|
RadGPT Simplifies Reports, CTC Vs. DNA Test, and Vendor Earnings
June 12, 2025
|
|
|
|
|
Together with
|
|
|
|
“Radiologists shouldn’t have to spend 10 minutes searching the medical record to figure out why a patient is here for a cardiac MRI. AI should do that.”
|
|
Tessa Cook, MD, PhD, of Penn Medicine in this week’s RadEqual/Intelerad webinar on AI and healthcare.
|
|
|
|
When it comes to informing patients of their imaging results, radiologists are caught between a rock and hard place. A new study in JACR shows how generative AI can help by drafting patient-friendly reports that are simple but accurate.
Patients must be informed immediately of their medical results according to a 2021 final rule under the 21st Century Cures Act that prevents medical information blocking.
- And while the technology exists to do that through tools like email and electronic patient portals, rapid notification can create confusion because the language physicians use to communicate with each other isn’t easily understood by anyone outside medicine.
Sure, radiology reports could be rewritten manually for patients, who typically read at about the eighth-grade level.
- But given today’s workforce shortages, who’s going to do that?
Generative AI and large language models offer a solution. In the new JACR paper, researchers from Stanford University led by senior author Curtis Langlotz, MD, PhD, described their development of RadGPT, an LLM designed to improve patient communication.
- To develop RadGPT, researchers started with OpenAI’s GPT-4 model and the RadGraph concept extraction tool to create an LLM that analyzes patient radiology reports and generates concept explanations and question-and-answer pairs.
How well did RadGPT work? The researchers tested it on 30 radiology reports generated at Stanford from 2012 to 2020, including different modalities and clinical applications.
- The LLM was asked to generate reports at a fifth-grade reading level (the level recommended by the Joint Commission for patient-facing healthcare materials).
Five radiology-trained physicians then rated the quality of RadGPT’s responses, finding …
- The average rating of RadGPT-generated concept explanations was 4.8 out of 5.
- 95% of concept explanations had an average rating of 4 or higher.
- 50% of concept explanations were rated 5, the highest possible rating.
- Questions and answers generated by RadGPT were also rated highly, with an average rating of 3.0 on a three-point scale..
The Stanford researchers told The Imaging Wire that their goal is to make RadGPT more widely available as part of a prospective evaluation with real patients.
- They are also developing a user-friendly interface in which patients can receive hyperlinked radiology reports.
The Takeaway
RadGPT and solutions like it fill a desperate need for tools that can save time for radiologists while helping patients better understand their reports and get more engaged in their care. The next step is to get technology like this into the hands of practicing radiologists.
|
|
Visit United Imaging at SNMMI 2025
Discover how United Imaging’s molecular imaging innovations are driving the cure for all patient types at SNMMI 2025. Hear from country music star Joshua Ray Walker, who will tell the story of his colon cancer diagnosis with United’s PET/CT technology.
|
|
Revolutionize the Reading Experience
With intelligent automation and AI‑powered workflow, PowerScribe One from Microsoft allows radiologists to generate and communicate high‑quality, consistent reports – and get more done in less time.
|
|
Unprecedented Insights Made Possible with AI
With the largest normative dataset of whole-body imaging in the world, Prenuvo’s AI researchers partner with the best academic minds to understand – like never before – what “normal” aging means. Learn about their work today.
|
|
- CTC Beats Stool Test for Colon Cancer: With Medicare paying for CT colonography, a new modeling study in Radiology found CTC is more effective than stool DNA testing for colon cancer screening. Researchers examined several screening strategies in a hypothetical population of 10k people; CTC with three-year follow-up for small polyps 6-9 mm and referral to optical colonoscopy for large polyps ≥ 10 mm was most cost-effective versus DNA testing (per-person cost of $3,913 vs. $6,011) while also reducing cancer incidence (70% vs. 49%).
- Infection Control Visualizes Viruses in VR: A virtual reality-based infection control program developed at Mass General Brigham educates hospital personnel on proper cleaning and disinfection techniques for equipment like ultrasound probes and mobile X-ray systems. The program employs head-mounted displays to help users visualize invisible contamination and has a gamified format that improves user interaction through tasks like cleaning a point-of-care ultrasound device. In a pilot study, 89% of users reported a positive experience.
- 4DMedical Powers Olympus Emphysema Screening: Endoscopy vendor Olympus this week released an emphysema screening program called SeleCT Screening that’s powered by lung imaging technology from 4DMedical. SeleCT Screening reviews existing CT scans to identify patients with severe emphysema who might benefit from bronchoscopic lung volume reduction with tools like Olympus’ SpirationValve System. The partnership is a major coup for 4DMedical given Olympus’ market presence in the bronchoscopy field.
- Cardiac MRI Shows Long-Term COVID Damage: Long-term COVID-19 infection can damage heart function in ways visible on cardiac MRI scans. Swedish researchers in JAMA Network Open scanned 59 people, finding that compared to healthy controls those with long COVID had reduced mean stress perfusion (2.80 vs. 3.43 mL/min/g), with lower left ventricular function as measured by mean global longitudinal strain (-17% vs. -19%) and mean global circumferential strain (-16% vs. -19%). Cardiac MRI could be detecting signs of coronary microvascular dysfunction.
