, Expanding Pill Monitors Stomach for up to a Month Ingestible, Expanding Pill Monitors Stomach for up to a Month

    Related Breaking News

    • Diagnostics
      FDA Clears Siemens Healthineers

      FDA Clears Siemens Healthineers' Aidan AI

      The latest technology enables new features and enhanced security.
      Siemens Healthineers 04.22.20

    • Glytec Receives Another Patent Allowance for Therapy Advisor

      Glytec Receives Another Patent Allowance for Therapy Advisor

      The company is expanding its software-as-a-medical-device platform to support the optimization of ALL diabetes medications.
      Business Wire 05.29.19

    • Diagnostics | Digital Health | Neurological | Software & IT
      Smartphone App Can Help Detect Huntington

      Smartphone App Can Help Detect Huntington's Early

      The app provides its user with a series of tests in order to check the presence of the symptoms.
      Kaunas University of Technology (KTU) 05.23.19


    • Diagnostics | Digital Health | Software & IT
      Using AI to Predict Breast Cancer and Personalize Care

      Using AI to Predict Breast Cancer and Personalize Care

      MIT/MGH's image-based deep learning model can predict breast cancer up to five years in advance.
      Adam Conner-Simons & Rachel Gordon, CSAIL 05.08.19

    • Cardiovascular | Surgical
      FDA Approves Boston Scientific

      FDA Approves Boston Scientific's VICI Venous Stent System

      New stent system is now available for treating patients with deep venous blockages.
      Boston Scientific 05.06.19

    Loading, Please Wait..

    Trending
    • FDA Clears LiveMetric's Wearable Blood Pressure Monitoring Tech
    • The Future Of Biomedical Engineering Advancements
    • A New Approach To Post-Market Surveillance
    • Philips Names New Supervisory Board Chairman
    • Healthcare Changes Prompt Medtronic To Merge Sales Force Medtronic, Inc. Is Combining Its U.S. Ca
    Breaking News
    • David Schnur Associates Adds ERI Group to Its Network of Partners
    • BlackHägen Design Promotes Jeff Morang to Director of Human Factors Engineering
    • Great Speech Wins First Place at ATA 2023 Innovator’s Challenge
    • FDA Clears Magstim Transcranial Magnetic Stimulation for OCD
    • 3D Virtual Simulation Helps Medical Students Diagnose, Treat Patients
    View Breaking News >
    CURRENT ISSUE

    March 2023

    • Sensing Technology Drives the Future of Medical Care
    • New and Developing Diabetes Technologies Offer 'Sweet Relief'
    • Discussing Matters of Substance in Medtech Materials
    • Medical Device Testers Are Caught Up in a Whirlwind
    • View More >

    Cookies help us to provide you with an excellent service. By using our website, you declare yourself in agreement with our use of cookies.
    You can obtain detailed information about the use of cookies on our website by clicking on "More information”.

    • About Us
    • Privacy Policy
    • Terms And Conditions
    • Contact Us

    follow us

    Subscribe
    Nutraceuticals World

    Latest Breaking News From Nutraceuticals World

    House Reps Try Again to Regulate CBD and Other Hemp Derivatives as Dietary Ingredients
    'What’s Up With Supps' Event at Expo West Raises Over $8,000 for Vitamin Angels
    IRI and NPD Rebrand as Circana Following Merger
    Coatings World

    Latest Breaking News From Coatings World

    BASF Invests in Expansion of Polymer Dispersions Plant in Daya Bay, China
    Stahl Completes Acquisition of ICP Industrial Solutions Group
    Interpon Powder Coatings on Dubai’s Latest Luxury Hotel ​
    Medical Product Outsourcing

    Latest Breaking News From Medical Product Outsourcing

    David Schnur Associates Adds ERI Group to Its Network of Partners
    BlackHägen Design Promotes Jeff Morang to Director of Human Factors Engineering
    Great Speech Wins First Place at ATA 2023 Innovator’s Challenge
    Contract Pharma

    Latest Breaking News From Contract Pharma

    BioNTech, OncoC4 Partner to Develop mAb Candidate ONC-392 in Cancer
    FLAMMA to Invest $200M Over the Next Three Years
    Pii Forms New Leadership Team
    Beauty Packaging

    Latest Breaking News From Beauty Packaging

    Selena Gomez Becomes Most-Followed Woman on Instagram
    Former Estée Lauder Execs Join Stila Cosmetics' Advisory Board
    Ole Henriksen's Eye Stick is Made from Limestone
    Happi

    Latest Breaking News From Happi

    L’Oréal Invests in Geno-led Venture for Biotech Ingredients
    Saltair Adds Santal Bloom Body Oil to Personal Care Collection
    Fredericksburg Farms Recalls Scented Candles with Glass Lids
    Ink World

    Latest Breaking News From Ink World

    Chris Scully Takes Over Koenig & Bauer UK
    Kodak Reports 4Q, Full-Year 2022 Financial Results
    Fujifilm Releases Uvijet HZ Thermoforming Ink for Acuity Prime Series
    Label & Narrow Web

    Latest Breaking News From Label & Narrow Web

    Imageworx showcasing Lombardi presses at new Innovation Center
    Schreiner PrinTrust supplies RFID-enabled labels for seventhings
    McDonald’s to publish report on reusable packaging
    Nonwovens Industry

    Latest Breaking News From Nonwovens Industry

    Weekly Recap: TerraCycle Launches Diaper Program, Ahlstrom to Divest or Shutter Stenay Plant & More
    Mativ Names CFO
    Indorama Ventures, Polymateria Partner on Biotransformation Technology
    Orthopedic Design & Technology

    Latest Breaking News From Orthopedic Design & Technology

    Study Highlights the Positive Clinical Outcomes of Medacta’s MySpine System
    NuVasive Receives Expanded Indications for Precice Limb Lengthening Solution
    Alex Jahangir Receives Highest Leadership Award From AAOS
    Printed Electronics Now

    Latest Breaking News From Printed Electronics Now

    LG Display’s 65-Inch OLED TV Panel Receives Carbon Footprint Certification
    Overcoming the Four Greatest MaaS Ticketing Challenges
    Jabil Posts Second Quarter 2023 Results

    Copyright © 2023 Rodman Media. All rights reserved. Use of this constitutes acceptance of our privacy policy The material on this site may not be reproduced, distributed, transmitted, or otherwise used, except with the prior written permission of Rodman Media.

    AD BLOCKER DETECTED

    Our website is made possible by displaying online advertisements to our visitors.
    Please consider supporting us by disabling your ad blocker.


    FREE SUBSCRIPTION Already a subscriber? Login