fbpx

MARCA Madness | Hyperfine Validation

“Radiology is not caviar anymore, it’s chicken tikka masala.”

Penn Medicine’s Saurabh Jha, MD on radiology’s expansion and commoditization.


I’m excited to share the latest Imaging Wire Show, featuring Blackford Analysis founder and CEO, Ben Panter.

Join us for a deep dive into imaging AI. We discuss how to solve AI’s assessment and deployment problem, AI’s downstream value, population health AI’s potential, and what it will take for AI to have its greatest impact.



Imaging Wire Sponsors

Arterys | Bayer Radiology | Blackford Analysis
Canon Medical Systems | Fujifilm Healthcare Americas
GE Healthcare | Novarad | Nuance
Riverain Technologies | Siemens Healthineers
United Imaging | Zebra Medical Vision



The Imaging Wire


MARCA Madness

The American College of Radiology announced plans to remain neutral on the Medicare Access to Radiology Care Act (MARCA), creating quite an uproar among its members.

About MARCA – If approved, MARCA would require Medicare to reimburse imaging services performed by radiologist assistants, as long as RAs work within physician-led teams (not independently).

ACR Neutrality – The ACR attributed its neutral position to “the widely diverse opinions of its members,” revealing that it won’t lobby Congress for or against MARCA 2021, but will reconsider its position based on future member feedback.

Member Negativity – Assuming ACR members really do have “widely diverse opinions” about MARCA, all those members with positive positions kept pretty quiet last week. The rest of the members…

About that RP Ad – Things took a turn for the worse when people discovered that the ACR was advertising a Radiology Partners job for a “Radiology Physician Assistant or Nurse Practitioner” that would be responsible for ordering and interpreting radiographic studies. It’s safe to say this got the folks on Radiology Twitter and the Aunt Minnie Forum riled up, as many saw this as confirmation of the ACR’s not-so-neutral allegiances (they eventually removed the job post).

The Silent Neutrality – The folks against MARCA were certainly the loudest, but you could still find some balanced perspectives if you looked hard enough. These people…

  • Argued that MARCA won’t stop scope creep
  • Shared their positive experiences with mid-levels
  • Noted that mid-levels “run” other departments and those physicians are still doing fine
  • Suggested that RAs should handle mundane tasks (e.g. CXR tube measurements)
  • Forecast that these mundane tasks will be handled by AI in the future anyway

The Takeaway – It’s pretty clear that this ACR MARCA uproar is about far more than whether RAs could safely handle additional responsibilities (it’s also about: career stability, corporatization, rising image volumes, falling reimbursements, AI on the horizon, etc.). However, it’s less clear how the ACR can remain neutral on MARCA for much longer.


CVIS’ Cloud Advantages

This Diagnostic and Interventional Cardiology article details the unique advantages of cloud-based CVIS systems (off-property access, team collaboration), with insights from one Mississippi-based cardiologist on the benefits of Fujifilm Healthcare’s VidiStar CVIS.

– Sponsored.


UCSD’s Case for Arterys Lung AI

See how UCSD reduced its lung nodule detection times and increased its reader performance using Arterys Lung AI.

– Sponsored.


