“It’s funny – when I’m asked whether I see AI replacing radiologists, I point out that in most of the rest of the world there aren’t any radiologists to replace.”
Qure.ai’s Pooja Rao on her experience using AI to fight TB across the developing world.
Imaging Wire Sponsors
- Carestream – Focused on delivering innovation that is life changing – for patients, customers, employees, communities and other stakeholders
- Focused Ultrasound Foundation – Accelerating the development and adoption of focused ultrasound
- Medmo – Helping underinsured Americans save on medical scans by connecting them to imaging providers with unfilled schedule time
- Nuance – AI and cloud-powered technology solutions to help radiologists stay focused, move quickly, and work smarter
- POCUS Systems – A new Point of Care Ultrasound startup, combining a team of POCUS veterans with next-generation genuine AI technology to disrupt the industry
- Qure.ai – Making healthcare more accessible by applying deep learning to radiology imaging
The Imaging Wire
Canon Focuses on Efficiency at RSNA
Canon Medical Systems announced a wide range of new systems and technologies at RSNA, revealing a major focus on AI-based image reconstruction, but less of a focus on AI diagnostic apps than many of its peers.
- AiCE Expands – Canon expanded its AiCE image reconstruction technology across much of its CT lineup (Aquilion ONE / GENESIS Edition, Aquilion Precision CT systems, Aquilion ONE / PRISM Edition, and Aquilion Prime SP CT) and into its MR portfolio (Vantage Galan 3T, Vantage Orian 1.5T MR), just a year after AiCE was added to Canon’s high-end CTs. AiCE reduces noise and boosts signal, producing higher quality images in less time.
A year – https://www.theimagingwire.com/2018/11/29/rsna/
- Aquilion ONE / PRISM Edition – Canon announced the Aquilion ONE / PRISM Edition spectral CT, highlighted by its included AiCE technology (previously limited to high-end CTs) for improving image quality and scan speeds, as well as its support for Canon’s FDA-pending Deep Learning Spectral Imaging (supports diagnosis with spectral insights).
- Compressed SPEEDER MR Technology – Canon Medical’s Compressed SPEEDER technology (pending 510(k) clearance) reconstructs full resolution MR images from highly under-sampled data, allowing 4-times faster scans while maintaining resolution and signal to noise ratio. The new technology will be available with Canon’s Vantage Galan 3T and Vantage Orian 1.5T systems.
- Ultrasound Update – Canon’s Aplio i-series ultrasound systems are now offered with a micro-convex wide bandwidth 2-in-1 iDMS traducer (abdominal, fetal, and pediatric exams) and a pediatric transducer (cardiac and pediatric exams), support optional integrated Breast Imaging Reporting and Data System (BI-RADS) and Thyroid Imaging Reporting and Data System (TI-RADS) tools, and works with Canon’s Attenuation Imaging (ATI) tool (previously only with Aplio i800).
- Security Suite – Canon Medical Systems unveiled its Gateway Platinum multi-level cybersecurity solution, combining a Barracuda NextGen firewall, InnerVision Plus protection (isolates Canon systems from network), remote diagnostic support, and a Secure VPN to Canon Medical Control Center.
- Virtual Imaging Realigned – The strategic realignment of Canon Inc.’s medical business reached another milestone with the transfer of Virtual Imaging from Canon USA to Canon Medical Systems USA, combining their portfolios, sales and service teams, and customer bases.
- Expanded Education – Canon Medical Systems updated its web-based 360° Learning portfolio, adding Canon Medical On Demand (educational webinars in clinical, technical, and practice-related imaging topics), Clinical Support Library (a library of online clinical support tools and use-case videos), and Healthcare Economics Learning Portal (a public portal to help clinicians understand changes in healthcare policy reimbursements).
Nuance’s Big PowerScribe One Update
Nuance rolled out a range of major enhancements to its PowerScribe One platform at RSNA 2019 that address usability, efficiency, and data integration.
- Report Quality, Consistency, and Efficiency – Nuance doubled-down on reporting with its latest PowerScribe One enhancements, adding a new Ambient Mode (automatically converts free-form dictation into structured reports), introducing a new feature that highlights relevant text in radiology reports, and integrating its new Smart Assist feature (eliminates manual data entry and automatically checks grammar/spelling).
- “Hey PowerScribe” – Nuance added virtual assistant functionality to PowerScribe One, giving radiologists another way to quickly access patient data from the PACS and EHR or use voice commands to complete tasks (e.g. “Hey PowerScribe, turn on SpotView,” or “Hey PowerScribe, send a message to Dr. Smith”).
