fbpx

Rad Blindness | Philips Prevails | CO-RADS Works

“. . . we need to be doing a lot more testing,”

Google Brain scientist, Alex D’Amour, on the underspecification problem that’s creating AI’s real world challenges.



Imaging Wire Sponsors

  • Arterys – Transforming medical imaging by reducing the tedium of interpretation and empowering physicians with our web-based AI platform.
  • 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


Inattentional Blindness

New Research out of the University of Utah showed that inattentional blindness (failure to spot obvious, but unexpected objects) exists across all radiologist experience levels, and suggested some ways to address this challenge.

  • The Study – Fifty radiologists evaluated seven CT scans for lung cancer. The seventh case included a large (9.1 cm) breast mass and lymphadenopathy.
  • The Result – Because of their focus on lung nodules, 66% of the radiologists did not detect the breast cancer and 30% missed the lymphadenopathy.
  • The Follow Up Study – When another 30 radiologists reviewed the same image set for a broader range of abnormalities, only 3% missed the breast mass and 10% missed the lymphadenopathy.
  • Equal Blindness – The researchers were unable to predict who would miss or spot the unexpected abnormalities based on radiologist experience, accuracy performing the primary tasks, or search behavior. In other words, even the most experienced / skilled radiologists are prone to inattentional blindness.
  • The Takeaway – Since inattentional blindness is a “universal cognitive truth,” this study suggests that new processes / techniques with a balanced focus on the expected and unexpected would be required to catch these abnormalities. To address this, they suggested performing a general assessment of a scan before looking for specific abnormalities or using a checklist to spot commonly missed findings. They also warned that AI (also mainly narrow) wouldn’t necessarily help solve this problem.

