Ongoing DBT | Auto Selecting AI | Radiation Risks


“In other words, we needed to develop an algorithm to develop an algorithm.”

A Twitter comment from Microsoft Research engineer, Neil Tenenholtz, about a new algorithm that streamlines image series collection — and it’s also about how we need to standardize series identifiers.



This is the last week to submit nominations for the 2020 Imaging Wire Awards. If you know someone who made outstanding contributions to radiology, nominate them today.



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



The Case for Ongoing DBT

The pile of evidence supporting DBT got a bit higher this week with a new study that found DBT catches more cancers than digital mammography through ongoing screenings. Here are some details:

  • The Study – The Italy-based research team performed initial baseline screenings on 32,870 women using DBT plus synthetic mammography (DBT+SM). Two years later, they screened roughly half of the women with DBT+SM and the other half with digital mammography (DM). The researchers compared the two groups, plus a control group (n = 28,680) that was exclusively scanned with DM for both their initial and repeat screenings.
  • The Results – Women who were screened with DBT+SM during both rounds had higher cancer detection rates than those who shifted to DM for their two-year screenings (8.1 per 1k vs. 4.5 per 1k), while both groups had higher CDRs than the DM control group (3.5 per 1k). The two-year screenings found a higher proportion of advanced cancers (≥ Stage II) among the women screened with DBT+SM than the women screened with DM (14.5% vs. 8.5%), although the DM control group had far higher advanced cancer rates (27.3%).
  • The Takeaway – There’s no shortage of pro-DBT studies these days, but this is prospective (most have been retrospective), includes outcomes from ongoing screening (most focus on initial screenings), and suggests that DBT catches more early stage invasive cancers and DM. That makes this study pretty notable.