- Nuance PowerCast – Nuance PowerCast is an extension of the proposed HL7 FHIRcast standard that enables real-time data synchronization between PowerScribe One, PACS, EHR, and other connected platforms. Nuance now supports two-way data-exchange with Visage, Sectra, BARCO, and Client Outlook, joining its previously-announced support for Siemens and Cerner PACS.
- Marketplace Integration – PowerScribe One now integrates to the Nuance AI Marketplace through the Nuance PowerShare image-sharing network, integrating AI results directly into radiology visualization and reporting workflows.
- Structured Terminology & CDE Support – PowerScribe One supports the use of structured report terminology and Common Data Elements (CDE), improving report accuracy/consistency, expanding data-sharing, and producing consistent and reliable data for use with AI models.
- Simplified UX – Nuance modernized and simplified the PowerScribe One user interface, adding a “dark mode” to reduce radiologist fatigue and an intelligent workspace that only displays data and tools when they are needed.
- Admin Dashboards – PowerScribe One also added graphical administrator dashboards, providing real-time access to system and radiologist performance data to assess workloads, productivity, and continuous improvement program status.
GE Healthcare Maintains AI Focus at RSNA
After an extremely active RSNA last year, GE Healthcare’s RSNA 2019 launches mainly focused on evolving its Edison AI platform, while also expanding its breast health and interventional portfolio.
- Edison Developer Program – GE Healthcare launched its Edison Developer Program, an expansion of GE’s Edison ecosystem that provides developers with a range of services to help speed up product development and then scale and deploy their solutions across GE’s customer base.
- Edison Open AI Orchestrator – Effectively serving as GE’s AI app marketplace and integration platform, Edison Open AI Orchestrator simplifies the implementation, deployment, support, and scaling of multiple AI applications (currently: Arterys, iCAD, Koios Medical, MaxQ AI, and Volpara) into a GE PACS reading workflow.
- MR Reconstruction – GE unveiled its new AIR Recon DL3 Edison deep-learning application, using MRI reconstruction to improve signal-to-noise and image sharpness and reduce scan times.
- Serena Bright – Calling it the first contrast-enhanced mammography (CESM) solution for biopsy, Serena Bright enables breast biopsy exams with the same mammography equipment, in the same room, and with the same staff as the screening mammogram.
- Revolution Maxima – Leveraging real-time depth sensing technology, Revolution Maxima simplifies and improves patient positioning for CT scans by spotting the center of the scan range and automatically aligning it to the isocenter of the bore.
- OEC Elite’s New UI – GE’s OEC Elite C-arm adds a new OEC Touch control panel located on the C-arm, eliminating the need to move to a separate workstation during surgical procedures.
- Embo Assist with Virtual Injection – The Embo Assist with Virtual Injection Edison app helps clinicians perform complex embolization procedures in the interventional suite, allowing them to analyze vasculature and simulate injections to avoid embolizing healthy tissues in the brain or prostate.
- Hospital industry groups followed through on previous threats, filing a lawsuit against the U.S. HHS in an effort to stop the upcoming healthcare cost transparency rule that would require hospitals to disclose negotiated rates with insurers. The suit argues that the rule violates the First Amendment (“provoking compelled speech”) and goes beyond the intended meaning of “standard charges” transparency in the Affordable Care Act. The industry group also threatened that the burden to comply with this rule would be “enormous,” actually create higher prices (LOL), and lead to more healthcare consolidation.
- A RSNA panel tackled the radiology corporatization trend revealing that corporate ownership of radiology practices increased from an estimated 6% in 2017 to roughly 10% now. The panelists detail the reasons that some practices decide to sell (hospital consolidation, market competition, after-hours call volumes, IT expenses, regulatory and compliance costs, and balancing medical practice with business management duties) and why others don’t sell (good health system relationships, low competitive pressure, high number of junior partners who wouldn’t benefit, partners who enjoy the business component of practice management).
- A well written guest post on thehealthcareblog.com from UPenn radiologist and Radiology Twitter star, Suarabh Jha, MD, detailed India’s massive tuberculosis challenge and how AI might help solve this problem by streamlining diagnosis efforts. Or as Jha puts it, “Artificial Intelligence is chest radiograph’s Sancho Panza.” The article specifically details how Mumbai’s own AI hero, Qure.ai, got its start and how it’s working to tackle TB diagnosis.
- A Washington University (St. Louis) team developed a new colon cancer imaging technique that identified tumors with 100% accuracy in its pilot study and might serve as a supplement to colonoscopy screening. The team’s new PR-OCT imaging technique combines optical coherence tomography (OCT) attached to a traditional colonoscopy endoscope and fed to a neural network model (trained on 26k OTC image frames of colorectal tissue samples) that achieved a sensitivity of 100% and a specificity of 99.7%.
