Hold the CXR

“I’m it.”

Rush County Memorial Hospital technologist, Eric Lewallen, after a COVID outbreak made him the only person at his Kansas hospital who could perform X-rays for over a week.




Imaging Wire Sponsors

  • Arterys – Reinventing imaging so you can practice better and faster.
  • Bayer Radiology – Providing a portfolio of radiology products, solutions, and services that enable radiologists to get the clear answers they need.
  • GE Healthcare – Enabling clinicians to make faster, more informed decisions through intelligent devices, data analytics, applications and services.
  • Healthcare Administrative Partners – Empowering radiology groups through expert revenue cycle management, clinical analytics, practice support, and specialized coding.
  • Hitachi Healthcare Americas – Delivering best in class medical imaging technologies and value-based reporting.
  • Nuance – AI and cloud-powered technology solutions to help radiologists stay focused, move quickly, and work smarter.
  • Riverain Technologies – Offering artificial intelligence tools dedicated to the early, efficient detection of lung disease.
  • Siemens Healthineers – Shaping the digital transformation of imaging to improve patient care.



The Imaging Wire


Hold the CXR

Despite our continued focus on determining how well CXR (or AI-aided CXR) diagnoses COVID-19 or predicts COVID outcomes, a new RSNA Radiology study confirmed that CXR screening is only necessary for patients with moderate to severe symptoms and high risk of deterioration.

  • The Study – Researchers out of Singapore reviewed 5,621 patients (17 to 60yrs, all CV19-positive) who were tested and treated for COVID-19 last spring, including 1,964 patients who received chest X-rays.
  • The Results – Of the 1,964 patients who received CXRs, 1,924 (98%) had normal scans and only 39 (2%) had signs of pulmonary infection in their CXRs. Most notably, only four patients (0.2%) had signs of pulmonary infection on their scans and later required supplemental oxygen and inpatient treatment (and all four already had symptoms at the time of their scans).
  • The Takeaway – This study seems to confirm that the Fleischner Society’s April recommendations had it right. CXR screening should be reserved for patients at risk of clinical deterioration, particularly if they are young or middle-aged with no or minimal symptoms.

