From Promise to Practice

THE PARADOX OF PROGRESS

More than 1000 AI tools approved. Fewer than 2% used. What went wrong?

At radiology conferences, the future looks bright. Booths are highlighting promises of fully automated diagnostics, intelligent workflows, and seamless clinical decision support. More than 1000 FDA-cleared AI tools - most for radiology - are showcased as if the future has already begun.

However the truth is, that little has changed. Fewer than 2% of U.S. radiology practices use these tools. Most radiologists have never tried one—and even fewer trust them. This isn’t due to a lack of innovation, funding, or regulatory approval. The real barrier is more fundamental—and more deeply embedded: the everyday structure of how radiology works.

Radiology’s data infrastructure is fundamentally misaligned with the logic behind scalable AI-solutions. Free-text reports, siloed systems, and absent standards create structural barriers that keep even the best algorithms from reaching clinical impact.

This article argues AI isn’t failing - our clinical environments are. If we want intelligence to scale, we need to start thinking structurally.

“DIAGNOSIS”: WHY AI ISN’T SCALING IN RADIOLOGY

If FDA clearance is the finish line, radiology AI would already be a success story. But the real race begins after regulatory approval - and most tools never make it out of the starting blocks.

The gap is striking: over 700 AI tools have been cleared in the U.S., yet clinical use remains rare. Only two—targeting coronary artery disease and diabetic retinopathy—have exceeded 10,000 uses nationwide. In most hospitals, even with advanced imaging systems, these tools remain shelfware: approved, available, and unused.

Why?

The main barriers to AI adoption aren’t technological—they’re infrastructural and cultural. AI tools often produce structured, high-quality data, yet must operate in workflows still dominated by free-text reporting. This mismatch creates friction: structured outputs require manual reconciliation, turning automation into additional work. In daily clinical practice, AI faces fragmented reporting environments, siloed systems, and incompatible formats—making seamless integration nearly impossible. These structural issues also limit AI’s long-term potential: without consistent, longitudinal data, models can’t improve over time1. Legacy IT infrastructures further compound the problem, blocking scalable integration.

Moreover, trust remains a central issue. Radiologists are rightly skeptical of tools that cannot explain their decisions or that function as black boxes. This is exacerbated when AI tools are injected into workflows without alignment to the way radiologists actually read, report, and communicate findings1.

The issue isn’t AI’s ability to read images - it’s its inability to read our clinical reality. Without reengineering the environment, scale will remain impossible.

THE ANATOMY OF AN INVISIBLE BARRIER

If we look past the surface-level explanations - cost, time, resistance - we find a deeper, more systemic issue: radiology was never built for AI.

Hospital data ecosystems remain deeply fragmented. PACS, RIS, and EMR-systems often operate in silos, limited by proprietary structures and weak adherence to standards like HL7 or FHIR. Even within a single hospital, departments may use separate vendors for imaging, speech recognition, and reporting - each with its own constraints. Vendor lock-ins and missing APIs turn even basic data handoffs into friction points. Trying to insert an AI tool into such an environment is like fitting a jet engine onto a bicycle. The infrastructure can’t carry it.

The problem becomes especially acute at the point of reporting. Most radiology reports today are written in free text, using inconsistent phrasing and individual reporting styles. For a human reader, this variability is manageable – even though it is inefficient. For an AI-system, it’s chaos. Extracting structured meaning from prose language requires error-prone NLP techniques that often introduce more uncertainty than insight.

Let´s imagine a different setup - one where radiology reports are created using structured templates: clearly labeled, standardized across vendors and sites, and machine-readable. Structured reporting forms the landing zone that enables AI to integrate directly into the diagnostic workflow - actively by inserting findings into predefined fields, and passively by learning from the high-quality data this structure provides.

It’s a shift in philosophy. Instead of building smarter AI to overcome noisy environments, we build smarter environments that allow even modest AI to function with precision. The barrier to AI isn’t code - it’s unstructured habits and disconnected systems.

WHAT STRUCTURED REPORTING CAN ACTUALLY ACHIEVE

Skeptics often ask: if structured reporting is so effective, why isn’t it the norm?

The answer is that - quietly, and in select institutions - it already is. At NYU Langone Health, a structured template was introduced for adnexal mass MRI reporting, known as O-RADS MRI. The result? Referring gynecologic oncologists rated the new reports as significantly clearer and more actionable. Patients, too, reported fewer misunderstandings. Structured reporting didn’t dilute the nuance of radiology - it amplified its communicative value.

At Cincinnati Children’s Hospital, the transition to structured reporting was completed in under two years. A cross-functional team - radiologists, IT, and operations - jointly selected templates and set integration milestones. Initial resistance faded as radiologists joined the design process. Reporting became more consistent, and referring physicians noted clearer communication and fewer follow-ups. Crucially, efficiency improved not by speeding up readings, but by reducing disruptions and delays in downstream care.

Structured reports benefit both human readers and AI training pipelines. Templates encode findings in a structured, machine-readable format. This provides clean, reliable data for training future models, reducing the reliance on costly and error-prone manual annotation. In short, structured reports enable a virtuous cycle: better data leads to better AI, which in turn improves diagnostic workflows.

THE CULTURAL RESISTANCE – AND HOW TO OVERCOME IT

A common concern is that templates reduce radiologists to rigid checklists. Many clinicians worry that nuance and narrative interpretation - essential tools of diagnostic storytelling - will be lost in a sea of drop-downs and pre-filled phrases. These concerns are valid. But they are also based on a false binary.

