Mandatory MOC | AI’s Lucky Winners | No Site Neutrality

“Clinically meaningless scam.”

A tweet from Howard Green, MD in response to a Harvey L. Neiman study on radiologist participation in the ABR MOC program.

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

Mandatory MOC

A Harvey L. Neiman Health Policy Institute study (n= 20,354 radiologists) found that although 56.4% of diagnostic radiologists participate in the ABR Maintenance of Certification (MOC) program, most only do it because they have to. This may not come as a big surprise to some, but the details are still notable.

Participation – An overwhelming 99.6% of radiologists who we’re required to participate in the ABR MOC did so, but only 13.9% of radiologists who had a choice participated. Participation among non-mandated radiologists was highest among radiology subspecialists (17% vs. 11.5% generalists), radiologists from academic practices (28% vs. 11.3% in private practice), and radiologists from groups with over 50 members (20.7% vs. 5% w/ <10 members).

Takeaway – The study took it easy on the ABR in its conclusion, suggesting that this “may reflect diplomate dissatisfaction or negative perceptions about MOC,” but public responses to this study weren’t as kind. Basically, critics think MOC is a shakedown, underscored by the class action lawsuit filed against the ABR this spring.

AI’s Lucky Winners

Dr. Luke Oakden-Rayner released another epic AI blog, arguing that AI competitions might be good for publicity, but they don’t produce useful models. Here’s why:

Multiple Hypothesis Testing – Like a coin flipping contest where five of the 100 participants get “heads” 8 out of 10 times, a certain number of “lucky” AI competitors are going to land the most accurate scores even if those models aren’t necessarily better.

Small Samples – Due to most competitions’ small test sets, there is almost no statistical difference between the winners and middle-of-the-pack competitors. Looking back at previous competitions’ leaderboards, the leader was only statistically superior to the 30th place competitor in the RSNA Pneumonia Detection Challenge, the 56th finisher in the MURA competition, and the 192nd finisher in the SIIM-ACR Pneumothorax Segmentation Challenge.

Goal Conflicts – Even though some may view these contests as a viable way to find good models or useful techniques, this is hard to do if you can’t confirm if a model was actually the best or if it “just happened to fit the under-powered test set.” Instead, the real goals of these contents are (or at least should be) “fun, community building, talent scouting, brand promoting, and attention grabbing.” Those are all solid goals, by the way.

No Site Neutrality

Hospitals landed a big Medicare victory last week when a U.S. federal court overturned the CMS “site-neutral pay policy.” The nearly 10-month old policy paid all healthcare providers the same for performing the same outpatient procedures, regardless of whether services were delivered at off-campus hospital clinics or at private practices. Hospital clinics were previously able to bill Medicare more than private practices, and with this ruling, they will be back to charging more pretty soon.

The Ruling – The judge found that the previous CMS rule didn’t effectively control unnecessary spikes in hospital use, which was its original goal, and it shouldn’t have been used as a price setting tool.

Happy Hospitals – The decision should help hospitals, as AHA estimates suggest that the rule would cost hospitals $380 million in 2019 and $760 million in 2020. At least for now, hospitals won’t be reimbursed for the cuts they incurred so far in 2019, but that might be coming.

Unhappy PhysiciansPhysician groups were not happy about the end of site neutrality, noting that the payment imbalance put private practices at a disadvantage, drove up healthcare costs, and propelled the healthcare consolidation trend.