The Wire

  • Rad Roles 2020: A radiologist survey published in Insights Imaging (n = 1,344 European rads, Jan-March) provided new insights into the role of radiologists in 2020. Most press coverage of this study focused on a revelation that 92% feel that sub-specialty expertise is important and improved radiologists’ clinical visibility. However, the survey also provided insights into non-radiology clinical training (80% view it as mandatory), patient communication (76% regularly communicate w/ patients, 25% received comms training, 82% feel comms training would help), and patient visibility (62% see low visibility as a risk to the specialty).
  • US Radiology Continues Expansion: US Radiology Specialists continued its recent expansion, acquiring Alabama’s ImageSouth (5 imaging centers, Birmingham’s largest imaging provider) just a few weeks after USRS expanded to Western NY through another acquisition. Since its creation through Charlotte Radiology’s private equity acquisition less than three years ago, US Radiology’s aggressive M&A pace has expanded the once-regional practice to ~150 outpatient imaging centers across 14 states.
  • CMS’s PA API: CMS proposed a new set of rules intended to streamline the prior authorization process, requiring payers in Medicaid, CHIP, and QHP programs to build APIs to support data exchange and prior authorization using HL7 and FHIR standards. These programs would allow providers to know each payer’s documentation requirements in advance and enable providers to send prior authorization requests and receive responses electronically. It would also require payers to make prior authorization decisions within 72 hours and explain any denials.
  • Lunit LC Excellence: A new study detailed the Lunit Insight CXR algorithm’s “excellent” performance detecting lung cancers in CXRs. The Radiology: Cardiothoracic Imaging study reviewed 218 CXRs (168 patients w/ previously undetected lung cancers, 50 patients w/ normal scans), and had four thoracic radiologists reevaluate the CXRs with and without the algorithm. Lunit Insight CXR outperformed the radiologists in classifying the CXRs (AUROC: 0.899 vs. 0.634–0.663) and localizing the lesions (AUFROC 0.744 vs. 0.619–0.651), while achieving higher sensitivity (69.6% vs. 47.0%) and specificity (94.0% vs. 78.0%). When the radiologists used the algorithm, their AUROC (0.685–0.724 vs. 0.634–0.663) and pooled AUFROC (0.636 vs 0.688) both substantially improved.
  • Lunit’s COVID Study: Lunit INSIGHT CXR continued its research momentum, as a recent PLOS ONE study highlighted the algorithm’s ability to detect COVID-19 in chest X-rays as part of emergency department triage workflows. The study used data from 279 COVID-positive patients (182 w/ CV pneumonia, 97 without pneumonia, from 6 centers), finding that INSIGHT CXR achieved 95.6% sensitivity and 88.7% specificity and detected COVID with similar accuracy as the centers’ radiology reports.
  • GE’s U.S. One-Stop Breast Clinic: GE Healthcare, Premier Applied Sciences, and St. Luke’s University Health Network (Pennsylvania) will pilot the U.S.’s first rapid diagnostic breast cancer center, coming a year after GE and Premier announced plans to expand their “one-stop” center model stateside. Using a coordinated and multi-modality approach (w/ training and materials from Premier), the rapid center will guide patients through diagnosis and treatment planning in one location, with one team, and within 36 hours (vs. up to 26 days).
  • Virtual Surgery Surge: The Daily Telegraph detailed the Proximie virtual surgery platform’s massive growth in 2020 (it allows remote specialists to guide on-site surgeons), which included an eightfold increase in users (to 2,500 users at 300 hospitals) and a fourfold increase in usage (6,000 operations). This might not specifically be an imaging story, but diagnostic and live medical imaging play a major role on the Proximie platform, so it might be worth paying attention to.
  • IR Rebound: An Omdia Healthcare report published on Auntminnie.com forecast that the interventional radiology market will bounce back in 2021 after COVID drove a 15% decline this year (it was originally slated for +5% growth). In fact, the firm forecasts IR spending to maintain an >5% compound annual growth rate over next five years, due in part to demand for minimally invasive and interventional cardiology procedures, and the continued adoption of hybrid ORs.
  • UV CT Cleaning: A new paper out of Johns Hopkins detailed how to use ultraviolet germicidal irradiation to decontaminate a CT gantry, helping to reduce transmission and cleaning times during the COVID emergency. Using a UV lamp, the team applied 350 to 580.9 μW/cm2 of UV (highest levels in the middle of the gantry) within 20cm of the scan plane. The team achieved at least “6‐log kill” of viral RNA (99.9999% eliminated) within 2 minutes and suggested that running the UV light for 3 minutes or 5 minutes would be even more effective.
  • fMRI AI Generalization Breakthrough: A team of Japanese researchers developed a machine learning technique that can to help diagnose major depressive disorder (MDD) using fMRI brain imaging, while overcoming the generalizability challenges that historically hindered fMRI MDD AI models (different sites produce very different fMRI images). Using fMRI scans from 713 participants and four imaging sites (149 w/ MDD), the team removed the fMRI variations using a new harmonization method, and then used that data to develop a machine learning classifier to identify MDD’s imaging biomarkers. Their model correctly identified patients with MDD with 70% accuracy against an independent fMRI set (521 participants, 5 sites).
  • Pandemic Impact: A new paper out of the Henry Ford Health System shared another account of the COVID-19 pandemic’s impact on radiologist workloads. Their review of a large subspecialized practice’s RVUs during H1 2019 and H1 2020 revealed a 51.5% overall decline during the COVID emergency, largely driven by breast imaging (-76.9%), musculoskeletal imaging (-75.3%), and neuroradiology (-67.5%).
  • TBShoNet: An MIT-led team earned attention at RSNA for their new TBShoNet deep learning tuberculosis detection model that analyzes smartphone photos of chest X-rays, and showing how imaging AI could help in low-resource regions. In a test using 662 CXR photos captured by five different smartphones (336 w/ TB), TBShoNet detected TB with an 0.89 AUC and achieved up to 81% sensitivity and 84% specificity.
  • The First 40 Walgreens Clinics: Walgreens’ primary care clinic aspirations got more tangible last week after the drugstore giant announced that the first 40 of its 500-700 planned on-site clinics will open by Q3 2021. These initial clinics will be in Houston (where some clinics are already open), Phoenix, El Paso and Orlando, while Walgreens plans to operate clinics in 30 American cities within 5 years. It appears that at least some of these sites will have on premise imaging, and all of the clinics (along with clinics from Walmart and others) could be part of a substantial shift in where/how health services are delivered.
  • DeepCOVID-XR: A Northwestern University team developed an AI algorithm that can detect COVID-19 on chest X-rays with similar performance as experienced thoracic radiologists. Their DeepCOVID-XR algorithm classified 2,214 test images (1,194 CV19-positive) with 83% accuracy and a 0.90 AUC, and then achieved 82% accuracy (AUC 0.88) on 300 random test images (134 CV19-positive) that were also read by five radiologists (consensus accuracy 81%, AUC 0.85).

The Resource Wire

– This is sponsored content.

  • This Hitachi blog post and associated white paper detail how multi-parametric MRI (mpMRI) can reduce unnecessary prostate biopsies by more than one third and increase disease detection by more than 50%.
  • These two independent Phase III trials provided new evidence supporting the accuracy and effectiveness of using Bayer’s Gadavist contrast agent for cardiac MRI.
  • In this quick video, Einstein Healthcare Network’s Peter Wang, MD describes how Nuance’s embedded clinical guidance delivers streamlined radiology workflow and improves report quality.
  • Register today for this webinar from Healthcare Administrative Partners where they will examine how the 2021 Medicare Physician Fee Schedule (MPFS) and Quality Payment Program (QPP) final rules will impact physicians, with a focus on radiologists.
  • The advancements of AI and 4D flow can shorten exam and segmentation times, speed up diagnosis and make clinical care more data-driven, intelligent, and patient-focused. Learn more in Arterys’ Transforming Cardiac MR whitepaper.
  • Learn how rapid breast cancer diagnosis clinics are reducing diagnoses-to-consultation turnarounds “from weeks to hours,” and how they are helping patients and clinicians in this new GE Healthcare Insight.

Get every issue of The Imaging Wire, delivered right to your inbox.

Join thousands of imaging professionals.