Over the past five years, artificial intelligence (AI) has been making its way into medical imaging, accelerating workflows and enhancing diagnostic precision. Generative AI is now creating new opportunities by synthesizing clinical context from images, reports, and patient records. Prof. Dr. med. Mathias Goyen, GE HealthCare’s Global Chief Medical Officer for Imaging and Advanced Visualization Solutions, explains how AI solutions are addressing real clinical challenges.
What do you see as the most groundbreaking innovation in medical imaging in the last five years, and how is it transforming clinical practice today?
One of the most transformative innovations in medical imaging over the past five years has been the clinical integration of artificial intelligence (AI) at scale. While AI has long been in development, we’re now seeing its real-world application making a measurable impact, particularly in enhancing workflow efficiency, image reconstruction, and clinical decision support.
In radiology, for example, deep learning-based image reconstruction techniques like AIR™ Recon DL are delivering higher-quality images with significantly reduced scan times. This not only improves the patient experience but also increases scanner throughput, which is critical in today’s high-demand environments.
Another key development is the expansion of quantitative imaging biomarkers. These biomarkers are helping transition imaging from qualitative interpretation to data-rich, quantifiable assessments. This is particularly impactful in oncology, cardiology, and neurology, where disease progression or treatment response needs precise, reproducible measurements.
Artificial Intelligence is rapidly being integrated into diagnostic imaging. What are some of the most promising AI applications currently in use or development at GE HealthCare?
At GE HealthCare, we focus on developing AI solutions that solve real clinical problems, such as addressing the need to improve diagnostic confidence, workflow, and outcomes. One notable example is our Critical Care Suite, which uses AI to detect pneumothorax on chest X-rays at the point of care and prioritize these cases for radiologist review. This has already shown the potential to reduce time-to-treatment in critical cases.
We’re also advancing AI in image acquisition. Tools like Auto Positioning and Auto Protocol Selection reduce variability and optimize exam quality with less technologist input, especially beneficial in resource-constrained environments. Additionally, our AI-enabled post-processing tools are helping clinicians quantify and interpret subtle findings, for instance, enabling automated segmentation of cardiac structures in echocardiography or volumetric assessments in CT.
Are generative AI tools, which enable radiologists to discuss images with AI, something coming soon to radiology?
Absolutely. Generative AI is already demonstrating potential in natural language processing tasks, and the application of such tools to radiology is a natural progression. We are actively exploring generative models that can synthesize clinical context from images, reports, and patient records, enabling radiologists to engage in intuitive, dialogue-based interactions with AI.
Imagine a scenario where a radiologist can ask the system: “Show me similar cases with histopathologic confirmation,” or “What was the prior growth rate of this lesion?” This level of interaction could dramatically enhance efficiency and clinical insight. While still early, such tools will likely evolve into digital assistants embedded in PACS/RIS systems - context-aware, always learning, and continually improving.
You’ve noted that AI’s most immediate impact lies in workflow optimization and triage, particularly in identifying critical findings such as pneumothorax and intracranial haemorrhage. How can radiology departments best implement these tools?
Successful implementation begins with clear clinical goals and robust change management. First, departments should identify bottlenecks or areas where delays have a measurable impact, such as emergency or stroke pathways. AI tools targeting these areas often yield the highest impact on care.
Second, integration into existing workflows is critical. AI is most helpful when embedded into the radiologist’s native environment (PACS, RIS, or EMR) so it supports rather than disrupts. For example, flagging critical findings should not require separate logins or software; it should happen automatically and with minimal user interaction.
Finally, continuous monitoring and feedback are essential. Institutions should track performance metrics – such as sensitivity, specificity, and time savings – and recalibrate as needed. Building trust among users through transparency and demonstrated value is key to sustaining adoption.
You claim that seamless integration of AI into PACS/RIS reduces friction. What specific design principles or examples at GE HealthCare illustrate this “AI-invisible” approach?
Our design philosophy centers on making AI “invisible,” working silently in the background while enhancing the clinical workflow. A great example is our integration of the Critical Care Suite into X-ray workflows. The AI runs on-device and sends priority alerts directly to PACS with no additional user input required. There’s no need for the radiologist to click on a separate AI tab or portal; the results are presented contextually and clearly within their reading workflow.
We also apply this principle in our Smart Subscription model, which allows providers to receive continuous software updates without interrupting clinical operations. And our AI orchestration layer helps to ensure that the right algorithms run on the right studies at the right time, based on clinical context, modality, and anatomy – seamlessly behind the scenes.
With AI supporting image interpretation, how do you see the role of radiologists evolving?
