- Watson Health on the Block: Axios reports that IBM is planning to sell Watson Health to one of several private equity firms or a “strategic buyer” by the end of January. IBM would reportedly gain $1B from the sale, while exiting an unprofitable business that still has some historical PR baggage. Despite the media’s infatuation with Watson Health’s previous AI challenges, it still has a valuable suite of products (many of which are only adjacently AI-related), some really sharp team members, and a solid customer list.
- MUSC’s Ambulance MRI: MUSC Health recently performed what might be the first ambulance-based MRI, as they evaluate adding mobile MR to their stroke ambulance fleet. MUSC’s demo used Hyperfine’ Swoop MRI system, finding that the system created diagnostic quality scans despite vehicle motion, and suggesting that ambulance-based MRIs could improve care coordination and cut treatment times (even compared to mobile CT). MUSC now plans to document its initial ambulance MRI demonstration and might launch a county-wide study.
- Low Value Pediatric Imaging: A new JAMA study (n = 49 children’s hospitals, 1.1M encounters) highlighted imaging as a leading driver of low-value pediatric care. Imaging’s low-value problem was greatest in pediatric EDs, with imaging exams representing all of the top seven low-value cost drivers (led by CT for abdominal pain & CT for minor head injuries). Imaging’s low-value role wasn’t as bad within the study’s hospital cohort, where imaging procedures represented “just” three of the top-10 low-value cost drivers (CXR for bronchiolitis, CXR for asthma, CT for abdominal pain).
- Milvue’s €8M: French AI startup Milvue completed an €8M Series A round from an investor group that included 58 radiologists, revealing plans to invest in ongoing product development, regulatory approval efforts, and its expansion into North America. Milvue’s core product analyzes X-rays for both osteoarticular and pulmonary pathologies, supporting triage and diagnosis.
- Siemens’ CXR Nodule AI Effectiveness: A new JAMA study found that Siemens Healthineers’ AI-Rad Companion Chest X-Ray algorithm improved radiologists’ pulmonary nodule detection accuracy. The researchers had nine radiologists review 100 CXRs with and without AI support, finding that the AI tool improved the group’s average nodule detection accuracy (75% vs. 69%), sensitivity (55% vs. 45%), and specificity (95% vs. 93%). Less experienced rads achieved a far greater AI sensitivity boost than senior rads (+12% vs. +9%), while the AI tool had a similar impact on the junior and senior rads’ specificity (+4% & +4%).
- TTG’s PE Acquisition: Imaging equipment and isotope sales/service company, TTG Imaging, was acquired by Sentinel Capital Partners, starting TTG’s “next phase of growth.” TTG Imaging already seemed to be in growth mode, completing nine acquisitions since early 2019, and its new PE-backed status suggests that it will accelerate its expansion going forward.
- A Touchless Future: A new Radiography Journal study suggests that most radiographers will be ready to embrace “touchless” controls (e.g. gesture and voice commands) as these features are integrated into future scanners. The survey of 155 radiographers and radiologic technologists (including students) revealed that 85% could see themselves using touchless imaging equipment features, while the majority view voice and gesture controls as “key in improving exam efficiency” (82% & 65%).
- COVID Cancellations: The latest COVID surge is once again forcing hospitals to delay or cancel elective procedures (and related imaging exams). A number of major hospitals have officially paused elective procedures (e.g. Cleveland Clinic, VCU Health, NIH Research Hospital, Advocate Aurora, U of Washington) and we’re now seeing state governments assuming a larger role in elective shutdowns, including Massachusetts’ statewide order to pause/cancel electives and government-enforced pauses at over 20 New York state hospitals.
- FDA Validation Gaps: Some FDA-cleared imaging AI/ML algorithms might be well-validated, but potential AI adopters would have a hard time finding sufficient validation data on the FDA’s searchable AI database. That’s from a recent Academic Radiology study that reviewed 118 algorithms on the FDA database, finding that 17 tools were posted without any validation claims or data, only 9 had validation datasets with more than 1,000 patients (most <500), and most validations lacked demographic data. The authors warned that potential AI adopters would struggle to properly assess AI tools within the FDA database, and encouraged AI firms to make additional validation data available through other non-FDA resources.
- DRF AI Accuracy: A new Journal of Digital Imaging study detailed a deep learning model that can diagnose distal radius fractures (DRFs) in wrist X-rays with extremely high accuracy. The researchers had the DL model and three orthopedic hand surgeons diagnose 150 frontal and lateral X-rays, finding that the model’s accuracy, sensitivity, specificity, and AUC (99.3, 98.7, 100, 0.993) were equal to or better than the orthopedic surgeons.
- Healthcare M&A Volumes Up: PwC released a report showing that US health services M&A activity was up 56% in 2021. Although this M&A trend goes well beyond medical imaging, radiology practice acquisitions definitely contributed to 2021’s surge in physician medical group acquisitions (+119% to 407) and imaging centers were included among the many “labs, MRI, and dialysis” center acquisitions (+62% to 102).
