#365 – The Wire

  • Samsung’s CXR Nodule AI: Samsung NeuroLogica announced the FDA clearance of its Auto Lung Nodule Detection AI tool, which is intended to help physicians detect 10mm-30mm lung nodules in CXR scans, and will come embedded in Samsung X-ray systems. Samsung has more AI on the way, revealing that its collaboration with Vuno will bring more chest CADe tools.
  • CAC Scores, Better with Age: A new JAMA study revealed that physicians might not be able to rely on CAC scores to rule out obstructive coronary artery disease (CAD), especially among younger patients. Analysis of 23,759 symptomatic patients with obstructive CAD found that many younger participants had no coronary artery calcium (Zero CAC = 58% of <40yrs, 34% of 41-49yrs, 18% of 50-59yrs), while these <60yr-old patients without CAC still had higher risks of heart attack and all-cause death.
  • Repair Rights Granted: After years of fierce OEM and ISO lobbying, the U.S. Copyright Office ruled that 3rd party service providers and device owners can legally access medical device software, as long as it’s necessary to diagnose, repair, or maintain a device. Unsurprisingly, OEMs and servicers have very different feelings about this ruling.
  • LDCT for HNC Survivors: A new JAMA study found that head and neck cancer survivors with a long smoking history have a 2.5 times higher rate of lung cancer than smokers who never had head and neck cancer. The study also found that low-dose CT screening was more effective than CXR screening for these patients, supporting their conclusion that regular LDCT screening is particularly important for HNC survivors. These screenings could also allow earlier detection of future head and neck cancers. 
  • GE & Circle NVI’s Stroke AI: GE Healthcare will integrate Circle Neurovascular Imaging’s stroke AI tools into the GE FastStroke processing platform, adding Circle NVI’s AI-based stroke triage and detection tools to FastStroke’s automated stroke CT workflow and communication features. Circle Neurovascular Imaging is a subsidiary of longtime GE partner, Circle Cardiovascular Imaging.
  • CXR’s Post-COVID Value: A new study out of Italy highlighted chest X-ray’s value for monitoring recovering COVID patients. The prospective study had 119 COVID patients undergo CXR exams three months after their hospital discharge, finding that roughly half of the patients still had lung abnormalities, and patients with higher CXR scores also had more severe cases while hospitalized.
  • AI Principles: The US FDA, Health Canada, and the UK MHRA published their 10 guiding principles for medical device “Good Machine Learning Practice,” setting the foundation for future international AL/ML regulatory collaborations (e.g. creating standards, resources, guidelines). The principles largely target: applying clinical expertise to product design, following software/security best practices, following training/evaluation/monitoring best practices, building for specific medical use cases and for AI+human operation, and providing users all necessary info. 
  • DBT Waste: A new Clinical Imaging editorial called for the NCI to cancel their $100m DBT trial (TMIST), suggesting that its unnecessary (DBT is already widely adopted & effective) and poorly designed (too small; should focus on reducing mortality; should include 40-44yr old women). Instead, the editorial argued that this funding should be used to improve breast cancer detection or for evaluating breast MRI. This isn’t the first criticism of the TMIST trial, reportedly leading to the NCI considering canceling it last year.
  • Atlantic Upskills: New Jersey’s Atlantic Health System (6 hospitals, 18k employees) announced a pair of interesting new radiologic technologist and pharmacy tech training programs intended to upskill its workforce. The radiology “tech school” will be offered to 120 current trained/certified RTs each year, allowing them to gain certifications for CT, MRI, mammography within 5 to 10 weeks, while giving Atlantic a multi-modal imaging workforce.
  • Pelvic CT Misreads: Radiologists commonly misread pubic root fractures as more serious anterior column fractures in pelvic CT scans, leading to unnecessary scans and evaluations. That’s from a new Clinical Orthopedics and Trauma study that analyzed 155 patients with pelvic or acetabular fractures at a single institution (83 & 72), finding that radiologists mistook 85% of the pubic root fractures as anterior column fractures.
  • AZmed Funding: French imaging AI startup AZmed secured a €3.5m funding round (total now €4.4m) that it will likely use to commercialize its Rayvolve AI tool (fracture and chest abnormality detection) and to support future product development.

You're signed up!

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

-- The Imaging Wire team