- MRI of White Matter and Cognitive Decline: Brain MRI scans revealed changes in white matter that correlated with declining cognitive function. In a new study in JAMA Neurology, researchers analyzed data from 4.5k people who got diffusion-weighted MRI scans and performed cognitive tests. MRI showed that people with higher levels of white matter free water had a greater decline in cognitive function, with the association greatest for free water in limbic tracts like the cingulum and fornix.
- PET Links Mood Disorders to Future Dementia: Mood disorders appearing in people later in life could be a precursor to neurodegenerative diseases like Alzheimer’s. Researchers in Japan used PET to scan 99 people, finding that those with late-life mood disorders were nearly 5X more likely than healthy controls to have gray matter accumulation of tau and beta amyloid, two substances linked to Alzheimer’s disease. Autopsy results of 208 subjects confirmed the link between late-life mood disorders and tau pathology.
- Telix Launches PSMA-PET Prostate Radiotracer: Telix Pharmaceuticals this week launched its Gozellix gallium-68-labeled radiotracer for PSMA-PET studies in the U.S. Gozellix received FDA approval earlier this year and is indicated for detection of PSMA-positive lesions in men with prostate cancer who have suspected metastases or biochemical recurrence based on PSA tests. Telix is touting the tracer’s six-hour half-life, which will make distribution easier than with other PSMA-PET agents.
- Positrigo Raises $9M for PET Launch: Swiss dedicated brain PET developer Positrigo raised $8.5M that the company will put to work commercializing its NeuroLF scanner. Positrigo developed NeuroLF as a more economical alternative to whole-body PET scanners for neurological conditions like Alzheimer’s disease, brain tumors, and epilepsy, especially as new pharmaceutical therapies create demand for diagnosis and treatment monitoring. Positrigo announced 501(k) clearance for NeuroLF in July 2024 and the CE Mark in October.
- Contrast Use Falls at Cardiac Cath Labs: Interventional cardiologists are using less iodinated contrast for percutaneous coronary intervention procedures. Writing in a study in JACC, researchers found that for 3.1M patients undergoing PCI from 2018 to 2022, mean contrast volume per procedure fell 11% (168 vs. 150 mL). But most of the lower use came from just 21% of physicians, while 77% didn’t change their use patterns at all and 1.9% actually increased their use per procedure. Iodinated contrast has been linked to acute kidney injury.
- CIVIE Debuts On-Demand Reading Platform: CIVIE this week announced the launch of its RadPod platform of on-demand radiology workflow software. CIVIE said the platform allows radiologists to log in and read studies on their own schedule – and get paid for it. RadPod includes AI-powered solutions for improving radiologist efficiency, including PACS, real-time voice recognition, worklist, an automated payment system, and automated quality assurance. CIVIE was formed earlier this year after PE firm WindRose Health acquired Collaborative Imaging and rebranded the firm.
- Merge Adds Cloud to Wash. Health System: Merge landed a contract to deploy its cloud-based enterprise image management solutions at MultiCare Health System, a network of 13 hospitals and over 300 clinics in the Pacific Northwest. The deployment is an expansion of an existing contract that started in 2024 for Merge Imaging Suite, and has been expanded to include cloud-based PACS. MultiCare is building an image exchange network throughout the region.
- CARPL Adds AI for ICH Detection: AI platform company CARPL.ai has added an AI algorithm from South Korean developer PurpleAI for detecting intracranial hemorrhage on non-contrast head CT scans to its platform offerings. Hyper Insight – ICH got FDA 510(k) clearance in 2024 to triage and notify radiologists of suspicious findings on CT studies.
- AI Helps Fracture Detection: A meta-analysis of eight studies on Gleamer’s BoneView fracture detection AI algorithm revealed positive findings. The review in European Journal of Radiology found that BoneView improved sensitivity in most studies, from radiology residents in Meetschen et al (77% vs. 58%) to senior radiologists in Dell’Aria et al (87% vs. 69%), with no difference in specificity. Radiologist reading speed was faster in three studies by a mean of 5.3 to 16 seconds, and was no different in one study.
- China Effect Persists for Imaging OEMs: The close of vendor earnings season sees most imaging companies still feeling the effects of slower sales in China as they brace for the impact of tariffs. Revenues were reported for Siemens Healthineers Imaging division (+8.7% to $3.75B), GE HealthCare Imaging and Advanced Visualization Solutions divisions (+2.7% to $3.38B), Philips Diagnosis and Treatment division (-4% to $2.24B), Fujifilm healthcare division (+9% to $2.04B), Canon Medical Systems (+2.3% to $901M), RadNet (+9.2% to $471M), Hologic Breast Health division (-9.7% to $356M), Guerbet (-7.1% to $205M), Varex medical division (+3.4% to $154M), Konica Minolta healthcare division (-6.5%% to $161M), and Butterfly Network (+20% to $21M).
|
|
Going Global with Enterprise Imaging
Merge is going global by bringing its enterprise imaging solutions to new regions, marketplaces, and customers. Learn more about its recent initiatives in Puerto Rico, the U.K., Latin America, and Canada.