Far from replacing radiologists, AI will elevate their role, decreasing repetitive tasks and enabling them to focus more of their time on high-level synthesis and patient engagement. Radiologists will increasingly serve as information integrators, combining imaging, clinical history, pathology, and genomics into unified diagnostic narratives.
Moreover, radiologists will play a central role in guiding AI development – labelling datasets, validating models, and ensuring ethical deployment. As multidisciplinary team members, they will also take on leadership in precision medicine initiatives and population health strategies.
You’ve predicted the emergence of “diagnostic medicine,” integrating radiology, pathology, and molecular diagnostics. How should current radiologists prepare for this shift?
Radiologists must expand their skill sets beyond imaging. Understanding molecular biology, pathology, and bioinformatics will be essential to remain central in the diagnostic chain. Cross-disciplinary training and collaboration will be key.
At GE HealthCare, we are building tools that unify diverse data types from imaging to genomics and present them in clinically meaningful ways. As these tools mature, radiologists who embrace lifelong learning and multidisciplinary integration will lead this transformation.
Radiology departments can also foster this evolution by establishing joint diagnostic boards and integrating data scientists into their teams.
As the newly appointed Global Chief Medical Officer for Imaging and Advanced Visualization Solutions (Ultrasound + Image Guided Therapies), what are your top priorities for the coming year?
My focus is threefold. First, I want to strengthen our clinical partnerships to ensure our solutions address real-world challenges and deliver measurable outcomes.
Second, we aim to accelerate our innovation pipeline, particularly around AI, molecular imaging, and precision diagnostics, while ensuring these tools are accessible globally, not just in high-resource settings.
Third, I’m committed to advancing health equity. This means working to reduce disparities in imaging access, quality, and outcomes, especially in underserved communities. We’re exploring ways to leverage mobile imaging, remote diagnostics, and AI to bridge these gaps.
You’ve observed healthcare systems around the world. How would you compare the pace and impact of digital transformation in different regions?
Digital transformation is advancing unevenly. In high-income countries, adoption is often driven by workforce shortages and efficiency demands. These systems tend to integrate AI and digital workflows rapidly, but they face regulatory and interoperability hurdles.
In contrast, many low- and middle-income countries are leapfrogging legacy systems, adopting cloud-native solutions and mobile diagnostics from the outset. This creates opportunities for innovation tailored to local needs.
That said, a common challenge everywhere is data silos. True transformation requires unified data platforms and cross-sector collaboration. At GE HealthCare, we are working globally to address these barriers and support scalable, region-appropriate solutions.
What are the biggest challenges healthcare providers face when adopting advanced imaging technologies and digital workflows, and how can they be overcome?
The biggest challenges are infrastructure limitations, user adoption, and change management. Advanced imaging tools often require robust IT systems, integration capabilities, and reliable connectivity, all of which can be unevenly available.
Equally important is the human factor. Clinicians need training, trust, and time to adapt to new technologies. That’s why we focus on validating and fine-tuning solutions with clinical partners and providing ongoing education and support.
Reimbursement and regulatory clarity also remain critical hurdles. Policymakers must keep pace with innovation to ensure sustainable adoption. We actively engage with regulators and industry partners to help shape this environment responsibly.
As AI takes on more diagnostic responsibilities, how should the industry address concerns around accuracy, accountability, and patient trust?
Transparency and validation are paramount AI tools we develop undergo rigorous testing and regulatory scrutiny. But that’s just the beginning. Post-market monitoring and real-world performance assessments are equally essential.
Accountability must remain human-centered. AI should augment, not replace, clinical judgment. Radiologists must be the final arbiters of diagnosis, supported by AI but not supplanted by it.
Building trust also requires educating patients and clinicians about how these tools work, their limitations, and their safeguards. Ethical AI principles, such as explainability and fairness in AI decision-making, are not optional; they are foundational.
What excites you most about the future of medical imaging, and what role will GE HealthCare play in shaping that future?
I’m most excited by the convergence of data, diagnostics, and personalization. We’re moving from images to insights and from static pictures to dynamic, data-driven stories about human health.
GE HealthCare is uniquely positioned to lead this evolution. With our deep domain expertise, global reach, and commitment to innovation with purpose, we’re building the tools and platforms that will define diagnostic medicine for the next generation.
Whether it’s AI that helps clinicians detect disease earlier, imaging biomarkers that help to predict therapy response, or platforms that connect radiology, pathology, and genomics information and data, our goal is clear: empower clinicians, advance care, and be a part of improving outcomes for every patient, everywhere.
