AI Needs Assessment | Space Ultrasound | SIIM 2021

“Our industry is littered with the bones of standards that got developed too early.”

Blackford Analysis CEO, Ben Panter, proposing a retrospective approach to creating AI standards.

Imaging Wire Sponsors

Arterys | Bayer Radiology | Canon Medical Systems | GE Healthcare
Healthcare Administrative Partners | Hitachi Healthcare Americas
Novarad | Nuance | Riverain Technologies | Siemens Healthineers
United Imaging | Zebra Medical Vision

The Imaging Wire

An AI Needs Assessment

A trio of AI heavyweights (Blackford’s Ben Panter, Nuance’s Karen Holzberger, and Philips’ Kevin Lev) took the SIIM 2021 stage to discuss “Who Should (and Will) Pay for” imaging AI.

However, the conversation quickly shifted to clarifying what needs to happen in order for imaging AI to reach its commercial and clinical potential… which would likely make questions about AI funding far more straightforward.

Here are some of the big takeaways:

  • Early AI – Imaging AI is very new, which means we’re still establishing many AI pathways, users, measurements, value propositions, and business models.
  • The Reimbursement Picture – There will be more AI reimbursements, but it’s unclear how they would be achieved and how much they’d be worth — and CMS/payers will want to see a lot more ROI evidence before they expand AI reimbursements.
  • Today’s AI Criteria – In the absence of reimbursements, imaging AI buying decisions are usually based on how/whether a tool improves radiologists’ workflow efficiency and clinical effectiveness.
  • Two Criteria Caveats – However, it’s still difficult to measure how many AI tools improve radiologists’ real-world efficiency and effectiveness — and this radiology-centric ROI focus doesn’t account for how AI adds value across care pathways (patients, referrers, etc.).
  • The Measurement Imperative – Given AI ROI’s role in justifying near-term business decisions and future reimbursement decisions, establishing a reliable and low-impact way to measure ROI is among the top AI challenges to solve.
  • The Onboarding Imperative – The resource costs associated with onboarding AI has become viewed with as much importance as the “actual cost of the tool,” highlighting the need for improved implementation processes and technologies.
  • What’s Working Now – Triage and other productivity AI tools are proving to have a lower entry barrier than interpretation-focused tools, because their value is straightforward and there’s less concern that they’ll slow radiologists down. We’ve also seen real adoption and reimbursement progress with stroke AI products, which are based on imaging but deliver value within stroke care workflows.
  • Thinking Beyond the Pixel – Because so much of radiology’s value “is in the report,” there’s a major opportunity in non-imaging AI (CDS tools, NLP-based tools, etc.), despite most AI players’ continued focus on pixel-based tools.

How to Ditch the Disk with Novarad CryptoChart

CD burning issues? Check out this one-minute video showing how Novarad’s CryptoChart image sharing solution allows patients to easily access and share their medical images using personalized, highly secure QR codes (not than CDs).

– Sponsored.

Biograph Vision Quadra’s FOV Impact

How can a larger field of view help improve PET/CT scans? See how an extended field of view allows for greater visualization in less time in Siemens Healthineers’ Biograph Vision Quadra PET/CT image gallery.

– Sponsored.

