Legal Transparency | Voice Assisted Guidance | Diagnostic Bias

“If I had a nickel for every time a resident/medical student/non-rad attending or even layperson/family member admonished me for choosing radiology bc ‘computers can already do it,’ my massive student debt would be nearly paid off…”

A post on the auntminnie.com message boards.


Happy 4th of July, Imaging Wire Readers!

Heads-up that this is going to be a one-issue week due to the Independence Day holiday. The Imaging Wire wishes you all a great and safe week (even if you’re not in the U.S.). Thanks for spending each Monday and Thursday morning catching up on the news with us and many thanks to our sponsors who make this publication possible.


Imaging Wire Sponsors

  • Focused Ultrasound Foundation – Accelerating the development and adoption of focused ultrasound.
  • GE Healthcare – Providing point of care ultrasound systems, from pocket-sized to portable consoles, designed to support your clinical needs and grow along with your practice.
  • 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.

The Imaging Wire



Legal Transparency

U.S. Hospitals lost their fight against healthcare cost transparency last week, when a US District Court rejected their claim that the federal government violated the first amendment by forcing hospitals to reveal the rates they negotiate with insurers. With the court decision (and unless the hospital groups’ appeal is successful), healthcare rates will be publicly available by 2021.

  • The Anti-Transparency Fight – A lot of water has passed under the bridge since this story began in mid-2019, when a Trump administration executive order introduced new transparency requirements in an effort to reduce overall healthcare costs. This was concerning enough that hospital groups and insurers even teamed up to fight it, arguing that the rule would create confusion, reduce competition, add administrative burdens, and (somehow) increase healthcare costs.
  • Hospital Impact – Hospitals will have to share healthcare cost info, including gross costs, insurer-negotiated rates, and out-of-pocket costs, while forcing all hospitals to publish a list of 300 “shoppable” services (all posted online).
  • Imaging Impact – Included among the list of 300 “shoppable” services are 13 schedulable imaging procedures (e.g. mammography, abdominal ultrasound, head CT & MRI, others).
  • Insurer Impact – The transparency mandate will require insurers to reveal negotiated rates with hospitals/clinics and create online tools that patients can use to calculate their out-of-pocket rates.
  • Patient Impact – It’s hard to predict how patients will benefit from this rule (a least at first), given that most don’t shop around for medical services. However, some view this as a step towards patients becoming more engaged healthcare consumers.



Voice Assisted Guidance

Anyone who uses a voice assistant device will tell you, its greatest value is the ability to access information without typing into a device and viewing results on screen. That may go for radiology too. A University of Pittsburgh study presented at SIIM20 revealed that Alexa-based programs could help provide diagnostic radiologists with clinical decision support, while allowing them to keep their focus on their patient images.

  • RAD Assistant – The team developed an Alexa Skill app, dubbed RAD Assistant, that’s programmed to listen for pre-defined user inquires and provides audible follow-up recommendations for incidental pulmonary nodule and ovarian cyst finding (tapping into Fleischner Society and Society of Radiologist in Ultrasound recommendations). In theory, this could replace or supplement web-based searches on the PACS workstation.
  • The Study – To validate the application, the team had two English speakers make 154 RAD Assistant inquiries about ovarian cysts and pulmonary nodules using three different devices (Echo Spot, Samsung phone, Apple iPhone). Of the 143 inquiries that were successfully completed, 100% provided correct pulmonary nodule and ovarian cyst recommendations with average respective response turnaround times of 49.5 seconds and 38.5 seconds.
  • The Takeaway – Dictation has long played a role in diagnostic reporting and we’ve seen some recent studies similarly supporting voice assistant guidance for interventional radiology procedures. This study shows that voice assistant solutions could also support diagnostic radiologists without taking up “expensive screen real-estate.” We’ll need more studies and HIPAA compliance before these solutions graduate from SIIM abstracts to clinical use. However, considering the growing role of voice assistant software at home and work, it’s reasonable to expect solutions like RAD Assistant to play a greater clinical role in the future.

