The session AI Integration in a Nutshell at the European Congress of Radiology explored key challenges and opportunities in AI adoption for medical imaging. Three experts—Willem Grootjans, Sergey Morozov and Kicky Gerhilde Van Leeuwen—shared insights on user expectations, workflow integration and AI marketplaces, highlighting the critical role of seamless implementation in maximising AI’s clinical value.
User expectations
Willem Grootjans focused on the practical aspects of integrating AI into clinical workflows, emphasising seamless integration as key to realising AI’s full potential. He highlighted the narrow scope of most AI tools, their lack of interactivity and the absence of standardisation across vendors, which creates integration challenges.
He illustrated how AI can assist in tasks such as scoliosis measurements, fracture detection and chest X-ray evaluations, stressing the importance of "proper configuration of your front-end and back-end systems." He warned that AI implementation should avoid disruptions and ensure users can modify results when necessary.
To assess AI’s impact, Grootjans discussed digital tracer labs that track eye movements and clicks to quantify efficiency gains. He concluded that “seamless integration is key. The integration will determine the added value of AI and how users will use it.”
IHE and AI workflow integration
Sergey Morozov presented the challenges of AI integration, focusing on interoperability and the role of the IHE (Integrating the Healthcare Enterprise) initiative. He criticised the current state of AI integration, likening it to a “zoo” of uncoordinated solutions and multiple pop-up interfaces that disrupt workflow efficiency.
Morozov stressed that many AI failures are not due to the technology itself but “because of integration, so inappropriate or failed technical integration eventually would completely disapprove AI.” He highlighted the importance of standardised solutions, emphasising that AI should be viewed as part of a broader software ecosystem, requiring proper implementation, regulatory compliance and seamless workflow integration.
He outlined IHE’s AI workflow profiles, which aim to standardise execution, result communication and validation. He closed by stating that “IHE provides essential tools for AI integration, standardised plug-and-play framework, new profiles… ensuring transparency.”
Role of AI-Marketplaces
Kicky Gerhilde Van Leeuwen explored the role of AI marketplaces in simplifying AI adoption for hospitals. She described how AI platforms streamline integration by offering a centralised hub for AI applications, handling aspects like contract management, billing and validation.
She pointed out the complexity of current AI adoption, stating that “hospitals are doing that one by one for each individual AI installation,” which is inefficient. She discussed how platforms address challenges like data orchestration, stating that “currently when we apply AI models, we see that it’s usually based on DICOM metadata… but also on the pixel data.”
She presented examples from Norway and the Netherlands, where national AI infrastructures are being tested to enhance scalability. She concluded that “the platform market is still a very dynamic market. Players are coming and going, they’re collaborating,” making it essential for hospitals to carefully evaluate long-term viability.
Effective AI integration in radiology requires more than just advanced algorithms; it demands robust interoperability, standardisation and scalable deployment solutions. From optimising workflow efficiency to enhancing AI accessibility through marketplaces, the speakers underscored the need for structured implementation strategies. As AI adoption grows, ensuring smooth integration will be key to unlocking its full potential in clinical practice.
Source & Image Credit: ECR 2025
