Hologic’s US Bid | Image Segmentation, Simplified | Breast MRI’s Early Benefits

“Head horns? Come on.”

Neurosurgeon, Dr. David J. Langer, in response to a widely criticized Australian study suggesting that young adults are developing horn-like bone spurs at the back of their heads due to prolonged smartphone use.


Imaging Wire Sponsors

  • Carestream – Focused on delivering innovation that is life changing – for patients, customers, employees, communities and other stakeholders
  • Focused Ultrasound Foundation – Accelerating the development and adoption of focused ultrasound
  • Medmo – Helping underinsured Americans save on medical scans by connecting them to imaging providers with unfilled schedule time
  • Nuance – AI and cloud-powered technology solutions to help radiologists stay focused, move quickly, and work smarter
  • Pocus Systems – A new Point of Care Ultrasound startup, combining a team of POCUS veterans with next-generation genuine AI technology to disrupt the industry
  • Qure.ai – Making healthcare more accessible by applying deep learning to radiology imaging

The Imaging Wire


Hologic’s Ultrasound Bid

Hologic entered exclusive negotiations to acquire French ultrasound maker, SuperSonic Imagine, potentially leading to a major modality expansion for Hologic and a decent exit for SuperSonic.

  • Breast Health Portfolio Play – Hologic positioned the $85 million bid (a ~43% premium) as a move to expand its breast health capabilities. This strategy drove Hologic’s 2018 acquisitions of Faxitron and Focal ($125m, $85m) as well as its handheld ultrasound OEM deal with Clarius, significantly expanding its breast health lineup in just over a year.
  • Modality and Geography Upside – The acquisition would bring Hologic into the large and growing cart-based global breast ultrasound market ($450m, high single digits CAGR), while expanding Hologic’s European presence (a key strategic goal) and giving SuperSonic Imagine’s ultrasound tech far greater access to the U.S. healthcare system (currently just 15% of SuperSonic’s revenue). The deal would also allow Hologic to support liver and prostate imaging, which probably wasn’t part of its strategy before now, and could be seen as both a growth opportunity and a potential distraction.
  • SuperSonic Imagine’s Exit – Although this surely isn’t what SuperSonic Imagine’s founders had in mind when they created the company 15 years ago, it could be seen as a necessary exit at this point, as the company’s flat revenue and profitability struggles (2018: $27.6m revenue, -$15.1m net loss) has forced it to continue to rely on financing to fund its operations ($43m in debt).


Image Segmentation, Simplified

MIT researchers created a new way to streamline the machine learning training process, using just a single labeled scan along with a set of unlabeled scans to create large databases of training images. Here’s how it works:

  • Image Segmentation – Key to this potential breakthrough is the ability to automatically generate data for the image segmentation process, using a CNN to analyze a range of unlabeled scans (different patients, equipment, etc.) “to ‘learn’ anatomical, brightness, and contrast variations.” The CNN then applies a random combination of these learned variations to a single labeled scan to synthesize new scans (reportedly realistic and accurately labeled), that are then fed into a different CNN that learns how to segment new images.
  • AI Benefits – In addition to massive potential time savings, the team hopes to make image segmentation more accessible in situations where sufficient training data is unavailable (e.g. uncommon pediatric brain conditions).
  • Synthetic Advantages – Although data augmentation is making progress towards a similar synthetic image goal, the researchers noted that their new system overcomes challenges with automation (creating augmentation guidelines, producing unrealistic synthesized images), requiring fewer source scans and creating more realistic synthesized scans with accurate labels.
  • Magic Background – This algorithm actually got its start as a way to synthesize training data for a smartphone app that could identify and retrieve information about “Magic: The Gathering” playing cards (value, ratings, etc.).


Breast MRI’s Early Benefits

Dutch researchers found that breast MRI is able to detect signs of cancer in women with a family history of cancer but no genetic mutations earlier than mammography, suggesting that MRI may help reduce chemotherapy treatment and mortality among this group, although it will also risk more false positives.

  • The team performed a randomized controlled study comparing MRI and DM screening among 1,355 women who fit this profile (who historically make up 15% of all breast cancer occurrences).
  • The women were randomized into a 675-woman MRI group a 680-woman DM group.
  • The MRI group revealed more breast cancers (40 vs. 15) and more invasive cancers (24 vs. 8).
  • The MRI group also had a much smaller median cancer size (9mm vs. 17mm) and earlier. detected tumor stages (45% vs. 7% stage T1a and T1b).