The Wire

  • Know Your No-Shows: A United Arab Emirates team explored the impact and causes of imaging no-shows. A review of their own no-shows revealed that X-ray had far less no-shows than more complex/concerning MRI scans (3% vs. 20% hourly avg.), while there are far less no-shows during the middle of the day compared to late at night (8%-19% vs. 36%-63% hourly avg.). The team blamed these no-shows on a combination of patient behavior (urgency, anxiety, etc.), cost concerns, environmental factors (weather, traffic, etc.), and scheduling (appointment set too long ago, inconvenient timing, etc.).
  • New DiA Apps: DiA Imaging Analysis announced the FDA and CE approval of its LVivo RV (assesses right ventricle dysfunction) and LVivo Bladder (automated bladder volume analysis) AI-based ultrasound applications. The solutions join DiA’s LVivo Toolbox, which now has six FDA/CE-approved ultrasound AI solutions.
  • Diagnostic Bias: A new study in RadioGraphics detailed the unconscious biases that could lead to missed breast cancers and misinterpreted breast lesions. The following cognitive processes lead to biased interpretations: 1) satisfaction of search (missing 2nd abnormality due to satisfaction of finding first); 2) Inattention blindness (missing abnormalities in areas that generally have fewer issues); 3) Hindsight (skewed confidence due review of previous studies); 4) Anchoring (over-relying on early findings throughout diagnostic process); 5) Premature closing (completing diagnosis before all information is reviewed); and 6) Satisfaction of reporting (over-relying on prior reports).
  • DBT’s Radiation Advantage: DBT can now add “reduced radiation exposure” to its list of advantages over full‐field digital mammography (FFDM). A team of Kuwaiti researchers scanned 200 women, finding that DBT had significantly lower median entrance surface dose (3.1-8.9 vs. 3.3- 9.1 mGy, depending on thickness) and lower median average glandular dose (median: 1.8-4.0 vs. 3.3-6.0 mGy, depending on thickness) than FFDM.
  • EHR Decision Support Works: An Einstein Healthcare study found that EHR-based clinical decision support prompts can help improve appropriate imaging rates. The researchers reviewed Einstein’s pediatric head trauma ER visits before and a year after launching the CDS feature, finding that CDS improved its ratio of “clinically necessary” pediatric brain CT orders from 56.8% to 76.2% of all scans.
  • CEUS Cancer Treatment Predictor: A Dallas and Philadelphia-based team developed software that uses contrast-enhanced ultrasound images to predict the effectiveness of transarterial chemoembolization (TACE) liver cancer treatment. The software combines image enhancement and machine learning to quantify morphologic liver cancer features (e.g. number of vessels, number of bifurcations, vessel to tissue ratio), and predicted long-term TACE response with 86% accuracy, 89% sensitivity, and 82% specificity when reviewing CEUS images from 36 patients prior to TACE treatment.
  • More AI Funding: After an AI funding slowdown, last week brought two more AI funding events. Taiwanese startup Deep01 raised $2.7m that it will use to fund the global expansion of its FDA-cleared software for locating intracranial hemorrhage in non-contrast CTs and expand its solutions portfolio. NLP startup Agamon completed a $3m seed round that it will use to commercialize and clinically-expand its software used to train hospital computers to “read” radiology reports.
  • AI TKR Predictor: A New York-based team developed a DL model that can predict knee osteoarthritis progression and likelihood of requiring total knee replacement (TKR) using knee radiographs. The researchers evaluated 728 patients’ knee X-rays (324 who eventually received TKRs), predicting whether patients would have TKR surgery within nine years with a 0.87 AUC (vs. 0.74 w/ traditional TKR prediction model).
  • Vuno’s Big CE Approval: South Korean AI developer, Vuno, announced the CE Mark approval of five of its AI solutions, allowing their use across 27 EU countries and the many global countries that recognize CE marking. The solutions include VUNO Med-BoneAge (assesses bone age based on left hand X-ray), VUNO Med-DeepBrain (segments/quantifies brain regions w/ MRI data), VUNO Med-Chest X-Ray (enhances screening for common thoracic diseases), VUNO Med-LungCT AI (detects/quantifies early pulmonary nodules), and VUNO Med-Fundus AI (detects/classifies ophthalmology lesions).
  • PA Follow Findings: A new UPenn study revealed that Pennsylvania’s law requiring imaging providers to notify patients about abnormal imaging results had a “small impact” on their institution and emphasized the importance of timely follow-ups and the benefits of involving ordering physicians. The researchers reviewed cases flagged for patient notification over a 1-month period (n = 235), finding that discussions took place with 87% of these cases (6 day avg. after exams), and follow-up care was provided in 74% of these findings (31.3 day avg. after exams).
  • COVID-19 Repository: RSNA announced plans to build a COVID-19 imaging data repository to support future research and educational efforts. The COVID-19 Open Radiology Database (RICORD) will feature annotated imaging data with supporting clinical information and will be assembled by “an international task force of scientists and radiologists.”
  • Exercise 4D flow MRI: New research from the University of Wisconsin found that cardiovascular 4D flow MRI can effectively measure blood flow during strenuous exercise, but not kinetic energy. The prospective study performed cardiac 4D flow MRI on 11 healthy adults (26yr avg., 9 completed study) at rest and during strenuous exercise, measuring blood flow in the ascending aorta (AAo) and main pulmonary artery (MPA) and quantifying kinetic energy (KE) in the left and right ventricle. A review found flow measurements were consistent (avg. differences: 10% at rest, 16% during exercise) and were highly repeatable (interobserver variability never >6%), while KE measurements had much wider differences (30%) and poor interobserver correlation.

The Resource Wire

– This is sponsored content.

  • It says a lot when a solution works so well for a radiology department that they decide to perform a study to quantify its benefits. In this Imaging Wire Q&A, University Hospital of Zurich’s Thomas Frauenfelder discusses his experience with Riverain Technologies ClearRead CT and his study on its effectiveness.
  • Rural hospitals have unique needs and most know that bigger doesn’t always mean better when it comes to healthcare. This Hitachi blog details why its combination of the right features, ROI, service levels, and philosophy make it the right partner for rural hospitals.
  • The GE Healthcare Venue Go features a uniquely adaptable design, a simple interface, and streamlined probe layout so you can go through your day quickly, efficiently, confidently.
  • Nuance’s latest blog details how the speed of radiology’s post-COVID road to recovery will depend on how the specialty accelerates its digital transformation, including its adoption of AI, NLP, and follow up tools.

Leave a Reply

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

Subscribe to this comment feed via RSS

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

Join thousands of imaging professionals.