|
|
Synoptic Reporting of Liver Masses
Contrast-enhanced ultrasound exams for focal liver lesions can be challenging to report in at-risk patients. Learn how Kailo Medical’s synoptic reporting solutions can make reporting easier while improving LI-RADS categorization.
|
|
Reach New Heights with Enterprise Imaging Cloud
Embrace the potential of AGFA HealthCare’s Enterprise Imaging Cloud, a fully managed Software-as-a-Service (SaaS) solution that will transform the handling, storage, and accessibility of medical imaging data. Learn how EI Cloud can help you today.
|
|
- Advancing AI-Driven Data Migration: Enlitic has joined forces with GE HealthCare to power the data migration feature in GE’s newly announced Genesis cloud portfolio. Learn how Enlitic’s AI-driven data migration facilitates large-scale transfers of high-quality medical imaging data.
- All Your Data in One Place: Mach7’s Vendor Neutral Archive (VNA) is a powerful data management and workflow orchestration technology built for the future. Learn how it can help you drive better patient care by centralizing all your data in one place.
- The Benefits of Operational AI: Explore the transformative potential of operational AI in healthcare in this on-demand webinar hosted by Blackford. Learn from the company’s partners how AI can help your practice operate more efficiently.
- AI Acceleration without the Hype: The Visage AI Accelerator program is an innovation magnet – an end-to-end AI solution that bridges research and diagnostic imaging on the same, unified platform. Learn how Visage 7 provides synergistic, smart workflows and flawless third-party AI integration.
- 95% Cost Reduction by Closing Sharing Gaps: Valley View Hospital eliminated image sharing friction by automating access for patients and out-of-network providers. The result? A 95% reduction in operational costs and a global reach that keeps care moving. Learn how PocketHealth helped Valley View streamline access and cut overhead.
- The Evolution of PACS: PACS has become the backbone of efficiency in modern radiology, evolving into sophisticated enterprise-wide imaging solutions. Discover how Intelerad’s IntelePACS can deliver the innovation, intelligence, and flexibility your imaging strategy needs by booking a demo.
- AI Tools for Lung Cancer Screening: CT lung cancer screening is gaining momentum around the world. Learn about AI-based nodule detection tools that can improve the accuracy of low-dose CT scans in this video from Riverain Technologies.
- A New Solution for Chest AI: Gleamer’s ChestView AI solution is a computer-aided detection (CADe) model cleared by the FDA for simultaneously detecting multiple findings on chest X-ray. Discover how it enhances explainability and confidence compared to traditional triage-focused CADt solutions.
- Maximize New CCTA Reimbursement with Philips CT 5300: Coronary CTA is the preferred noninvasive exam for detecting and ruling out CAD. Updated guidelines and improved reimbursement reinforce its value for stable or atypical chest pain. The CT 5300 from Philips delivers fast, high-quality, low-dose cardiac imaging using AI and zero-click motion correction.
- Clarity, Speed, and Confidence for MRI Efficiency: Radiologists have used a variety of methods to improve efficiency, but many of these methods come with drawbacks. Find out in this article how SpinTech MRI takes on the challenge of MRI efficiency with its STAGE software.
- Radiology Advocacy, Challenges, and Innovation: In this episode of The Radiology Report podcast, Medality CEO Daniel Arnold sits down with Reno Radiological Associates CEO Anthony Dispenziere to discuss the business behind independent radiology practice and what it takes to stay competitive in an increasingly complex healthcare landscape.
- Give Patients a Clear Path to Accessing Medical Data: Clearpath is a simple integration that empowers digital delivery of medical records and images. Request a demo today to find out how you can ditch the disc and give your patients and third parties instant access to digital data.
- Say Goodbye to On-Premises Costs: Free up resources with cloud-based solutions from Optum for medical imaging. Visit this page to see how you can say goodbye to on-premises costs!
- An Update from DeepHealth: What are the latest developments at DeepHealth? Check out this video interview with company executives Kees Wesdorp and Niccolo Stefani, who discuss the company’s recent highlights.
- AI Echo in the Netherlands: Us2.ai is now being used at two leading hospitals in the Netherlands, bringing advanced AI technology to echocardiography patients in the country. Learn more about these two milestone deployments on this page.
- A New Benchmark for Tomo Imaging: There’s a new benchmark for digital breast tomosynthesis 3D images with MAMMOMAT B.brilliant from Siemens Healthineers. The system’s 50° wide-angle tomosynthesis helps you achieve excellent outcomes for your patients, radiologists, and breast care professionals.
- Presenting Unboxing AI: Check out CARPL’s video series, Unboxing AI, featuring experts discussing AI and its future in radiology. The next episode on June 12 features Jin Wook Choi, MD, PhD, of PurpleAI – reserve your seat today.
- Integrating AI into Clinical Practice: AI has the potential to revolutionize healthcare, but it requires a collaborative effort between clinicians and AI experts. In this micro-learning course from Calantic by Bayer, learn about the latest developments in healthcare AI.
|
|
|
|
|
