|
Simpler Radiology Reports, Fatal MRI Accidents, and Stationary CT
February 16, 2026
|
|
|
|
|
Together with
|
|
|
|
“[W]hy aren’t we providing all women with the best information available on their risk, early detection, and potential prevention of breast cancer and, no less, that of heart disease? If not now, then when?”
|
|
Eric Topol, MD, advocating for the routine clinical use of mammography AI.
|
|
|
The landmark MASAI study of mammography AI showed that ScreenPoint Medical’s Transpara AI algorithm can reduce radiologists’ workload while improving cancer detection. We talked to MASAI researcher Jessie Gommers about the latest MASAI results, this time focusing on interval cancers that surface between breast screening rounds.
|
|
|
|
Can large language model AI algorithms write simpler radiology reports for patients than clinicians? A study published in European Radiology found that LLM-produced reports were more readable, but there are areas of concern that will require fine-tuning.
Patients are taking greater interest in managing their own healthcare, requesting direct access to medical information like images and reports.
- That’s a good thing, but it creates challenges for healthcare professionals more used to communicating with other providers.
Taking the time to draft a report just for patients is a non-starter for many radiology professionals in a time of workforce shortages.
- But this could be an excellent use case for AI, especially the LLMs that have sprung up over the past few years.
So researchers from Germany tested three LLMs to draft patient-friendly versions of 60 radiology reports from X-ray, CT, MRI, and ultrasound modalities.
- The LLMs included the ubiquitous ChatGPT-4o, as well as two open-source LLMs (Llama-3-70B and Mixtral-8x22B) that had been deployed on-premises within their hospitals.
The authors wanted to know not only how well the LLMs performed in drafting patient reports, but also whether there were differences between the black-box ChatGPT 4o and the two open-source LLMs.
- The LLMs were instructed to generate layperson summaries at the eighth-grade reading level, preserving key clinical information.
In comparing original radiology reports to LLM-produced summaries, researchers found…
- Original reports had much lower ease-of-reading scores on the Flesch readability scale (17 vs. 44-46).
- Original reports were judged much less understandable on a five-point scale (1.5 vs. 4.1-4.4).
- The two open-source LLMs had higher rates of critical errors that could lead to patient harm (8.3%-10%), while ChatGPT 4o had no critical errors.
- Original reports had shorter total reading time versus LLM versions (15 vs. 64-73 seconds).
- There was no difference in understandability based on modality.
The findings on critical errors are particularly concerning.
- Clinicians may see on-premises open-source LLMs as having patient privacy advantages over cloud-based ChatGPT 4o, but such models may require more clinical oversight to avoid patient harm.
The Takeaway
The new study on LLM-generated patient radiology summaries is encouraging, pointing to a future in which a cumbersome task could be offloaded to generative AI algorithms. But much work remains to ensure patient safety and privacy before this can happen.
|
|
Radiology Automation Simplified
CARPL is an enterprise-grade radiology AI validation and deployment platform with 250+ AI applications across 85+ AI vendors that empowers healthcare providers to access, assess, and securely integrate imaging AI in their practice. Book a demo today.
|
|
The World’s Happiest Customers – 13 Straight Years
Sectra continues to top the charts when it comes to customer satisfaction. The company recently won seven Best in KLAS awards, including the 13th straight for its Sectra PACS enterprise imaging solution. Find out why on this page.
|
|
- Fatal MRI Accident in Finland: Government authorities in Finland last week released a report on the 2025 death of a patient while under anesthesia for MRI. The patient was referred for shoulder pain, and sedation was ordered because of claustrophobia. But the facility “did not have readiness to safely perform MRI scans under sedation” and the patient was given a dose equivalent to general anesthesia, causing respiratory depression. MRI safety expert Tobias Gilk believes the real story is that the accident received a formal government investigation – a practice the FDA should emulate.
- Child Dies During MRI Scan in India: Anesthesia is also believed to be behind the death of a child during an MRI scan in India. As reported by India Today last week, the six-year-old boy was given a “heavy dose” of anesthesia before the procedure at a private imaging center in northern India. His condition began to deteriorate and he lost consciousness. He died after being rushed to a nearby hospital.
- 3D Cardiac MRI Protocol Reveals Heart Defects: A new 3D volume rendering protocol for MRI displays both heart tissue and blood flow simultaneously, and could be useful for visualizing pediatric heart conditions. In a paper in Radiology: Cardiothoracic Imaging, Pennsylvania researchers described the protocol, which combines 3D imaging, 4D cine, and 4D flow cardiac MRI. In a small study with four patients, the technique enabled visualization of the myocardium and valves in less than one second, and supported simultaneous depiction of valve dynamics, regurgitation, and stenosis.
- New Flexible Pediatric MRI Coil: Philips launched a new flexible 3T MRI coil for pediatric applications through a collaboration with InkSpace Imaging. The Snuggle coil features a blanket-like design that wraps around the patient and a high-density array for high-resolution images. Snuggle recently received FDA clearance, and the companies believe it can reduce the need for repeat scans by making exams more comfortable for children.