- iCAD introduced ProFound AI Powered by Panorama, a future expansion of its Profound AI platform that helps clinicians correlate patients’ prior DBT images and breast cancer risk assessments over time. ProFound AI for DBT gained FDA clearance late last year and is currently used to review each DBT image/slice and provide radiologists with key information (e.g. Certainty of Finding lesion, Case Scores) to support decision-making and improve reading efficiency.
- A new study from reaction data (n= 900 organizations) revealed that healthcare organizations are increasingly shifting towards single source radiology IT partnerships (64% of organizations adopted it, 71% prefer it) and away from a “best in breed” deconstructed PACS approach (only 54% adopted a single vendor approach in 2016). Organizations using a single vendor were most likely to work with GE (followed by: Change, IBM/Merge, Philips, Sectra, Agfa), while organizations using a “best in breed” approach were most likely to use Change for PACS (followed by: GE, Philips, Fujifilm), IBM for VNA (followed by GE, Change, Philips), and Change for their viewer (followed by: IBM, Philips, GE).
- RSNA president, Valerie P. Jackson, MD, kicked-off the meeting with a call for radiologists to venture out of the reading room and engage with their patients and peers, pointing to the mounting evidence that the current reading room standard no longer works (more time available due to AI efficiencies, pervasive unnecessary imaging, rise of team-based care, high burnout). To get out of the reading room, Dr. Jackson suggested that radiologists challenge any beliefs that their personalities aren’t fit for patient interactions, work to expand their understanding of the patient experience, and increase interaction/collaboration with referring physicians and members of the care team.
- A Google Health team achieved “expert-level” accuracy with four chest X-ray AI models developed to detect pneumothorax (0.95 AUC), nodule or mass (0.72), airspace opacity (0.91), and fracture (0.86), with the models performing “on par with” four board-certified radiologists. The team used over 600,000 images from two de-identified datasets (1: India’s Apollo Hospitals, 2: NIH’s publicly available ChestX-ray14) and then used a deep learning model to extract image labels from the ChestX-ray14 set and apply them to the 560k-image Apollo Hospitals dataset, revealing a possible way to label and leverage massive datasets.
- MaxQ AI had a busy week at RSNA, announcing a partnership with Flywheel (supports data management and curation for AI development), adding its Intracranial Hemorrhage (ICH) AI application to the Blackford Platform and GE’s Edison Open AI Orchestrator, and announcing the company’s ISO 27001 certification for cybersecurity management.
- An RSNA presentation suggested that coronary CTA and FFR-CT is as effective at supporting heart disease treatment decision making as coronary angiography, and may be the better option given that it’s not invasive. The study of 223 patients with left main or three-vessel coronary artery disease, revealed a high 0.82 treatment decision rate between CTA+FFR and conventional angiography (Cohen’s kappa 0.82), while CTA+FFR CTA changed the treatment decision in 7% of the patients.
- A guest editorial in The New York Times introduced the idea of AI bias to much of its readership last week, detailing how racial bias is pervasive among both humans and AI models, but revealing the good news that AI bias (which is really due to biased data) can be fixed. The article highlighted a prototype solution that fixes algorithmic bias and revealed that bias-fixing service has already been provided to health systems, prompting the radiology AI Twitter community to share additional thoughts on how to address AI bias.
- Neusoft’s RSNA booth brought the introduction of the NeuViz Epoch CT (512-slice, 16cm detector, 0.259s rotation, 25ms, temporal resolution) highlighted by its organ-specific AI algorithms and Iterative Reconstruction software, and the debut of its MDaaS AI solutions portfolio (details are limited). NeuViz Epoch CT and MDaaS aren’t available in the U.S. and would almost surely hit Neusoft’s other regions before the company considers expanding them stateside.
The Resource Wire
- This Healthcare IT News piece details how Summa Health improved its incidental lung nodule findings follow-up rates by 662% in six months using Nuance’s AI, analytics, and NLP solutions.
- If you weren’t already following along, Qure.ai, RAD-AID, and Saurabh Jha recently had an excellent tweetchat about the use of imaging AI to fight TB and other diseases in developing regions.
- This Carestream Special Report details how providers can get the greatest ROI from their X-ray technology as radiography demands increase and budgets head the other direction.
- The Focused Ultrasound Foundation’s 2019 State of the Field Report details FUSF’s initiatives and achievements and the state of the focused ultrasound market.
- POCUS Systems’ forthcoming ultrasounds will combine ease of use, durability, and reliability, allowing clinicians to focus on their patients.
- Yale University research reveals that the average patient drives past six lower-cost providers on the way to an imaging procedure, due in large part to patients’ and physicians’ limited cost consciousness. Medmo helps address this issue by letting patients enter what they can afford for their scan, then booking them at a nearby imaging center willing to accept that rate.