The Wire

  • Hyperfine Validation: Hyperfine Research landed a favorable Nature Communications study validating its portable MRI’s (mMRI) ability to detect and characterize intracerebral hemorrhage (ICH) at patient bedsides, and suggesting that pMRI neuroimaging “could be useful in a broad range of clinical settings.” The Yale study had two neuroradiologists interpret 144 Hyperfine pMRI exams (56 ICH, 48 acute ischemic stroke, 40 healthy controls), detecting ICH with 80.4% sensitivity and 96.6% specificity. They also found that pMRI segmented / estimated ICH volumes similarly to conventional CT/MRI exams.
  • Revolutionizing Whiplash: Australia’s Kolling Institute and a team of US collaborators developed a cervical MRI analysis technique that they say might revolutionize whiplash diagnosis and treatment planning. The researchers developed a deep learning model to streamline cervical spine muscle segmentation and composition analysis (30 seconds vs. 4-8hrs), suggesting that its combination of speed and accuracy could allow earlier interventions for severe whiplash.
  • DiA’s $14m: Ultrasound AI startup DiA Imaging Analysis announced a $14m round (increasing total to $25m) that it will use to fund commercial expansion, product development, and seek more vendor and channel partnerships. The ultrasound AI space is heating up, as this round comes just a few weeks after Ultromics’ $33m Series B ($58.9m total) and a year after Caption Health’s $53m Series B ($60.9m total), both of which also emphasized commercialization.
  • Opportunistic Sarcopenia Screening: A new AJR study detailed a deep learning model that could be used to opportunistically screen existing chest and abdominal CT scans for sarcopenia (age-related muscle loss). The researchers used the DL model to analyze abdominal CTs from 9,223 asymptomatic adults, finding that image analysis at the L1 vertebral level was more accurate than L3 analysis and established risk scores for predicting hip fractures and death (hip fracture ROC = 0.717 vs. 0.709 vs. 0.710; death ROC = 0.737 vs. 0.721 vs. 0.688).
  • The U.S. Navy Wants Fujifilm: Fujifilm Medical Systems landed a 10-year/$10m sole source contract to provide the U.S. Navy with X-ray systems, following a two-year competitive evaluation. Starting later this year, Fujifilm will begin equipping Navy vessels with its DR systems.
  • Rad Underestimates: A study in the Journal of Trauma and Acute Care Surgery found that radiologists commonly underappreciated rib fracture severity in chest CTs, encouraging surgeons to evaluate CTs on their own before making patient management decisions. The single-center study reviewed chest CTs from 410 adult patients (2,337 rib fractures, 1-5 severity scores), finding that radiologists’ mean severity scores were well below surgeons (1.58 vs. 1.78), radiologists missed 5.9% of fractures in initial CT exams, and they struggled detecting displaced rib fractures (sensitivity 54.9%, specificity 79.9%).
  • Northern Ireland’s Diagnostic Review: The NHS Northern Ireland Trust’s review of 9k patients’ radiology exams that were performed by a questionable locum consultant radiologist has uncovered 43 incorrect diagnoses so far, including six extremely serious cases. The Northern Trust began its review in June and they’ve reviewed 7,902 of 13,030 flagged exams so far.
  • CT Dose Reduction Faceoff: A new study out of South Korea suggests that ClariPI’s ClariCT.AI image denoising model can reduce CT dosage as well as GE’s TrueFidelity DLIR solution, while maintaining similar image detectability. The research team (which included ClariPI’s CTO) processed CT images using both solutions, finding that ClariCT.AI achieved similar or greater dose reduction than TrueFidelity (ClariCT.AI: 86% vs. TrueFidelity: 87% at high strength, 76% at mid, 60% at low).
  • AI for Diagnostic Training: A new UPenn study detailed an AI-augmented radiology education approach that improved non-experts’ ability to diagnose diffuse parenchymal lung diseases (DPLD) in chest CTs. The study had 25 non-experts (radiologists, residents, pulmonologists) list the top three potential diagnoses in a set of DPLD CTs, before and after using the pattern-based training algorithm. The non-experts’ accuracy with their first-choice diagnosis and top-three differential diagnoses increased significantly after training (before: 32.5% & 49.7%; after: 41.2% & 65%), bringing them close to the study’s five thoracic radiologists (48% & 64.3%).
  • Prenatal Ultrasound Disparities: A new Academic Radiology study revealed widespread obstetrical ultrasound imaging utilization disparities. The Saskatchewan-based study (80.5k pregnancies, 57.8k individuals) found that prenatal ultrasound was performed during the 2nd trimester of 87.7% of pregnancies overall. However, patients had far lower odds of undergoing 2nd trimester ultrasound if they lived in rural areas (-30% vs. urban areas), were members of Canada’s First Nations (-50% vs. not First Nations), or lived in the lowest-income neighborhoods (-86% vs. highest).
  • Stanford’s AI Recommendations: Stanford AI leaders detailed the policy barriers that must be overcome in order for imaging AI to achieve its potential, encouraging policymakers to work with medical societies to develop stronger regulatory processes and standards. The paper suggested addressing regulatory gaps by: 1) Making sure algorithms are distinguished from their diagnostic tasks; 2) Defining algorithmic performance beyond accuracy (e.g. transparency, and auditability); 3) Expanding the AI evaluation process; and 4) Encouraging independent 3rd party testing.

Take the Canon AiCE Challenge

Take the AiCE challenge and see why half the radiologists in a recent study “had difficulty differentiating” images from Canon Medical Systems’ Vantage Orian 1.5T MR using its AiCE reconstruction technology compared to standard 3T MRI images.

– Sponsored.



The Resource Wire

  • See how Einstein Healthcare Network reduced its syringe expenses, enhanced its syringe loading, and improved its contrast documentation when it upgraded to Bayer Radiology’s MEDRAD Stellant FLEX CT Injection System.
  • When Birmingham Radiological Group-GV adopted Nuance PowerScribe One, the practice eliminated 60-75 minutes in daily reporting time and reduced calls to the radiology reading room by 80% by getting its reports to clinicians faster. See how in this Nuance Case Study.
  • This European Radiology study highlighted Riverain Technologies’ ClearRead Xray – Detect as one of just two imaging AI products to achieve the FDA’s most stringent premarket approval level. See how they measured up against the other 99 AI tools here.
  • Room for more efficiency in your breast imaging operations? Check out this GE Healthcare post detailing how new technologies are improving patient experiences and making breast imaging teams more efficient.
  • This Medical Image Analysis study detailed an uncertainty-based learning framework, called Uncertainty Vertex-weighted Hyper- graph Learning (UVHL), that identified COVID-19 from CAP with CT images with 90% accuracy.
  • Despite significant interest, there’s still confusion about the value of imaging AI. This Blackford Analysis white paper explores the key cost considerations and ROI factors that radiology groups can use to figure out how to make AI valuable for them.


Leave a Reply

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

Subscribe to this comment feed via RSS