The Wire

  • Ultrasound in Space: UltraSight is coming aboard the International Space Station, where Astronauts will use its cardiac ultrasound AI guidance platform to study how heart anatomy changes during space travel. Although outer space is hard to beat, handheld ultrasound’s small size and expanding AI guidance capabilities have also earned it headlines for its use on the battlefield, on Mount Everest, and at Antarctic research stations.
  • CDS AI Works: Research presented at SIIM 2021 detailed how Yale’s appropriate imaging performance improved after adding an AI-supported text input to its Clinical Decision Support system (CDS). After adding a text section that leverages AI to suggest structured indications (rather than making rads search for indications), Yale reduced its ratio of orders without CDS scores (46% to 31%) and increased its appropriate orders (41% to 50%).
  • PET Avoids cHL Radiotherapy: Post-chemo PET exams could eliminate radiotherapy for 78% of patients with early-stage bulky classic Hodgkin lymphoma (cHL). Dana-Farber researchers performed PET scans on 94 patients after two chemotherapy cycles, finding that patients with less disease uptake than liver uptake could complete their treatment through four more chemo cycles (vs. 4 chemo + radiotherapy) and still have higher survival rates without progression (93.1% vs. 89.7%).
  • IBM’s 3D Update: IBM Watson Health updated its iConnect Access diagnostic viewer and image exchange platform, introducing a new 3D interactive segmentation option (creates 3D anatomic models) that will support the platform’s future 3D printing capabilities. IBM also announced an end-to-end 3D printing partnership with Ricoh USA that will produce these 3D models.
  • COVID’s Uneven Screening Impact: A JAMA study detailed COVID’s uneven impact on Washington’s breast cancer screening rates. Overall exams fell by 49% between April – December 2019 and April – December 2020 (55.6k to 27.5k), with larger declines among minority women (-53.9% to −64.2% vs. -49.2% white), women in rural areas (roughly -59% vs. -50% urban), and women who self-paid or were covered by Medicaid (roughly -70% & -62% vs. -50% private coverage).
  • Artrya’s Cardiac CT Funding: Australian AI startup Artrya completed a $15m AUD funding round that it will use to develop and commercialize its Salix cardiac CT software. Using cardiac CT scans, Salix produces a 3D cardiac image, a report analyzing patients’ coronary artery disease risk (plaque, stenosis, calcification), and Fractional Flow Reserve (FFR) analysis.
  • Enteric Tube Placements Shift to IR: Although enteric tube placement procedure volumes fell by 32% between 2010 and 2018 (157k to 107k), fluoroscopy-guided tube placements and replacements increased significantly (+18% & +55%). That led to a lot more enteric tube procedures for interventional radiologists, who performed almost 92% of fluoroscopy-guided placements/replacements in 2018.
  • Surveillance Mammography’s Illusive Improvements: Despite major technology advancements, surveillance mammography has achieved only “minimal improvements.” A Radiology Journal study compared surveillance screening outcomes over two decades (32k women, 1996–2007 vs. 2007–2016), revealing stable or only slightly improved cancer detection rates (6.8 vs. 8.5 per 1k), interval cancer rates (both 3.6 per 1k), sensitivity (65.4% vs. 70.4%), and specificity (98.3% vs. 98.1%).
  • KA in the USA: KA Imaging will make its Reveal 35C dual-energy X-ray detector available across Alpha Imaging’s 14-state central/northeast U.S. territory. Alpha Imaging has plenty of X-ray partners, but the Reveal 35C’s unique capabilities (3 images per scan, upgrades standard X-rays, lower dose/cost than dual energy-systems) should help it stand out.
  • Improving TI-RADS with AI: A SIIM 2021 study detailed how Koios Medical’s thyroid nodule AI system (produces TI-RADS assessments w/ added risk analysis) could improve reader accuracy and efficiency. Fifteen readers evaluated 650 nodules (130 malignant) with and without AI, posting significantly improved average AUCs (+0.083), sensitivity (+8.4%), specificity (+14%), and interpretation times (-23.6%) when they had AI support.
  • ID’ing IVCs: Another SIIM 2021 presentation detailed a deep learning model that can automatically identify patients who never had their temporary IVC filters removed. UCSF researchers developed the model with 5,225 annotated CXRs (30% with IVC filters), achieving high sensitivity and specificity with both an internal dataset (15% of database, 96.2% & 98.9%) and a 1,424-image external set (97.9% & 99.6%).
  • Canon Medical Marketplace: Canon Medical Systems announced its new Medical Marketplace, an online storefront that imaging providers can use to quickly order Canon parts for 15% below list.
  • Incidental Fatty Liver Screening: Abdominal ultrasound and CT exams could serve as an “opportunistic screening” pathway for hepatic steatosis (aka fatty liver disease), a common incidental that’s not always reported or addressed. That’s from a new JACR study that added follow-up recommendations to 1,256 radiology reports with incidental hepatic steatosis findings, resulting in 52% of these patients receiving follow-up testing/management.
  • 18F PET/CT Prostate Planning Advantage: A new Emory study found that 18F-fluciclovine PET/CT imaging (Blue Earth’s Axumin) can “significantly improve” post-prostatectomy radiotherapy planning and decision making. The researchers performed 18F-fluciclovine PET/CT or conventional imaging (bone scan and either CT or MRI) on 165 men, finding that PET-based radiotherapy planning led to far higher event-free survival rates (3yr 75.5% vs. 63%; 4yr 75.5% vs. 51.2%).

University of Colorado Hospital’s Case for ClearRead Bone Suppress

This Riverain Technologies case study details how the University of Colorado Hospital enhanced its chest X-ray workflow with ClearRead Bone Suppress.

– Sponsored.

The Resource Wire

  • Know how your practice measures up? In this post, Healthcare Administrative Partners details the key benchmarking quality metrics and how they can help radiology practices improve.
  • See how physicians are leveraging GE Healthcare’s Edison Open AI Orchestrator platform and Icobrain’s AI-enabled applications to monitor and treat MS and Traumatic Brain Injury.
  • See how and why Zebra Medical Vision sees a much bigger opportunity in public health AI than many of us imagine in this Imaging Wire Q&A with company CEO, Zohar Elhanani.
  • Canon Medical System’s new Aquilion Exceed LB CT radiation therapy planning system has a lot to boast about, combining AiCE reconstruction technology with an industry-leading bore opening (90cm), field of view (90cm), and detector coverage (4cm).

You might also like

You might also like..

Select All

You're signed up!

It's great to have you as a reader. Check your inbox for a welcome email.

-- The Imaging Wire team

You're all set!