- Color-Coded MRI: A new study in AJNR showed that color-coding lesion changes in MRI exams significantly improves brain metastases follow-up assessments. Two radiologists read 121 pairs of follow-up T1 MRIs (color-coded & conventional), achieving far higher diagnostic accuracy (91.3% vs. 74%), diagnostic certainty (Likert score: 4 vs. 2 out of 5), and interrater agreement (κ = 0.80 vs. 0.46) with the colorized exams. The radiologists also interpreted the color-coded exams much faster (51.8 vs. 74.2 seconds), although the conventional exams were faster when factoring in image processing time (91.5 vs. 79.4 seconds).
- Qure.ai’s qXR-BT 510(k): Qure.ai announced the FDA 510(k) clearance of its qXR-BT solution, which analyzes chest X-rays to automate clinicians’ breathing tube placement / measurement processes (including both endotracheal and tracheostomy tubes). The fact that qXT-BT automates a process that’s common, necessary, and somewhat mundane makes it the type of tool that even AI-hesitant radiologists might be open to.
- 68Ga-FAPI PET/CT for Gastric Cancer: A new study out of China detailed 68Ga-FAPI PET/CT imaging’s strong performance for gastric cancer diagnosis, suggesting that it could even replace 18F-FDG PET/CT (the current standard). The small study of 20 patients with gastric cancer found that 68Ga-FAPI PET/CT outperformed 18F-FDG PET/CT for primary tumor detection (100% vs. 74%) and tracer uptake (P < 0.05). 68Ga-FAPI PET/CT was also superior for both patient-based and lesion-based evaluations, with the exception of metastatic lesions in supradiaphragmatic lymph nodes and ovaries.
- Standardizing AI Development: A Nature review of 72 healthcare AI studies highlighted the need to adopt a standardized AI development framework, proposing a framework that includes: (1) data preparation, (2) model development, (3) model validation, (4) software development, (5) impact assessment, and (6) healthcare implementation.
- Whole-Body CT AI Incidentals: A team of German researchers showed how AI could be used to detect unknown pathologies in emergency whole-body CT scans. The researchers analyzed 105 emergency whole-body CTs using a prototype version of Siemens Healthineers’ AI-Rad Companion Chest CT, identifying patients with cardiomegaly (15 patients), coronary plaques (17), aortic ectasia (34), missed vertebral fractures (13), and suspicious lung lesions (2). However, the AI tool also produced a high number of false positives (especially w/ lung lesions and vertebral fractures).
- Positron & Neusoft’s PET/CT Alliance: Positron announced an alliance with Neusoft Medical Systems that will bring Positron into the North American PET/CT arena, after historically targeting the cardiac PET segment. Positron will rely on Neusoft Medical Systems to manufacture and distribute a new cardiac and oncology-focused Positron-branded PET/CT system (based on Neusoft’s NeuWise PET/CT).
- Loma Linda’s Liver CT Improvement: A new paper from Loma Linda University Medical Center detailed how they right-sized their abdominal CT utilization. After identifying particularly high use of abdominal CT with and without contrast for liver imaging, LLUMC (1) updated their liver imaging protocols to align with LI-RADS guidelines, (2) held clinical team education sessions to introduce the changes, and (3) shared their progress internally. As a result, LLUMC’s median monthly abdominopelvic CTs exams with and without contrast fell from 74 to 52 and their contrast-enhanced abdominopelvic exams increased from 11 to 37.5.
- AI3C: A pretty impressive list of U.S. healthcare and life science organizations announced the formation of Artificial Intelligence Industry Innovation Coalition (AI3C), created with the goal of transforming healthcare through responsible AI adoption. The AI3C and its members (including: Microsoft, Cleveland Clinic, Duke Health, Intermountain, Novant, Brookings Institution, and more) will develop tools and programs to drive AI development and adoption.
- SubtleMR Matches Full-Speed Brain MRI: A recent study from UCSF and Stanford showed that Subtle Medical’s SubtleMR image enhancement software allows 45% faster 3D brain MRI exams, without sacrificing image quality. The researchers performed 3D volumetric brain MRI exams on 32 patients at standard and 45%-faster speeds (sequences: 3D T1 precontrast, 3D T2 FLAIR, and 3D T1 postcontrast), processing the faster sequences with SubtleMR. Four neuroradiologists found that AI-enhanced scans were noninferior for all image quality metrics, while improving signal to noise ratio and restoring high spatial resolution of small structures.
- Probo Acquires REMETRONIX: Probo Medical continued its M&A spree, acquiring major imaging equipment installation and deinstallation service provider REMETRONIX. The acquisition gives Probo install / deinstall capabilities across the US and in the UK, and follows a string of acquisitions that significantly expanded Probo’s imaging sales and service capabilities over the last few years.
- POCUS Matches CXR for COVID: A new JUM study found that point-of-care ultrasound can be as effective as chest X-ray for diagnosing and managing COVID patients. The researchers performed POCUS and CXR exams on 479 ED patients with suspected COVID (87% tested COVID-positive) finding that POCUS findings with B-lines, irregular pleural line, and small sub-pleural consolidations were common among the COVID-positive patients (80%, 60%, 55%). These POCUS findings were non-inferior to CXR, which correctly identified roughly 82% of the COVID-positive patients.
