FDA clearance and CE marking are both essential for market access and regulatory compliance of AI radiology tools. These evaluations confirm that a device is safe, performs as intended, and complies with required quality standards.
Yet, in practice, these approvals tell only half the story. That’s because they don’t guarantee that the AI tool will fit into your workflow, be used by your radiologists, and give you a good ROI.
So the real question is no longer “Is it cleared?” but “Does it work in my practice, for my patients, with my team?”
In this blog, we’ll cover what these approvals mean and five things to consider when evaluating FDA or CE-marked AI tools. We’ll also highlight how tools within the Rayvolve® AI Suite meet these expectations while also making sure they perform well in real-world settings.
What are FDA clearance and CE marking in medical imaging AI?
FDA clearance in the United States
The US Food and Drug Administration (FDA) classifies medical devices into three categories based on risk. Most imaging AI systems fall under Class II, meaning they require a 510(k) or De Novo submission.
- 510(k) clearance demonstrates “substantial equivalence” to an already cleared device (called “the predicate”).
- De Novo applies when there’s no existing equivalent, requiring stronger evidence of safety and effectiveness.
FDA 510(k) approval is quite rigorous and requires extensive documentation, testing, and interactions with the FDA.
But once the tool is cleared, FDA clearance signals that it meets baseline safety and performance standards. While it sets a critical foundation, it's still not a complete measure of clinical reliability. We’ll explain more about this later in the blog. But first, let’s also understand CE marking.
CE marking in Europe
AI medical devices in Europe are regulated under the Medical Device Regulation (MDR). Each manufacturer seeking to obtain a CE mark must complete a conformity assessment that evaluates:
- Clinical performance and safety through a Clinical Evaluation Report (CER)
- Risk management, usability, and cybersecurity
- Compliance with quality system requirements under the current version of the ISO 13485
The CE framework emphasizes continuous monitoring and ongoing clinical evaluation, ensuring the product continues to perform in real-world use.
While FDA clearance or CE marking is critical for any AI tool used within your healthcare center, this does not mean that it will necessarily be adopted in daily practice. For instance, there are additional metrics you should consider when evaluating these AI tools.
How to evaluate a radiology AI tool? 5 things to consider
Published, peer-reviewed studies highlight that the tool performs reliably across diverse patient populations and imaging conditions.
When reviewing AI vendors, look for:
- Multicentric studies showing generalizability
- Metrics such as sensitivity, specificity, and AUC
- For quantitative tools such as bone age estimation and orthopedic measurements, regression metrics like mean absolute error (MAE)
- Reader studies showing how AI impacts radiologist performance
Consider the study from Luiken et al. (2025), “Evaluation of commercial AI algorithms for the detection of fractures, effusions, and dislocations on real-world clinical data: A prospective registry study,” for example. It is an independent study that examined three popular AI radiology tools on the market.
Among the evaluated tools, AZtrauma, a module within the Rayvolve® AI Suite, achieved the highest performance for the detection of all fractures (AUC 84.88%, sensitivity 79.48%), led in examinations with osteosynthesis material and in cases with multiple fractures, and recorded the highest AUC for joint dislocation.
Such evidence gives radiologists confidence that the tool performs well when compared to its counterparts.
Workflow integration and usability
It is important to note that any AI tool cleared by the FDA or bearing a CE mark is only beneficial to you if it can easily integrate with your current systems and does not interfere with radiologists’ day-to-day workflows.
When considering the integration aspects of AI tools, you should look at:
- Compliance with PACS/RIS systems and DICOM standards
- Ease of use: zero or minimal clicks on the AI tool interface will keep you, the radiologist, in control of the image
- Human-in-the-loop design that allows easy verification of AI results
- Flexible deployment options (fully on-premises, VPN-secured connection to cloud, or a hybrid model) so your IT team can align the setup with data-governance and latency requirements
If it takes your team an extra few hours per week just to use an AI tool, it’s probably not worth the effort. That’s one of the reasons why Rayvolve® AI tools can seamlessly integrate into your current imaging system, with no need for additional log-ins or switching between platforms.
Rayvolve® AI results are displayed within the PACS interface, so your team can focus on interpretation rather than dealing with AI tool interface management. The AI tools are designed to augment the work of the radiologist, not replace the radiologist.
Analyze cost and return on investment (ROI)
It's critical for organizations to create detailed KPIs to measure how successful AI is following implementation in order to accurately determine its impact on revenue.
For instance, how much faster is the turnaround time? Has the AI tool been successful in helping to identify previously missed diagnoses? Has it decreased overtime costs for staff?
With these types of measurable metrics, you can then evaluate the success of the AI after implementation using the established pre-AI performance metrics to gain a true understanding of the ROI associated with using AI in your organization.
Real-world deployment and user feedback
The validation of clinical applications identifies whether a diagnostic tool works under controlled testing (research), but ultimately, it is through real-world implementation that you will determine whether such tools are useful in your facility.
For this, look for transparent post-market performance data and peer feedback from other hospitals or radiology groups.
