#362 – The Wire

  • Automating AI with Multimedia Reporting: A Philips paper in JACR proposed a new radiology reporting process that they believe would address AI’s labeling, training, performance, and monitoring challenges. The novel annotation and dictation system would automatically create hyperlinks that connect report text with specific findings on medical images. They suggest that these links would produce stronger labels for AI training without hindering radiologist efficiency, while also creating an automatic feedback loop for monitoring and retraining already-deployed algorithms.
  • Samsung’s Photon Counting CT Supply: After years of joint R&D, Samsung and 5N Plus announced that 5N Plus will supply Samsung with detector substrates used in Samsung NeuroLogica’s forthcoming photon counting CT imaging systems. Although Samsung has made its PCCT development efforts public in recent years, this supply announcement has greater relevance given Siemens Healthineers’ new PCCT FDA 510(k) and the recent PCCT detector maker acquisitions by Canon and GE.
  • CT AI for Emphysema Severity: A new MGH study found that Siemens Healthineers’ AI-Rad Companion Chest tool can assess emphysema severity in chest CT scans with similar accuracy as radiologists. The researchers had the AI tool and two blinded thoracic radiologists assess 113 adults’ emphysema severity using non-contrast chest CTs (severity levels: mild, moderate, or severe), achieving very similar accuracy (AUCs: 0.77 AI & 0.76 rads).
  • Siemens and Mizzou’s Online Alliance: The University of Missouri and Siemens Healthineers are teaming up to develop one of the first online clinical engineering programs in the U.S., following through on one of the main initiatives outlined in their 10-year strategic alliance. The University of Missouri will use a $2.6m state grant combined with Siemens’ support and technology to train, certify, and recruit future generations of clinical engineers.
  • Radiology Zoom School: Speaking of online education, a new Insights into Imaging study suggests that far more radiology education will be held online in the future. The 244 radiologists from 31 countries overwhelmingly agreed that radiology is “particularly well-suited for online teaching” (91%) and estimated that 50% of radiology training is appropriate for online learning. That would be a big shift considering that only 12% of radiology training was online before the pandemic.
  • Low-Field MRI for MS: A new UPenn-led study suggests that ultra-low-field portable MRI systems (ULF-MRI; in this case the Hyperfine Swoop) could eventually support point-of-care multiple sclerosis screening and monitoring. The researchers performed contrast-enhanced brain scans using the Swoop and a high-field MRI on 17 patients (15 w/ MS), finding that the Hyperfine scans had 100% sensitivity for detecting >5mm white matter lesions. The Swoop was able to detect Leptomeningeal contrast enhancement with 33% sensitivity, which is low, but could be improved in the future.
  • Why Not Fear AI: A new JMIR paper detailed why future radiologists shouldn’t be as concerned about AI’s impact on their careers as many recent surveys say they are. The authors shared a long list of reasons why these AI concerns might be overblown (at least in the short term) including: 1) AI’s current performance limitations; 2) Healthcare’s disconnected and heterogeneous digital infrastructure; 3) Lacking AI literacy and training; and 4) Liability and ethical barriers.
  • RamSoft Adds Imaging Sharing: RamSoft became the latest PACS player to expand into image sharing, making its new QR Snap app a standard feature in its PowerServer RIS/PACS platform. QR Snap allows imaging facilities to share patient images / reports with physicians using a QR code, while also allowing practices to provide patients with their QR codes. There are quite a few image sharing options, but this is the first we’ve heard of that comes standard with a PACS system.
  • Neck US Consultations: A new Clinical Imaging study shared a solid case in favor of radiologist-patient consultations following neck ultrasound exams. The prospective/randomized study (109 patients, 44 who had rad consultations) found that radiologist consultations reduce patient anxiety during and after exams (P = 0.040 & 0.027), are viewed as important by most patients (92.7%), and do not significantly prolong exam times (7.6 vs. 7.3 minutes). 
  • Imaging Centers Liable in Indiana: An Indiana court expanded the state’s medical malpractice rules to hold imaging centers “vicariously liable” for the negligence of independent entities’ employees and contractors (yes, they mean independent radiologists). This liability expansion is the result of a lawsuit that held both an imaging center and a contracted radiologist responsible for a missed diagnosis (spinal arteriovenous fistula), rather than just the radiologist. Indiana’s hospitals have been vicariously liable for this type of non-hospital negligence for over 25 years.
  • PSU’s Photoacoustic Helmet: Penn State researchers are developing a wearable head scanner that uses photoacoustic imaging to help diagnose brain function and behavior health. With the help of a two-year $900k NIH grant, the team will develop a prototype wearable scanner, which they believe could become an attractive MRI alternative (no contrast, lower-cost, no extended in-bore periods, less infrastructure requirements) while achieving better image resolution and depth than current optical imaging approaches.

You're signed up!

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

-- The Imaging Wire team