“The future doesn’t happen all at once…”
Koios Medical CEO, Chad McClennan, in response to the WHO’s big tuberculosis AI recommendation.
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.
- Novarad – Transformational imaging technologies that empower hospitals and clinicians to deliver clinical, operational and fiscal excellence.
- 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.
- United Imaging – Our mission, Equal Healthcare for All, pushes us beyond conventional boundaries to help clinicians expand modern, digital, intelligent care to more people within their communities.
- Zebra Medical Vision – Transforming patient care with the power of AI.
The Imaging Wire
The WHO’s Big TB Recommendation
The World Health Organization just recommended using computer-aided detection AI software to identify and triage tuberculosis in chest X-rays — as an alternative to human interpretations.
- Recommendation & Rollout – The WHO’s updated TB screening guidelines recommends using CAD tools for screening/triaging TB in individuals who meet certain criteria (e.g. ≥15 years, regional prevalence, exposure history, comorbidities, etc.). The WHO also released a suite of digital tools to help countries launch and operate their CAD-based TB screening programs.
- TB AI’s High Ceiling – The WHO’s new TB screening guidelines should significantly expand the use of TB CAD solutions, given that about a quarter of the world’s population is infected with TB and 10m become ill/contagious each year. That could mean a whole lot more X-rays and automated TB diagnoses, and a lot of opportunities for the companies who make TB CAD solutions (e.g. Qure.ai, Lunit).
- A Long Time Coming – Chest X-rays have played a key role in TB screening and triage for decades, but there’s never been enough human clinicians to support CXR screening programs. In more recent years TB AI developers and TB health organizations have actively (and successfully) worked to drive real-world TB CAD adoption and validation, leading to the WHO’s new recommendation.
- An AI Milestone – The WHO’s new guidelines could also have ramifications beyond tuberculosis, as it’s the first major endorsement for AI-only image interpretation and it could serve as a blueprint for other screening efforts in low-resource regions. Whether it leads to AI-only discussions in developed regions is unclear, but that’s worth keeping an eye on too.
The Wire
- Ultrasound Mindreader: A Caltech team is using functional ultrasound to develop noninvasive brain–machine interfaces (BMIs) that can interpret neural activity and then transmit instructions to machines or computers (e.g. allow a paralyzed person to move a robotic arm). The Caltech scientists view their functional ultrasound-based BMI as a potential breakthrough, since BMI currently relies on surgically implanted electrodes or less-effective external modalities like fMRI (too bulky, expensive) and electroencephalography (low spatial resolution).
- COVID Ultrasound AI Perfection: The Lawson Health Research Institute developed an ultrasound AI model (trained w/ 243 patients, 612 recordings, 121,400 frames) that identified COVID pneumonia with “perfect” accuracy. The model achieved this “perfect” score using ultrasound scans from 25 patients with COVID and other lung diseases, outperforming 61 ultrasound-trained acute care physicians who identified the same COVID scans with accuracy that was “no better than a coin toss.”
- Hold on to Contrast Monitoring: A new JACR opinion paper argued that radiologists should think twice before making technologists responsible for monitoring contrast administration for severe reactions (as proposed in a separate JACR editorial). The paper encouraged radiologists to keep this responsibility because: 1) The number of severe reactions is “far from trivial;” 2) Radiologists have unique medical knowledge that would be difficult to transfer to RTs; 3) This change would undermine rads’ goal to increase patient interaction; 4) It could cut reimbursements.
- Identifying X-Rays: All those anonymized chest X-ray datasets might not be as anonymous as we think. A team of German researchers used a public CXR dataset and an X-ray classification system to develop a deep learning-based reidentification model that was able to identify two images from the same person with 95.55% accuracy. In theory, hackers with access to unanonymized X-rays could use reidentification models like this to gain access to patient healthcare data that’s included in anonymized images (diagnoses, histories, site of care, etc.).
- Automated Reminders Work: Automated imaging appointment reminder systems are just as effective for avoiding no-shows as conventional phone/mail reminders performed by imaging staff. That’s from a Michigan Medicine study that compared CT and MRI imaging appointments during the months before (using calls + written letters) and after adopting an automated appointment reminder system (using automated texts + calls + patient portal notifications). Patients reminded with traditional communication methods had a 2.82% “missed care opportunity” rate (292/10,348), while patients reminded with the new automated communication methods missed 2.44% of their appointments (262/10,719) one month later.