Practical indicators you can consider include:
- Number of active installations
- User satisfaction and training requirements
- Support responsiveness and update frequency
AZmed, for instance, has developed tools that are currently used by over 2,500 healthcare facilities around the world, including trauma centers, emergency departments, and teleradiology networks.
Below is a statement from Henrik Michaely, Owner and Chief Radiologist at MVZ Radiologie Karlsruhe, regarding the use of the Rayvolve® AI Suite:
“We use Rayvolve® for the detection of fractures and thoracic pathologies. The solution seamlessly integrates into our local PACS and our workflow. Thanks to its results, our radiologists can significantly speed up their overall reporting while simultaneously increasing their accuracy. We therefore consider it our AI-based second opinion, which boosts our overall quality and performance.”
This reflects real-world reliability.
Why routine radiography benefits most from regulated AI
Standard radiographs remain the most important and frequently used modality in clinical imaging. This is due to the relatively low dose to the patient, low cost, and universal availability.
But this access comes at a cost.
Most radiologists experience burnout and have to deal with heavy workloads. This, in turn, increases the risk of fatigue-related errors. AI radiology tools help here by decreasing their manual workload, so radiologists can focus on interpretation.
Yes, FDA or CE markings matter, but the choice also depends on whether the tool can consistently support fast, accurate, and reliable reads across varied cases.
When assessing AI for routine radiography, look for tools that:
- Handle high volumes efficiently without slowing workflow.
- Deliver accurate detection across different types of exams.
- Provide clear, structured outputs that can be audited.
- Integrate easily into PACS and reporting systems.
AZmed’s Rayvolve® AI Suite includes four regulated tools designed with these priorities in mind:
- AZtrauma: an FDA-cleared tool that detects fractures and a CE-marked tool that detects fractures, dislocations, and joint effusions to support trauma and emergency care.
- AZchest: CE-marked chest X-ray AI that identifies major lung and cardiac findings. In the U.S., two FDA clearances cover lung nodule detection and triage for pneumothorax and pleural effusion.
- AZboneage: This CE-marked tool calculates pediatric bone age automatically using the Greulich & Pyle method.
- AZmeasure: Characterizes osteo-articular geometries, providing precise measurements for orthopedic evaluations. This is also CE-marked.
Each module supports real-world efficiency, consistency, and compliance. Read on to know more.
AZmed: FDA- and CE-marked AI tool for routine radiography
Rayvolve® is a suite of tools for radiology that use AI and are designed to integrate into radiology workflows.
The algorithms have been validated through clinical research and are currently deployed in over 2,500 healthcare facilities globally. The clinical studies conducted show a clear difference in the amount of time saved, alongside a high level of consistency and accuracy in diagnosis across use cases such as trauma X-rays, chest imaging, orthopedic assessments, and pediatric bone age.
We believe that the initial step toward adopting an AI system is gaining regulatory approval.
The ultimate benefit of AI comes from reliable validation, ease of use, and consistent performance over time.
Hospitals utilizing the software report faster response times, increased consistency in results, and greater confidence while working in on-call or emergency settings. In addition, this software works directly within existing PACS workflows, allowing radiologists to maintain control of their own interpretations.
Here’s how Navid Faraji, MD, MSK radiologist at Cleveland University Hospital, sums it up:
“Rayvolve® demonstrated high stand-alone accuracy, aided diagnostic accuracy, and decreased interpretation time. When extrapolated over an entire population, one can see quickly how using this tool can really help decrease medical errors and healthcare costs.”
Book a call with our experts to find out why hospitals and radiologists choose AZmed tools over other FDA and CE-marked tools and how you can do the same.
Regulatory information
EU - Rayvolve: Medical Device Class IIa in Europe (CE 2797) in compliance with the Medical Device Regulation (2017/745). Rayvolve is a computer-aided diagnosis tool, intended to help radiologists and emergency physicians to detect and localize abnormalities on standard X-rays.
US - Medical devices Class II according to the 510K clearances.
Rayvolve: is a computer-assisted detection and diagnosis (CAD) software device to assist radiologists and emergency physicians in detecting fractures during the review of radiographs of the musculoskeletal system. Rayvolve is indicated for the adult and pediatric population (≥ 2 years).
Rayvolve PTX/PE: is a radiological computer-assisted triage and notification software that analyzes chest x-ray images of patients 18 years of age or older for the presence of pre-specified suspected critical findings (pleural effusion and/or pneumothorax).
Rayvolve LN: is a computer-aided detection software device to assist radiologists to identify and mark regions in relation to suspected pulmonary nodules from 6 to 30mm size of patients of 18 years of age or older.
Caution: The data mentioned are sourced from internal documents, internal studies and literature reviews. It is for distribution to Health Care Professionals only and should not be relied upon by any other persons. Testimonial reflects the opinion of Health Care Professionals, not the opinion of AZmed. Carefully read the instructions for use before use. Please refer to our Privacy policy on our website. For more information, please contact contact@azmed.co.
AZmed 10 rue d’Uzès, 75002 Paris - www.azmed.co - RCS Laval B 841 673 601
© 2025 AZmed – All rights reserved. MM-25-56