The Wire

  • Reprioritization Works: Head CT intracranial hemorrhage (ICH) worklist prioritization algorithms improve wait times. That’s from a new University of Texas Southwestern Medical Center study that reviewed the impact of their ICH prioritization tool from September 2017 to March 2019, finding that the tool’s reprioritization feature reduced wait times for both negative (15.45 minutes) and positive scans (12.02 minutes). However, the tool’s pop-up widget and flagging features did not improve wait or turnaround times.
  • KA Imaging’s Reveal Subscription: KA Imaging will make its Reveal 35C dual-energy X-ray detector available through a subscription model (includes: detector, service, updates, 5yr upgrades), giving hospitals a flexible way to adopt the unique portable detector during the COVID-19 pandemic. The Reveal 35C’s COVID advantages go beyond its new cost flexibility, as it brings dual-energy imaging to patient bedsides (more sensitive than standard X-ray, more portable / economical than CT) and it enables bone and soft tissue differentiation without motion artifacts.
  • Adult Appendicitis CT, Still Rising: Although pediatric CT use for abdominal pain evaluations declined in U.S. emergency departments (likely due to an ultrasound-first strategy for suspected appendicitis), emergency CT use on adults with abdominal pain has continued to rise. An AJR study found that emergency CT use for adults with abdominal pain increased from 3.9% in 1997 to 37.8% in 2016, while pediatric CT use increased from 1.2% in 1997 to 16.6% in 2010 and then declined slightly through 2016.
  • Butterfly Goes Public: Butterfly Network is headed to the NYSE after a merger with PE-backed “blank-check company” Longview Acquisition Corp. The merger valued Butterfly at $1.5b and will give the company another $584m to drive healthcare adoption and fund product development. Considering that Butterfly raised $370m before now, this move gives it a lot more funding to work with.
  • Automating GGO Volumes: New research out of France detailed an automated lung volume measurement technique for ground-glass opacities that could help determine COVID-19 severity from CT scans and predict 7-day outcomes. The researchers developed the automated algorithm based on voxels X-Ray attenuation to measure abnormal lung parenchyma volume and compared it to conventional semi-automatic software using CT scans from 37 COVID-19 patients. The study revealed that the automated software performed similarly to the conventional semi-automated approach (but in 15 minutes vs. 93) and both achieved statistically significant association between automated lung volume measurements and maximum clinical severity scores.
  • GE’s CT Step Forward: GE Healthcare took a “substantial step forward” with its acquisition of Swedish photon counting CT detector (PCCT) startup, Prismatic Sensors. Prismatic Sensors’ Deep Silicon detector technology could significantly improve CT image quality and lower radiation dosage for a range of applications including oncology, cardiology, and neurology. The acquisition comes three years after GE Healthcare first bought a stake in Prismatic Sensors and represents a key step in the company’s plan to bring PCCT detectors with Deep Silicon to the market “in the near future.”
  • Geriatric POCUS Improvements: A UPenn study detailed how a brief handheld ultrasound training course significantly improved four geriatric medicine fellows’ POCUS confidence and capabilities. An initial 4-hour training workshop doubled the fellows’ overall comfort / confidence using a Butterfly Ultrasound device, which almost tripled following an extended curriculum that included an hour of supervised scanning each week.
  • Subtle’s $12m: Subtle Medical wrapped up a $12.2m Series A round (increasing its total to $13.3m) that it will use to expand its imaging enhancement portfolio. Subtle Medical is more established than many AI players at the time of their Series A round, with an established name in the image reconstruction space, CE and FDA approvals, and use at over 50 sites.
  • CO-RADS Evidence: New research out of the Netherlands confirmed the benefits of using the CO-RADS CT reporting data system, finding that the 1-5 scoring system allows fast and reliable COVID-19 diagnosis (especially among patients with ≥48hrs of symptoms). The study of 1,070 patients who arrived at five emergency departments with suspected COVID-19 pneumonia (536 CV19-positive), found that CO-RADS identified the COVID-positive patients with a 0.87 AUC (vs. RT-PCR tests) and patients with ≥4 CO-RADS scores had a 25.9 odds ratio for testing positive for COVID-19.
  • Lacking LCS: Last week brought two more studies detailing the U.S.’ low and uneven lung cancer screening rates. A JAMA review of 15 cohort studies (n = 16,863) revealed a 55% overall lung cancer screening adherence rate among high risk individuals in the U.S., with lower adherence rates in the 50-64yr age group, among current smokers, and minority and non-college educated individuals. The American Lung Association’s ‘State of Lung Cancer’ Report found that just 5.7% of the 8 million people in the U.S. who are eligible for LCS were actually screened in 2019, while the best LCS rates were in the Northeast (MA 18.5%, VT 13.8%, NH 12.1%) and the worst rates were in the Southwest (NV 1%, CA 1.2%, NM 1.6%).
  • ACR Calls for Action: The ACR urged its members to advocate in favor of the “Holding Providers Harmless from Medicare Cuts During COVID-19 Act of 2020” (H.R. 8702), which would halt the 2021 evaluation and management (E/M) Medicare Physician Fee Payment Schedule changes, that some say will reduce radiologist reimbursements by 11%. So far ACR members sent 1,914 letters to Congress in support of H.R. 8702, representing only 4% of ACR radiologists and the lowest participation rate the ACR has ever experienced.

The Resource Wire

– This is sponsored content.

  • In its latest Q&A, Nuance sat down with Nines CEO David Stavens, PhD, and medical director Dr. Jean-Paul Dym, MD, to discuss their experiences with PowerScribe One and their perspectives as a Nuance innovation partner.
  • Let Siemens Healthineers’ AI-powered assistants provide automatic post-processing and automation of routine workflows and tasks – so that you can focus on more critical issues.
  • This UCSD study shows that fully automated ventricular segmentation using Arterys Cardio AI provides contours and volumes that can be used to aid expert segmentation.
  • GE Healthcare shares why it is so excited about acquiring Prismatic Sensors AB, suggesting that its newly acquired Deep Silicon detectors technology could give it the depth and purity to set a new standard for CT Imaging.
  • They say that in times of crisis, you get to know who your real friends and partners are. This Q&A session details how Healthcare Administrative Partners stepped up to guide their client Triad Radiology Associates through the challenges presented by the COVID-19 pandemic.
  • This Riverain Technologies case study details how Duke University Medical Center integrated ClearRead CT into its chest CT workflows, reducing read times by 26% and improving nodule detection by 29%.

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

Join thousands of imaging professionals.