- MRI Startup Wellumio Raises Funds: Dedicated brain MRI developer Wellumio raised $7.3M to commercialize its Axana scanner for stroke imaging. Axana is a cart-based 0.1T scanner that can be moved to the patient’s bedside for rapid assessment in the emergency department and other settings and speed treatment within the “golden hour” after stroke onset. Wellumio will use the funds for a Phase 2 clinical study to support an eventual regulatory application to the FDA.
- CCTA Plaque Measures Predict MACE: European researchers developed a set of features to characterize high-risk plaque on CCTA scans, and tested it for predicting major adverse cardiovascular events in European Radiology. The combination includes CAC scores ≥ 400, low-attenuation plaque, positive remodeling, and napkin-ring sign. Among 1.7k patients studied from the DISCHARGE trial, those with the features had almost 4X higher risk of MACE (HR = 3.81). If they also had obstructive CAD their risk was nearly 9X (HR = 8.78).
- Stationary CT Research Moves Ahead: Researchers from China continue to advance their work on a novel CT scanner with a stationary detector ring rather than the rotating gantries found on conventional systems. NanoVision Medical Technology’s system positions 24 stationary X-ray sources and 64 detector modules in two rings around the patient. In a study with 30 patients published in Radiology, observers rated stationary CT’s image quality for fine pulmonary structures 42% better. The findings echo research presented at RSNA 2024.
- Quality Control in Lung Cancer Screening: CT lung cancer screening has been a success story in Kentucky, generating a rapid shift toward earlier detection. In a new paper in CHEST, clinicians share their QUILS System 1.0 for quality control in LDCT screening with a focus on providing screening in diverse and low-resource areas. The program includes a quantitative evaluation system, an audit and feedback loop, online training tools, and coaching and technical assistance. They hope QUILS will help overcome barriers to wider lung screening adoption.
- Are Low-Resource Communities Ready for Lung Screening? As U.S. health systems ramp up CT lung screening programs, the question is whether historically under-resourced communities are ready. Researchers surveyed leaders from a Black community in Florida, finding “minimal” awareness of screening, with respondents noting limited outreach despite available services. Screening readiness is driven “less by hesitancy and more by invisibility,” highlighting the need for outreach to Black communities, which see later lung cancer diagnoses and higher mortality rates.
- Median Ties Up with Tempus: Hot off receipt of FDA 510(k) clearance for its eyonis LCS software for CT lung cancer screening, Median Technologies cemented an alliance with Tempus AI, which will distribute the solution through its Tempus Pixel platform. The Tempus software enables clinicians to analyze, track, and quantify lesions, a good fit for Median’s focus on detecting and characterizing lung lesions found on LDCT screening exams.
- AGFA Highlights Enterprise Imaging Contracts: AGFA HealthCare is highlighting several new U.S. contracts it received for its cloud-based enterprise imaging software. The company received a large contract for its Enterprise Imaging Cloud Teleradiology solution from a multistate radiology group that serves 11 sites in four states and wants to unify its software on a single cloud-based platform. AGFA also expanded its enterprise imaging installation with Tampa General Hospital in Florida. AGFA this month received three Best in KLAS awards for its enterprise solutions.
- HeartLung’s AI-CVD Predicts Heart Failure: The AI-CVD algorithm from HeartLung Technologies predicted heart failure more accurately than clinical criteria in a new study. Researchers used AI-CVD to extract additional data from CT-derived coronary artery calcium scans of 6.7k people without symptoms. Compared to the PREVENT-HF clinical criteria, AI-CVD had higher AUC at 10 years (0.83 vs. 0.79) and the algorithm performed consistently across age, sex, and race/ethnicity. AI-CVD got FDA clearance in December.
- Bracco Launches Contrast Injector in Europe: Bracco’s ACIST Medical Systems division is launching a new contrast injection system in Europe – just in time for next month’s ECR 2026. The ACIST Pro injector was granted the CE Mark under MDR rules and represents the next generation of iodinated contrast technology after the company’s ACIST CVi injector for cardiac cath labs and CT. It replaces manual injection with automated functionality, with clinical studies showing it reduces contrast use and rates of acute kidney injury, while saving an average of five minutes per case.
- Serac Gets Fast Track for SPECT/CT Tracer: A radiotracer for visualizing lung inflammation on SPECT/CT scans has been granted fast-track designation by the FDA. Serac Healthcare’s technetium-99m maraciclatide is being developed to diagnose interstitial lung disease, a condition that can cause progressive lung inflammation and fibrosis. The radiotracer is being tested in the PREDICT-ILD Phase 2 clinical trial, with clinical results expected later this year.
- Sectra Lands St. Luke’s Enterprise Imaging Deployment: Sectra deployed its cloud-based enterprise image management system at St. Luke’s University Health Network, which serves 16 campuses and over 350 outpatient sites in Pennsylvania and New Jersey. The installation integrates radiology, cardiology, ophthalmology, and other modalities into a single storage environment that streamlines clinical workflows.
