“I Have Seen My Death”
Anna Bertha Ludwig’s response after her husband, William Roentgen, performed the world’s first X-ray on her hand 125 years ago this week.
Happy Holidays Everyone!
This is the final issue of the year, so keep an eye out for your next one on January 4th. Huge thanks to all of our readers and sponsors who make this newsletter possible. Wishing all of you an excellent and safe holiday season.
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
CXR + EMR COVID Predictions
A Mount Sinai team developed a deep learning model that uses CXR scores and “clinical variables” collected during their ED admissions (and pulled from EMRs) to accurately predict COVID-19 outcomes. Here are some details:
- The Study – The CXR + EMR model was trained with data from 338 COVID-positive patients (CXR scores, clinical variables, 30-day outcomes) and then tested with data from 161 other COVID-positive patients.
- The Results – The CXR + EMR model predicted intubation (0.86 AUC) and death (0.82) with far greater accuracy than the CXR-based model (0.66 intubation, 0.59 death) and the clinical variables-based model (0.64 intubation, 0.59 death).
- The Takeaway – This study suggests that AI models that combine imaging and clinical data can better predict COVID outcomes than models that only use imaging or clinical data. That makes a lot of sense, and its part of a recent wave of AI research focused on combining images with clinical data (there are a few more in this issue and one earlier this week) to produce more holistic assessments than just using images.
The Wire
- “A Hammer in Search of a Nail”: That’s how a recent Radiology Journal editorial described the wave of imaging AI models developed to diagnose/segment/distinguish COVID-19, suggesting that these efforts have been largely unproductive and unnecessary. Instead, the editorial called for more efforts to develop COVID-19 AI models that combine imaging, laboratory, and clinical information to provide more actionable insights. The folks behind the Mount Sinai study at the top of today’s issue might agree.
- Imaging M&A: Imaging companies are well represented among FierceBiotech’s Top-10 Medtech M&A targets, with Caption Health (AI-guided ultrasound), Hyperfine Research (portable MRI), Butterfly Network (handheld POCUS), and Zebra Medical Vision (imaging AI) all making its list. Although it’s unclear how much journalistic rigor went into this list, FierceBiotech cited a Jeffries analyst who suggested that just about all of these companies are evaluating offers.
- DBT’s Survivor Screening Benefits: A new MGH study detailed the advantages of using DBT to screen breast cancer survivors. The researchers reviewed screening mammograms from 8,170 women (~9k DBT exams, ~23k DM exams), finding that BC survivors screened with DBT had fewer abnormal interpretations (5.8% vs. 6.2%) and higher specificity (95% vs. 94.7%) compared to DM, while achieving similar cancer detection rates (8.3 per 1k exams vs. 10.6 /1k) and screening-detected invasive cancers rates (74% vs. 72%). This might not be surprising given DBT’s research momentum, but there wasn’t much data on using DBT to screen survivors before now.
- Canon SPEEDER’s FDA: Canon Medical’s Compressed SPEEDER technology just gained FDA approval for 3D sequences with Canon’s Vantage Orian 1.5T MR system, reducing scan times during 3D MR exams by reconstructing full resolution images from under-sampled data. Compressed SPEEDER is available as an option with Canon’s M-Power software.
- LUS for Neonatal Pneumothorax: Chest X-ray might be the standard for diagnosing neonatal pneumothorax, but a new study review out of China suggests that lung ultrasound is a better option. The review of eight studies (529 newborns) found that LUS diagnosed neonatal pneumothorax with higher specificity (98% vs. 96%), sensitivity (99% vs. 82%), and diagnostic odds ratio (920.01 vs. 44.54) than CXR. Plus, LUS was faster in five of the eight studies and it doesn’t expose the babies to radiation.
- HAP & Edison Partner: Edison Radiology Group selected Healthcare Administrative Partners to be its full-service radiology revenue cycle management (RCM) provider, including billing, coding, carrier credentialing, and MIPS Measure Assurance Services. The New Jersey-based radiology practice (three locations, ~16 radiologists, in business 30-years) highlighted HAP’s organizational fit and responsiveness in the announcement.
- RSNA’s COVID Dataset: RSNA published the first dataset from its RSNA International COVID-19 Open Radiology Database (RICORD). The initial RICORD dataset includes 120 annotated COVID-positive chest CTs from four international sites and is available to global research/education communities for AI development. RSNA will continue to expand its RICORD database with the goal of creating the world’s largest open database of anonymized COVID-19 medical images.
