Cardiovascular LDCT | Radiology’s Lane | Unnecessary Surveillance

“Health Equity Is Radiology’s Lane”

Geraldine McGinty, MD, MBA, FACR on radiology’s right and responsibility to address health disparities.

Imaging Wire Sponsors

Arterys | Bayer Radiology | Canon Medical Systems | GE Healthcare
Healthcare Administrative Partners | Hitachi Healthcare Americas
Novarad | Nuance | Riverain Technologies | Siemens Healthineers
United Imaging | Zebra Medical Vision

The Imaging Wire


RPI and MGH researchers developed a deep learning model that uses low-dose CTs to accurately predict cardiovascular disease (CVD), providing another example of how AI could expand current population health screening pathways.

  • The Model – The researchers developed the model with over 30k LDCTs from the NLST database, using one CNN to isolate the cardiac region in LDCT scans and a second 3D CNN to detect CVD and automatically quantify mortality risk.
  • The Study – Using a 2k-patient NLST test set, the model outperformed three standard CAC-based screening methods for detecting CVD (0.871 vs. 0.684-0.753 AUCs) and predicting mortality risk (0.768 vs. 0.650-0.698 AUCs). Then with a 335-patient MGH set, the model detected CVD with an impressive 0.924 AUC, which they attributed to MGH’s superior scanners and labeling (compared to the NLST scans).
  • LDCT Screening Implications – Given lung cancer screening participants’ higher CVD risks and considering that the U.S. could soon expand its LDCT screening population by over 80%, algorithms like this could have a major population health impact.
  • Eye on Population Health AI – Population health imaging AI is gaining momentum, with growing research attention on ways to better utilize current screening pathways (LDCT, DEXA, etc.), increased strategic focus from major AI players like Zebra Medical Vision, and the continued shift to value-based care. It might still be early in the population health AI ballgame, but no earlier than many other imaging AI segments.

HAP’s Benchmarking Best Practices

Know how your practice measures up? In this post, Healthcare Administrative Partners details the key benchmarking quality metrics and how they can help radiology practices improve.

– This is sponsored content.

UC Davis’ Whole-Body PET Highlights

See what UC Davis has to say about United Imaging’s EXPLORER Total Body Scanner, the first whole-body PET.

– This is sponsored content.