The Wire

  • UNC researchers developed a new method for creating PET radiotracers, which uses existing drugs, but could make it possible to attach radioactive tags to compounds that previously have been difficult or impossible to label. The researchers attached the common radioactive molecule Fluorine-18 into drug molecules, which could allow new ways to screen patients’ drug responses, track cancer progression, and most notably support drug development research.
  • A new report from Moody’s finds that providers who most commonly treat out-of-network patients (radiologists – as well as hospitals, physician staffing companies, labs, and ambulance providers) will absorb the greatest financial impact if one of the various surprise billing legislation efforts is signed into law. The bills that require an in-network guarantee (cover out-of-network charges at in-network rates) would present the greatest challenges to radiologists (and other physicians and ancillary service providers) who aren’t employed by the hospital.
  • Philips announced the version 5.0 update to its EPIQ CVx cardiac ultrasound platform, providing a number of speed and ease-of-use improvements. The platform now includes automated applications for 2D assessment of the heart (reducing clinician touches by 21%) and 3D right ventricle volume and ejection fraction measurements (reducing measurements to as little as 15 seconds). While expanding the EPIC CVxi’s diagnostic capabilities to allow the interventional system to also be used in the echo lab.
  • Research from Boston’s Brigham and Women’s Hospital found that technologists can efficiently and accurately assign protocols for abdominal CT and MRI exams, giving radiologists more time for value-added activities. The study had two CT technologists and two MRI technologists assign protocols (generally done by residents and fellows) over a 15 week period, assigning 22 CT protocols per hour and 19 MRI protocols per hour, with only 0.18% of the CTs (3/1,650) and 0.88% of the MRIs (5/569) reported to have quality issues by radiologists and none requiring call-backs due to protocol error.
  • Research from the Hospital of the University of Pennsylvania and BU Medical School found that machine learning may be able to achieve “near complete” information extraction from radiology reports (findings, recommendations, clinical history, procedures, imaging indications and limitations, etc.), producing the info as “complete, contextualized facts.” Using 120 abdominal and pelvis radiology reports (50 CT, 48 MRI, 22 ultrasound), the team developed an information schema that they combined with custom labeling software and ML extraction architecture, labeling over 5,000 facts and 15,000 pieces of information, including previously unseen facts.
  • Esaote introduced its new cardiovascular ultrasound systems to the U.S. market, including its MyLabX7 compact console system and MyLab Omega high-end portable system, touting their improved image quality, workflow, and ergonomics. The MyLabX7 adopts some of the premium features found in the flagship MyLab 9 system and completes the MyLab X series’ U.S. rollout (joining the X5, while the X6 is not expected to come stateside), while the MyLab Omega sits atop Esaote’s portable lineup.
  • A new paper detailed Northern Ireland’s successful risk-based breast cancer screening program, which may serve as a model for future screening programs and adds context to the debate between risk and age-based screening. The 2013-2017 program achieved ongoing improvements in participants (162 to 372), attendance rate (68.9% to 77.7%), and recall rates (14.2% to 8.6%), while cancer detection rates varied between 21.5 per 1,000 women screened to 30.9 per 1,000 women screened.
  • Visage Imaging announced the global release of Visage 7 Enterprise Imaging Platform version 7.1.14, including 120 enhancements to product usability and to the “dexterity” required to support complex migration and integration scenarios. Highlighted improvements include: comprehensive One Viewer support for multi-modality enterprise viewing, enhanced support for complex multi-identifier-domain environments, enhanced support for combined multi-accession reports, and a scripted interface for automatically populating custom fields in Nuance PowerScribe 360 reports.

The Resource Wire

– This is sponsored content.

  • Headed to SiiM this week? Meet the Qure.ai team at booth #416 and mark your calendar to catch Chief Commercial Officer, Chiranjiv Singh, at the SiiM “Economics of AI” panel on Wednesday June 26th at 11am.

  • This POCUS Systems video details how its forthcoming ultrasound helps patients and providers, and reduces reliance on MSK MRI use.
  • Schedule a demo with Nuance at SIIM and discover the company’s newest AI-powered imaging solutions including PowerScribe One, PowerShare Image Sharing, PowerScribe Follow-up Manager, and AI Marketplace for Diagnostic Imaging. It’s the promise of AI, delivered.
  • With the focused ultrasound adoption curve now at its inflection point, there’s a major shift underway from research-oriented activities to commercialization and patient treatment. Here’s The Focused Ultrasound Foundation’s 2019 Spring Progress Report with details on how and where that shift is taking place.

Leave a Reply

Basic HTML is allowed. Your email address will not be published.

Subscribe to this comment feed via RSS

Get every issue of The Imaging Wire

Delivered right to your inbox. We will never sell or share your information.