- GE’s New C-Arm: GE Healthcare announced the FDA clearance of its new OEC 3D mobile C-arm surgical system, highlighting the flexibility of its combined 3D and 2D imaging, and the CT-level quality of its 3D volumetric images.
- No Explanation Necessary: There’s been many calls for greater AI explainability, but a recent JACR editorial argues “that explainability might not be sufficient to promote trust” and it “might not even be necessary.” That’s because: 1) Humans can’t even adequately explain their own decisions; 2) AI explainability might come at the expense of accuracy; 3) Trustable AI goes far beyond the model (also data, workflow, designers, ecosystem, etc.); and 4) We still don’t have “continuous learning” models that could explain decisions with historical context. Until we have explainable AI sorted out, the authors suggested that we should foster AI trust through educating rads on AI’s limitations, enforcing AI best practices, and educating patients about AI.
- Lucida Funded: UK imaging AI startup Lucida Medical completed a multimillion-pound Seed round that it will use to develop and validate its AI solutions that analyze MRI scans for signs of cancer, initially focusing on a prostate cancer solution. Lucida’s MRI analysis software combines radiogenomics, machine learning, and image processing.
- ScanVan’s Screening Success: A new Clinical Imaging study detailed the New York-based ScanVan mobile mammography screening program’s success, suggesting that no-cost exams and mobile access are key to improving screening among underserved women. In 2019, the ScanVan screened 3,745 women (66% Hispanic & African American, 43% uninsured, 15% Medicare), identifying 17 new breast cancers and resulting in 258 recalls.
- PMV Acquires Mindshare: Precision Medical Ventures (PMV) acquired lung CT AI company, Mindshare Medical, revealing plans to accelerate the commercialization of Mindshare’s RevealAI-Lung CADx lung nodule characterization software. PMV will do business under the Mindshare Medical name, while bringing in new well-connected leaders, new funding, and likely new strategies.
- AI Fear & Understanding: The more radiologists understand AI, the less they’re scared of being replaced by it, although plenty of rads are still concerned. That’s from a survey in European Radiology (n = 1,041 radiologists/residents in 2019) that found that 48% view AI positively and 38% fear replacement by AI, while optimistic and pessimistic responses were directly correlated with the respondents’ knowledge of AI. The authors (who appear to be among the AI optimists) called for expanded AI education.
- Two Sides to Pediatric Brain MRI Incidentals: Pediatric brain MRI scans result in far more incidental findings that many might think, but very few of clinical significance. That’s from a new JAMA study that reviewed brain MRIs from 11,679 9-10yrs-old children and discovered incidentals in 2,464 (21.1%) of them. However, only 431 of the incidentals were assigned category 3 (3.7%, “consider referral”) and 20 were assigned category 4 (0.2%, “consider immediate referral”).
The Resource Wire
– This is sponsored content.
- See why the Medical University of South Carolina decided to use Siemens Healthineers’ AI-Rad Companion with every chest CT scan that the hospital system performs.
- Learn how Zebra-Med’s AI Coronary Calcium solution can provide an ‘early-warning’ system for patients with coronary artery disease using existing CT scans.
- 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 (rather than CDs).
- On March 21st, United Imaging celebrated its tenth year in business. Since its founding in 2011 the company has over 13,000 global installations and continues to grow. A video on LinkedIn celebrates United Imaging’s employees and customers worldwide.
- Learn more about Arterys’ new AI assistant for neurovascular emergencies, which can perform stroke detection on CT and brain tumor diagnostics on MR (among other things).
- This Hitachi blog details how a complete CVIS solution drives efficiencies and increases productivity.
- Learn how Saint Joseph Mercy Health System doubled its follow-up recommendation identification and tracking and achieved 1st-year ROI with Nuance PowerScribe Follow-up Manager.
- Properly documenting and coding interventional radiology Evaluation and Management (E&M) services is a collaborative effort. Check out this Healthcare Administrative Partners guide detailing how to do it right.
- This Riverain Technologies case study details how Einstein Medical Center adopted ClearRead CT enterprise-wide (all 13 CT scanners) and how the solution allowed Einstein radiologists to identify small nodules faster and more reliably.
- In this 60-second GE Healthcare video, Dr. Elizabeth Bergey of Quantum Imaging and Therapeutic Associates discusses the critical role of workflow orchestration in radiology efficiency.
- Learn how Salem Regional Medical Center improved its radiology workflows and cut service and syringe costs after adopting Bayer’s MEDRAD Stellant FLEX system.