The Wire

  • The Case Against Routine Lumbar X-Rays: A JAMA study found that routine lumbar spine radiography exams “provide limited value” for pain management decision-making. The researchers examined data from 443 middle-aged UK women over six years, finding that radiographic changes (K-L grade, osteophyte, disc space narrowing) had no association with pain–related disability.
  • Radiologists’ MSSP ACO Participation Way Up: A new JACR study found that the share of Medicare-participating radiologists affiliated with Medicare Shared Savings Program (MSSP) accountable care organizations (ACOs) more than tripled between 2013 and 2018 (10.4% to 34.9%), warning nonparticipating radiologists to prepare for this shift to continue.
  • LucidHealth Expands to Green Bay: LucidHealth acquired Green Bay Radiology (20 radiologists), representing its fourth acquisition in Wisconsin since late 2019 (following Riverside, Madison, RA of Fox Valley). These acquisitions quickly made LucidHealth a leader in Wisconsin, serving 29 hospitals and 100 clinics in the state, and staffing more subspecialized radiologists than any other group in Wisconsin.
  • Unnecessary Surveillance: Many patients treated for head and neck cancers (HNCs) receive far more surveillance imaging exams than recommended. That’s from a new JAMA study (n = 136) that found HNC survivors received an average of 14 imaging studies over a 3.2-year median surveillance period (asymptomatic surveillance is recommended for six months), creating $36.8k in average imaging expenses per patient.
  • CEREB B-Mode Assist: Czech scientists are developing an ultrasound AI technique that would allow primary care physicians to assess patients’ Parkinson’s disease risks without neurology training. The CEREB B-Mode Assist system detects subtle variations in brain ultrasounds, evaluates echogenicities in any predefined brain area, and tracks changes over time.
  • Another Breakthrough Delay: CMS once again postponed the Breakthrough Device payment rule until December 15, which would have automatically granted Medicare reimbursements for “breakthrough devices” as soon they gain FDA approval. CMS’ December revision is expected to include new language requiring products to target Medicare populations and giving CMS continued authority if new risks emerge.
  • United Imaging Considering $1B IPO: United Imaging Healthcare is reportedly considering a $1b IPO in Hong Kong within the next year, following a massive $500m round in 2017. United Imaging was already reportedly working towards a China-based IPO, so this report at least adds evidence that the company is planning to go public.
  • Multi-Modal Image Fusion: Chinese researchers developed a deep learning-based image fusion method that extracts and combines information from multiple modalities (in this case MRI, CT, SPECT) to create clearer images with more quantitative details. The method involved pre-processing images (noise removal, registration, standardization) that are used to train a fusion calculation model.
  • Unnecessary Tests: A Michigan Medicine study found that preoperative testing before low-risk procedures is still very common, even though it’s usually unnecessary. Roughly half of Michigan’s patients who underwent three common surgeries (n = 40k; 2015-2019; gallbladder removal , groin hernia repair, or cancerous breast tissue removal) had at least one preoperative test (including imaging) and 13% had three or more tests.
  • Imaging Hack Case Dismissed: A U.S. District Judge dismissed a class action case against Arizona-based imaging provider, Assured Imaging, finding that the plaintiffs couldn’t show damages from a 2020 cyberattack that exposed patient information. The plaintiffs (former Assured patients) alleged that the hack placed them at higher risk of future identity theft.
  • AI and Radiology’s Data Deluge: An AuntMinnie article is getting some social media buzz from the pro-AI crowd for highlighting how AI will help radiologists keep up with rising imaging volumes. The article shares accounts from a number of radiologists who detail their own workflow constraints, how AI is helping them today, and the AI applications they are most excited about (specifically: triage, NLP, workflow, and augmented intelligence).
  • Clarius V8.0: Clarius released its V8.0 Ultrasound App, highlighted by its ability to automatically detect body anatomy and optimize imaging for that anatomy. The app also adds new features to capture and document photographs, anonymously share ultrasound images on social networks, and new workflow features (e.g. TI-RADS reporting, Lower Extremities Doppler, Labor and Delivery).
  • Reporting Errors in NZ: New Zealand authorities cited a radiologist for reporting errors that led to a woman’s delayed liver cancer diagnosis and eventual death. The radiologists noted two “difficult to characterize” liver lesions in the woman’s CT, but did not provide a differential diagnosis, communicate possible malignancy, or recommend follow-up imaging, despite existing clinical evidence (physical symptoms, high liver enzyme levels, lesions also identified via ultrasound).
  • Fetal Blood Flow Breakthrough: Toronto-based researchers developed a new ultrasound technique that could better predict fetal blood flow problems by measuring umbilical artery (UA) wave reflections. The researchers tested the technique using ultrasounds from 86 pregnant women (26-32 weeks; 40 healthy, 16 w/ fetal circulation problems, 30 w/ maternal placenta circulation problems), finding that the healthy controls had significantly lower UA wave reflections.

Arterys, The CMR Post-Processing Favorite

See why American Hospital of Paris radiologist, PR Olivier Vignaux, says Arterys is his “favorite post-processing tool for cardiac MR,” in this user interview.

– This is sponsored content.

The Resource Wire

– This is sponsored content.

  • With Turbo Suite Excelerate by Siemens Healthineers, you can reduce MRI exam times by up to 50%. See how it’s possible in these videos featuring example hip, knee, and brain scans.
  • See why AI-powered cardiovascular screening’s detection and prevention benefits make it the new frontier of population health.
  • This Hitachi blog details how a complete CVIS solution drives efficiencies and increases productivity.
  • Radiology is among the top five specialties reporting significant burnout. Join The Nuance Challenge to discover how to optimize workflow to improve provider satisfaction and accelerate care delivery.

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!