- NIH Funding Still Favors Male Researchers: Despite some recent progress, NIH funding of radiology research is still heavily weighted toward men. A new study in Current Problems in Diagnostic Radiology reviewed 9.4k NIH radiology grants from 2009 to 2023, finding that 75% were awarded to male principal investigators, who also received 3.9X more total funding than women and higher average funding per grant ($704k vs. $536k). That said, the gender funding gap in total funding shrank from 7X in 2009 to 2.8X in 2023.
|
|
The Next Generation of Universal Remote Imaging
Step into the next generation of universal remote imaging with LUMINOS Q.namix R from Siemens Healthineers. Designed for intuitive operation and patient comfort, this award-winning system sets a new standard in fluoroscopy. Explore its precision and efficiency.
|
|
Discover AI-Powered Radiology Screening Solutions
Visit DeepHealth at ECR 2026 to learn more about how they are transforming population health and cancer screening with innovative AI solutions. Book a meeting today or stop by at Expo 506.
|
|
Leadership in Life Sciences
Quibim is committed to accelerating its development within the life sciences sector and strengthening collaborations with leading pharmaceutical companies. Learn about recent leadership developments that are moving the company forward.
|
|
- How AI Is Redefining Data Migration: Enlitic’s Migratek data migration services – combined with AI-enabled ENDEX data standardization – is changing the game for data migration projects. Discover how it can benefit you in this article.
- A Bold Transformation in Client Experience: Intelerad is transforming how it supports customers and partners with clients through a company-wide Client Obsession initiative. The company is making investments in new tools, technologies, and staff to remove friction and deliver value – find out how it works on this page.
- Putting Patients Back in Charge: In this episode of The Radiology Report, Medality’s Daniel Arnold sits down with Laurel Lemasters MD, founder of Imago MRI, where patients have clear access, transparent pricing, faster answers, and a radiologist who still believes in being a doctor first.
- Radiology Case Report: A man in his 50s presented with syncope with minor head trauma and unassociated risk factors. Find out how MRI helped provide a diagnosis in this case study.
- Redefining What’s Possible in MRI: Since 2018, BlueSeal technology from Philips has redefined what’s possible in MRI – from a bold engineering idea to the world’s first fully sealed, helium-free 1.5T MR system. Learn about the technology behind the innovation on this page.
- Powering Reporting Progress with Efficiency and Interoperability: Radiologists are looking to maximize their efficiency through new reporting tools that integrate easily with their PACS and RIS. Watch this Imaging Wire Show interview with Kailo Medical’s Jason Mercieca and Dieter Smith about new developments in reporting.
- AI-Empowered CT Workflow: CT systems from United Imaging are designed for high image quality and low dose, and their AI-empowered workflow enables fast and reproducible positioning, helping you image patients with confidence. Learn more on this page.
- Innovation for the Imaging Workflow You Deserve: Radiologists are overwhelmed by complex systems and growing caseloads. Discover how innovative workflow orchestration from Merge integrates various PACS, RIS, and EMR systems into a single AI-driven workspace.
- Uncovering Hidden Cardiac Amyloidosis with Echo: Us2.ai’s automated echocardiography analysis technology was used in the AI-Screen-CA study to surface previously unrecognized cardiac amyloidosis in an unselected clinical population. Learn about its transformative impact on this page.
- A Best in KLAS Hat Trick: AGFA HealthCare was named Best in KLAS in three enterprise imaging segments this year: PACS under 300k studies, universal viewer, and vendor-neutral archive. Find out what makes customers keep coming back on this page.
- Enterprise Imaging Done Differently: Legacy radiology solutions were not designed to carry healthcare organizations into the future. From their first line of code, Mach7 Technologies was designed to meet the imaging needs of the entire healthcare enterprise. Learn more about their unique approach today.
- Cut Repetitive Manual Tasks: Discover how Rad AI’s radiology reporting software helps you speak less and say more by reducing dictation times up to 50% and words dictated up to 90%. Join 9 of the top 10 radiology practices using Rad AI to improve efficiency.
- Shifting the Stage in Lung Cancer Screening: Watch this video from Riverain Technologies to learn how their ClearRead CT solution for lung cancer screening can drive enrollment, earlier detection, and seamless management of incidental findings.
- An AI-Powered Solution for Fracture Detection: Gleamer’s BoneView provides radiologists and clinicians with an instant and automatic second reading of trauma bone X-rays, fully integrated into the reading workflow. See how it works today.
- The Workstation of the Future: A dedicated team of radiologists shapes every aspect of the functionality and design of MosaicOS from Mosaic Clinical Technologies. Learn how every feature was built to eliminate distractions, amplify focus, and enhance the radiologist experience.
- One Viewer for All: Achieve ultrafast image interpretation with greater efficiency and precision with Visage 7 from Visage Imaging. Visage’s one viewer philosophy enables all end users – from radiologists to clinicians – to access powerful tools based on clinical need.
|
|
|
|
|
