“If there is actually an elephant in the room, you as a human would likely notice it.”
York University researcher, Amir Rosenfeld, on the challenges that AI neural networks have when an unexpected object is included in a study image.
The Imaging Wire is Brought to You By
- 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
- Pocus Systems – A new Point of Care Ultrasound startup, combining a team of POCUS veterans with next-generation technology to disrupt the industry
Keep these companies in mind each time you enjoy The Imaging Wire. Check them out and see how they’re driving our industry forward, and shoot them a note if you’re interested in learning more. They’re all great companies run by solid people.
The Imaging Wire
Silver Tsunami Warning
HIMSS CEO, Hal Wolf, issued a Silver Tsunami Warning to the healthcare industry, suggesting that the world’s aging population will shift even more healthcare costs to the government (in US from 49% to 53%) and exacerbate the healthcare professional shortage (currently a 7 million global shortage). The executive emphasized that this shift will require the healthcare industry to innovate at a far greater speed and scale than today, with a focus on better using available data and an expectation that patients will increasingly take health monitoring and management into their own hands.
AI Fails the Vision Test
A study from York University and University of Toronto in Canada found that that artificial intelligence systems fail a vision test that children can easily accomplish, highlighting a key challenge with current AI computer vision / object detection technology. In the study, a neural network system correctly identified objects in an image of a room (chair, couch, etc.) but things went haywire when it viewed a second image of the room that happened to have an elephant in it (get it…). As a result of the new unexpected object, the system labeled the chair a couch, called the elephant a chair, and was unable to even see items that it correctly identified in the previous image. Here’s why: humans can do a double-take in cases where we see an unexpected object in a scene, and computer vision systems and neural network technology can’t (at least for now). Coverage of the study didn’t specifically reference medical imaging (more of a focus on autonomous driving and similar applications), but medical imaging AI certainly relies on neural networks and this study is worth noting.
Mercy Health Invests
Mercy Health continued what is shaping up to be an ambitious approach to IT for a healthcare system, making a strategic investment in NucleusHealth. Mercy already uses NucleusHealth’s cloud-based image exchange and just adopted NucleusHealth’s Nucleus.io as its back-up PACS. With this investment, the hospital system becomes a minority stakeholder in the San Diego-based company. Mercy Health is among the select group health systems that funds some vendor partners that it believes will benefit the global health care community (like NucleusHealth), and last week it’s IT division (Mercy Technology Services) joined an even more select group of healthcare provider divisions offering services to other hospitals, after announcing plans to go to market with a “PACS as a Service” offering.
IHS Markit Forecasts Mobile MRI Growth
IHS Markit forecasted strong (but unspecified) growth for mobile MRIs, largely attributed to the segment’s overall ease of adoption and much lower up-front investments (they can be rented or shared, don’t require an MRI room, etc.), making mobile MRIs relevant to rural and budget-conscious healthcare providers. Mobile MRI growth is further supported by technological (MRIs becoming lighter, smaller, self-shielding), economic (hospitals increasingly budget-conscious), and societal (the population is aging) factors. Although not specifically tied to the mobile segment, the IHS Markit report also provided some interesting MRI growth numbers, forecasting a 5.1% revenue CAGR for 1.5T MRIs (7% in China) and a -2.7% CAGR decline in average selling prices across all MRI segments through 2022.
- Research on PACS preferences from the University of Michigan found that radiologists place the greatest value on stability, reliability, interoperability, fast load times, and minimizing repetitive and non-value-added work. Meanwhile, “niche add-ons” like mobile device compatibility, connectivity to non-affiliated sites, and integrated instant messaging were reported to be the least valuable PACS features.
- The NIH’s National Heart, Lung and Blood Institute awarded Case Western Reserve University (CWRU) School of Medicine a $1.1 million grant to explore a new imaging approach for diagnosing peripheral arterial disease (PAD) that does not rely on gadolinium-based contrast agents and could therefore be used on patients with kidney disease. The CWRU researchers are specifically targeting non-contrast-enhanced magnetic resonance angiography (MRA), noting that this method has already shown promising results, but has to overcome a number of challenges that limit its effectiveness for PAD and other applications.
- DiA Imaging Analysis announced that its AI-powered LVivo EF cardiac decision-support software is now available for use with GE Healthcare’s Vscan Extend handheld mobile ultrasound. LVivo EF (FDA and CE cleared) is the first AI-powered ejection fraction automated app intended to support mobile ultrasound and is used to detect patterns in ultrasound scans (traditionally done visually).
- Despite digital breast tomosynthesis’ longer image acquisition and interpretation times, research from the UW School of Medicine found that facilities that have adopted DBT have not experienced reduced monthly breast cancer screening volumes (this was a concern with DBT). The research found that facility-level volume was slightly improved 15 months after DBT adoption, suggesting that DBT may lead to slightly higher screening volumes because the technique results in fewer recalls and requires shorter diagnostic workup times.
- Laurel Bridge rolled-out enhancements to its Exodus Migration and Consolidation Controller, improving healthcare providers’ ability to internally manage their own archive migration and consolidation. The update brings improvements to Exodus’ interface, migration status monitoring tools, migration speed (via load balancing), study-based query performance, and support for a Source-of-Truth migration approach.
- NYU researchers developed an artificial intelligence algorithm to correctly distinguish between adenocarcinoma and squamous cell carcinoma in histopathology slides with 97% accuracy, while also determining the presence of abnormal versions of six genes linked to lung cancer with 73% to 86% accuracy (depending on the gene). The research suggests that AI could help physicians instantly determine cancer subtype and mutational profile and start patients on targeted therapies sooner.
- A UCSD team developed a wearable ultrasound patch capable of continuously monitoring central blood pressure in arteries and veins up to 4cm under the skin, allowing earlier and more precise cardiovascular diagnosis, and potentially replacing the use of tonometers. There’s still work to do before this new technology is clinic-ready, and the researchers will next work to integrate a power source, data processing units, and wireless communication capability.
- Careful what you delegate. Research from the UK reveals that radiologist trainees are responsible for 52% of all reporting errors, even though trainees only produce 18% of all reports.
- ViewRay’s MRIdian Linac MRI-guided radiation therapy system continued its expansion with an installation at Istanbul’s Acibadem Maslak Hospital, becoming the first and only MRI-guided radiation therapy system used in Turkey. The MRIdian Linac system has gained placements in France, Illinois, Wisconsin, and Missouri in recent months.
- An SEC filing revealed that Mobius Imaging completed a $25 million funding round this summer, but there are no details on how Mobius will use the investment. The company makes a range of imaging equipment and appears to lead with its CT systems.
The Resource Wire
– This is sponsored content.
- Yale University research reveals that the average patient drives past SIX lower-priced providers on the way to an imaging procedure, due in large part to patients’ and physicians’ limited cost consciousness. Medmo helps address this issue by letting patients enter the price they can afford for their scan, then booking them at a nearby imaging center willing to accept that price.
- The Focused Ultrasound Foundation’s 2018 State of the Field Report documents the technology’s progress over the last year, with details on research and regulatory status, as well as insights into the various treatments being developed.
- In this Carestream video, an orthopedic surgeon opens up about why he decided to add the OnSight 3D Extremity System and how his practice benefits from the weight bearing CT.
- POCUS Systems is now approved as a Veteran Owned Business with the US Government Office of Veterans Business Development, paving the way for partnerships with the federal healthcare delivery systems.