- V7 Labs’ $3m: V7 Labs completed a $3m seed round that it will use to expand its AI training automation platform. V7 Labs’ SaaS/cloud/browser-based computer vision platform helps automate AI developers’ training data workflows (reportedly 10-100x faster) with tools to build automated image/video data pipelines, organize and version data sets, and train and deploy AI models.
- Screening Guidelines’ False Negatives: A new MIT-led study (n = 3.9m women) suggests that the debate over what age to start breast cancer screening is “missing a key point,” because women who decide to start screening at 40-years are less likely to have cancer than same-aged women who don’t participate in screenings (and therefore don’t contribute data). The researchers suggest that if an equal number of women from these two groups each started screening at 45-years, the women who don’t currently follow screening guidelines would represent three quarters of all breast cancer deaths. Because of that, the study suggests that future screening guidelines should balance age-based factors with risk-based factors.
- Medicare Cuts, Cut: Radiology’s looming Medicare reimbursement cuts might not be as bad as expected, as the current version of 2021 Consolidated Appropriations Act would scale-back the reductions to ~4% (vs. 10%) in 2021 with other adjustments taking place in the future. This last minute reprieve is made possible by a $3b injection into the Medicare Physician Fee Schedule (and months of lobbying), increasing Medicare payments in light of COVID’s impact on healthcare providers. This looked like a done deal on Tuesday, but as of Wednesday night Trump is threatening to veto the bill for a number of reasons that have nothing to do with radiology.
- Surprise Billing, Legislated: After years of debates and stalemates, it looks like we now have legislation addressing surprise medical billing. The 2021 Consolidated Appropriations Act (the same one that softened the above radiologist payment cuts) included new surprise billing rules that will protect patients from receiving balance bills for emergency care, most out-of-network care at in-network facilities (e.g. radiology), and air ambulances (all starting in 2022). The final structure finds a middle ground between providers and insurers, as it starts with 30 days of negotiations that will use in-network rates and exclude lower Medicare/Medicaid rates (providers’ legislative preference), before moving on to arbitration that will not take into account providers’ “billed charges” (insurers’ preference). This looked like a done deal on Tuesday, but as of Wednesday night Trump is threatening to veto the bill for a number of reasons that have nothing to do with radiology.
- GE & ENDRA Renew: ENDRA Life Sciences renewed its collaboration agreement with GE Healthcare through the end of 2023 (it started in 2016) as ENDRA works towards commercializing its Thermo Acoustic Enhanced UltraSound (TAEUS) technology. As part of the agreement, GE Healthcare will introduce ENDRA’s TAEUS fatty liver application to its clinical ultrasound customers in return for rights to manufacture and license the technology.
- The CXR-LC Model: An MGH-led team developed a CXR-based deep learning system that might be able to identify more high-risk smokers who should participate in LDCT lung cancer screening. The team developed their CXR-LC model using data from the PLCO trial (n = 41,856; CXR, age, sex, smoking status) and then validated it on a pair of datasets featuring 5,615 (12-year follow-up) and 5,493 smokers (6-year follow-up). CXR-LC identified patients who would develop lung cancer better than current CMS eligibility system (0.755 vs. 0.634 AUC) and missed 30.7% fewer lung cancers.
- IR’s Safety Commitment: A Boston University duo called for better efforts to understand interventional radiology’s adverse events / complications and make a “recommitment to patient safety.” The authors called for widespread cultural changes (blame-free error reporting, prevention education, better understanding of errors), suggesting that this will lead to improved safety practices.
The Resource Wire
– This is sponsored content.
- Discover the biggest trends driving CT imaging today and tomorrow in this Siemens Healthineers overview.
- Learn how Arterys’ Cardio AI can improve your Cardiac MRI workflow and cut LV/RV function segmentation by up to 25 minutes.
- This Bayer Radiology white paper details how the right people, plan, and systems can help imaging teams achieve their dose management goals.
- With orthopedic care growing with the aging population, orthopedic imaging is growing along with it. This Hitachi blog details how more orthopedic practices are bringing imaging in-house and what factors they should take into account as they decide how/whether to add imaging to their suite of services.
- Ever wonder how secure your ultrasound systems are? This GE Healthcare Insight details how outdated operating systems might make many ultrasound systems more vulnerable than other devices and outlined the steps organizations can take to keep their ultrasounds protected.
- In this Nuance video, Penn Medicine professor, Warren B. Gefter, shares how PowerScribe One leverages AI, structured data, and automation to drive improved patient care.
- CPT updates, E/M services changes, CDS, and MIPS are just some of the topics covered in Healthcare Administrative Partners’ upcoming 2021 MPFS Updates & Radiology Reimbursement Impact webinar. Register today!
- 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%.