The Wire

  • A team of China-based researchers performed cone-beam breast CTs (CBBCT) on 409 patients who underwent conventional mammography a week earlier, finding that CBBCT is more comfortable than digital mammography. The patients evaluated the comfort of the scans on an 11-point scale (lower is more comfortable), finding that both non-contrast and contrast-enhanced CBBCT was more comfortable than mammography among most patient groups, particularly <60 year old patients and patients with a BMI above 18.5 kg/m2.
  • Butterfly Network expanded its addressable market last week with the launch of its new iQ Vet veterinary ultrasound. Although Butterfly was sure to point out that the iQ Vet is “an entirely new product” with software presets for a range of animals and their body parts, the device still leads with the company’s ultrasound-on-a-chip technology and sub-$2,000 price tag.
  • Following numerous studies on breast density notification letters’ (BDN) readability challenges, a Johns Hopkins team clarified that adopting simplified BDN language can make a big difference. The team revised Maryland’s state-mandated BDN to a 4.9 Flesch-Kincaid grade reading level (vs. 12.2 original) and surveyed 500 patients (split 50/50 between revised and original version) to evaluate comprehension. Just 2.2% of women who received the revised BDN overestimated the link between dense breasts and lifetime cancer risk (vs. 56.6% w/ original BDN) and 96% of patients who received the revised BDN were likely to discuss breast density with their providers (vs. 32.8% w/ original).
  • Google officially absorbed Alphabet’s DeepMind Health business 10 months after first announcing plans to integrate the AI group and three years after acquiring DeepMind with the agreement that it would remain autonomous. DeepMind will become part of Google Health, which DeepMind suggests will improve its access to Google’s expertise and resources and help with future product development.
  • Aidoc released its “complete AI package” for the identification and triage of ischemic and hemorrhagic strokes in CT scans, combining Aidoc’s CE-marked Large-Vessel Occlusion AI module (for ischemic strokes) and FDA/CE-cleared intracranial hemorrhage AI module (for hemorrhagic strokes) within a single package.
  • ViewRay is facing a class action lawsuit on behalf of investors who held company stock between March 15, 2019 and August 8, 2019. The lawsuit alleges that ViewRay failed to disclose that demand for its systems declined due to changes to Medicare reimbursement, overstated its order backlog, and made false/misleading statements about its business and financial prospects. The lawsuit also noted unusually high ViewRay trading volume (16x above avg.) during the 10 days before its August 8th announcement that revealed operational challenges and reduced its 2019 guidance that led to a 56% stock price decline on August 9th.
  • A study from the University of Colorado School of Medicine found that “fast MRI” is an effective method to identify pediatric traumatic brain injuries, potentially giving physicians an alternative to CT’s radiation exposure and conventional MRI’s anesthesia requirements. The 2015-2018 study of 225 children <6 years-old who underwent emergency CT scans followed by a fast MRI scan within 24 hours found that the fast MRI results matched CT in over 90% of cases and were completed successfully in 99% of cases (6:05 median scan time). CT was still better with skull fractures/breaks, but fast MRI performed better at imaging the brain and the space between the brain and skull.
  • Philips partnered with the non-profit PURE (Point-of-care Ultrasound in Resource-limited Environments) to provide diagnostic ultrasound training to health workers in Rwanda. Leveraging Philips’ Lumify POCUS system’s live-streaming tele-ultrasound capabilities, the program will link Europe and U.S.-based specialists with physicians at Rwanda’s University Hospital of Kigali, who will then train medical workers across the country.

The Resource Wire

This is sponsored content.

  • This Carestream case study compares images of foot trauma captured using the OnSight 3D Extremity System to images captured on 2D X-rays.
  • Qure.ai CCO, Chiranjiv Singh, shares about the importance of building trust in healthcare AI in this editorial.
  • Did you know that imaging patients are most likely to no-show for their procedures on Mondays and Saturdays? By partnering with Medmo, imaging centers can keep their schedules full, despite the inevitable Monday no-shows.
  • Focused ultrasound researchers at the University of Michigan were awarded a $2.4M grant from the National Institute of Biomedical Imaging and Bioengineering (NIBIB) to study using histotripsy for brain applications. Histotripsy is a form of focused ultrasound that mechanically destroys tissue without heat.

You might also like

You might also like..

Select All

You're signed up!

It's great to have you as a reader. Check your inbox for a welcome email.

-- The Imaging Wire team

You're all set!