The Wire

  • Auto Selecting for AI: An MGH and BWH team detailed a new technique that uses DICOM metadata to automate selecting relevant MRI image series for algorithm development. The approach produced high generalizability and efficiency (less than 0.4 ms/series) and achieved 97.4% – 99.96% accuracy with a pair of 20k-image brain MRI datasets. This study has some buzz going for it, as it solves a real AI problem, suggests similar techniques could work for other modalities, and is getting its share of online praise. On the other hand, some might suggest that this study is really a sign that radiology needs to standardize series identifiers.
  • Ambra’s New ProViewer: Ambra Health got a jump-start on the RSNA 2020 launch season, announcing its Ambra ProViewer cloud-based diagnostic image viewer (FDA-cleared). Ambra highlighted its new viewer’s image management features (diagnostic and photos/videos), web-based sharing and mobile / home access capabilities, ease-of-deployment, customizability, and advanced visualization and analysis tools.
  • No Satisfaction: A new Brigham and Women’s Hospital study found that its radiologists’ overall burnout levels were no worse than their non-rad colleagues, but the rads aren’t particularly satisfied either. The study surveyed faculty from 15 BWH clinical departments (questions on 0-4 scale, burnout = ≥2), revealing that its radiologists were less likely to find their work meaningful (-2.77351), feel happy (-2.669) or valued (-2.5486), or believe they are treated with respect and dignity by leadership (-3.44149).
  • NTAP on Wired: Wired.com recently profiled CMS’s decision to reimburse providers when they use Viz.ai’s ContaCT to triage suspected strokes and Digital Diagnostics’ IDx-DR for diabetic retinopathy. These CMS NTAPs might seem like old news to you, but it’s new news to Wired’s millions of tech-savvy readers who reside outside of the imaging AI bubble, and coverage like this has a way of helping niche policies become more mainstream.
  • AI’s Slow Start in the UAE: A new survey out of the United Arab Emirates revealed that the country’s radiologists and radiographers (n = 34 & 119) are still figuring out imaging AI. Although 32% percent are “excited” about imaging AI, 85.6% didn’t believe that AI will play an important role in radiology, 40% weren’t familiar with imaging AI technology, and 74.5% have never studied AI. Although AI awareness surely varies across regions, this is a solid reminder of how early we are in the global imaging AI ballgame.
  • RapidAI & Penumbra’s PE Partnership: RapidAI and Penumbra launched a partnership to develop pulmonary embolism diagnosis solutions. The companies will collaborate on several clinical and communications modules, using RapidAI’s AI tech to streamline PE triage and decision-making based on analysis of CT scans, while Penumbra’s PE products would presumably be used to treat these patients.
  • Radiation’s TGCT Risk: A new study out of UPenn found that men exposed to diagnostic radiation in the lower abdomen and pelvic regions have a 59% greater risk of developing testicular cancer. The study of 315 men with testicular germ cell tumors (TGCT) and 931 men without TGCT, found that the risk of TGCT was far greater among men who had at least three abdominal /pelvic x-ray or CT scans (1.78 odds ratio) or lower GI series or barium enema (4.58 odds ratio). Men who were first exposed to diagnostic radiation before the age of 11 also face higher TGCT risks than those exposed at 18 or later (2.00 odds ratio).
  • GE’s 7T MRI FDA: GE Healthcare announced its SIGNA 7.0T MRI’s FDA approval, calling it the most powerful FDA-cleared MRI (5x more than most MRIs), and emphasized its versatility (brain/joints, clinical/research), resolution, and detail. GE also highlighted the SIGNA 7.0T’s UltraG gradient technology (its most-powerful whole-body gradient coil) and its use of the company’s familiar SIGNAWorks applications platform.
  • Keeping Electives Open: A new Duke study in JAMA detailed a clinical decision support tool that hospitals can use to predict available resources and help manage elective procedure scheduling (particularly during the COVID pandemic). The CDS tool (developed w/ 42k patients’ EHR data) was able to predict hospital resource metrics such as length of stay and ventilator use with AUCs ranging from 0.76 to 0.93. This study didn’t directly mention imaging, but imaging volumes would certainly benefit if a tool like this helped keep electives open.
  • BrainTale’s €1.7m: BrainTale completed a €1m seed round that it will use to support its development and commercialization strategy. BrainTale’s CE-marked brainQuan software uses MRI scans to quantitatively measure brain microstructure and predict traumatic brain injury recovery.
  • Reconstructing Thin Slice MRI: New research out of Korea found that deep learning–based MRI reconstruction can produce thin-slice (1mm) pituitary MR images that are diagnostically superior to thicker slice conventional images. By addressing the image noise challenges that are often associated with high spatial resolution pituitary MRI, the DL-reconstructed 1mm images were able to outperform the conventional 3mm slices.
  • GE & GenesisCare’s Big Partnership: GE Healthcare announced a 5-year, $130m partnership with GenesisCare that will provide the cancer and cardiac care company’s 440+ locations with GE imaging technology, solutions, and services. The partnership will span across GenesisCare’s locations in Australia, the U.K., Spain, and the company’s recently acquired U.S. locations.
  • AI Buyers Guide: A new Emory paper detailed the key factors providers should consider when evaluating AI software (decision makers, data ownership / privacy, costs, KPIs, and ROI), and went a step further detailing the majority of the AI products available (119 products, 55 companies). There’s a lot of info in this paper, so check it out if you’re interested.
  • Konica Minolta & Dia Imaging’s Cardiac Expansion: About 1.5 years after making DiA Imaging Analysis’ LVivo Cardiac Toolbox available on Konica Minolta’s Exa Cardio PACS, the companies announced the addition of new clinical and financial benefits. Exa Cardio PACS users can now leverage Dia Imaging’s’ LVivo EF tool to analyze the left ventricle for cardiac dysfunction in COVID-19 patients, and receive reimbursements in the U.S. for myocardial strain imaging performed with Dia’s LVivo Strain tool.

The Resource Wire

– This is sponsored content.

  • Learn how GE Healthcare’s analytics and AI could be the key to precision health success.
  • Nuance’s latest blog details how Radiology Alliance (Tennessee, a Mednax affiliate) deployed Nuance PowerScribe Follow-up Manager in the middle of the COVID-19 emergency, allowing them to identify and communicate 2,288 incidental findings since their April launch.
  • Learn how Arterys’ Cardio AI can improve your Cardiac MRI workflow and cut LV/RV function segmentation by up to 25 minutes.
  • Documentation is critical in order to be properly reimbursed for yttrium-90 (y-90) radioembolization procedures. Learn how to document these procedures for maximum reimbursement in this blog article from Healthcare Administrative Partners.
  • Check out Riverain Technologies’ on-demand webinar demonstrating how its AI solutions integrated into LucidHealth’s radiology workflow and sharing best practices on how to combine AI with radiologist expertise.

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

Join thousands of imaging professionals.