Structured reporting doesn’t have to mean sterile reporting. In fact, modern systems allow for flexible, hybrid designs: standardized data fields for key findings, paired with open commentary sections when needed. The goal isn’t to reduce expression - it’s to reduce ambiguity.

Resistance is also fueled by distrust in AI itself. Radiologists are trained to understand pathology, not probability distributions. Introducing a black-box tool into their diagnostic process without clear explanations or control mechanisms feels like a threat to their expertise. This fear is amplified when AI tools are introduced without context, consultation, or workflow alignment.

Yet scepticism often fades when clinicians are involved early. At Cincinnati Children’s Hospital, radiologists were initially reluctant to adopt structured templates. Yet once they helped shape the system, participation turned into advocacy. Adoption followed co-creation.

The lesson is clear: resistance to structure is not a fixed trait. It is often a reaction to being excluded from the process. Invite radiologists to the table, and what once looked like a constraint begins to feel like a tool.

THE STRATEGY PROBLEM: WHY HOSPITALS ARE SOLVING THE WRONG PROBLEM

Faced with the promise of AI, many hospitals pursue the most visible goal: the next algorithm. They pilot stroke detection tools, lung nodule classifiers, or triage systems. Some even launch innovation centers and AI-sandboxes. But while these efforts are well-intentioned, they often skip the first and most essential step: making the clinical environment ready for AI.

Without standardized data formats, connected systems, and workflows that support integration even the best algorithm becomes a high-maintenance accessory. It won’t integrate. It won’t scale. It will remain a siloed solution in a fragmented system.

The problem is not the algorithm - it’s the architecture.

Leading radiology societies have begun to recognize this. A 2024 multi-society statement from the ACR, RSNA, ESR, and others outlines a framework for responsibly implementing AI in radiology. It doesn’t start with choosing a vendor. It starts with governance, data readiness, interoperability standards, and clinician involvement. Similarly, the ESR recommends incentivizing structured reporting as a foundational layer for scalable AI deployment - because without clean data, even the most powerful tool will fail.

Scaling AI requires aligning systems, teams, and data into an ecosystem AI can actually use—built on clean data structures, technical interoperability, and workflows suited for real-world clinical practice. But structural readiness isn’t the only barrier. Financial and organizational dynamics play a role, too: reimbursement rarely rewards downstream benefits like fewer clarification calls or faster care escalation, and procurement often prioritizes short-term ROI over foundational investments like structured reporting or interoperability. As a result, hospitals frequently implement the next algorithm without fixing the data layer—trapping innovation in a cycle of underperformance and unrealized potential.

Hospitals that invest in infrastructure before intelligence won’t just scale faster. They’ll scale smarter.

WHAT HOSPITALS CAN DO NOW

Scaling AI in radiology doesn’t begin with procurement - it begins with preparation. Hospital leaders can take tangible steps today:

• Conduct a data infrastructure audit to assess the interoperability and structure of reporting systems.
• Engage radiologists early in structured reporting initiatives to reduce cultural resistance and foster ownership.
• Establish an AI governance board involving IT, clinical, and operations stakeholders to ensure alignment across departments.
• Prioritize vendor-neutral structured reporting systems to maintain flexibility and future-proof integration efforts.
• Pilot structured reporting in high-volume modalities (e.g., Head-CT, Knee-MRI) not only to prepare clean data pipelines - but to gain real-world experience, identify workflow obstacles early, and develop local strategies for long-term success.

These actions shift the focus from tool selection to ecosystem readiness - and lay the basis for responsible, scalable AI deployment.

WHAT HAPPENS IF WE IGNORE THE STRUCTURAL PROBLEM?

Radiology AI doesn’t lack innovation—it lacks alignment. Hospitals face a flood of vendor pitches and demo tools, but smart algorithms can’t thrive in disjointed environments. Every failed pilot reinforces skepticism, feeding the belief that “AI doesn’t work.” But it does. What fails is how we deploy it. Without standardized data, interoperable systems, and aligned workflows, AI will remain stuck in pilot mode. High-tech centers may advance, but most will fall behind—scaling innovation in isolated pockets rather than system-wide.

There’s still time to change course, but it requires a shift in mindset: from chasing algorithms to building clinical infrastructure. Structured reporting isn’t a technical detail—it’s the foundation for scalable, trustworthy AI. Leaders must decide: implement AI as a superficial add-on to a broken system, or invest in the infrastructure it truly requires. The difference will determine whether AI becomes a cost burden or a catalyst for quality, safety, and efficiency.

Diagnostic intelligence isn’t held back by the lack of smarter algorithms—but by the absence of smarter environments. And those begin with structure.

REFERENCES

1. Obuchowicz R, Lasek J, Wodziński M, Piórkowski A, Strzelecki M, Nurzynska K. Artificial Intelligence-Empowered Radiology—Current Status and Critical Review. Diagnostics. 2025;15(3):282. doi:10.3390/diagnostics15030282
2. Wu K, Wu E, Theodorou B, et al. Characterizing the Clinical Adoption of Medical AI Devices through U.S. Insurance Claims. NEJM AI. 2024;1(1). doi:10.1056/AIoa2300030
3. FDA has now cleared more than 1,000 AI models, including many in cardiology. January 10, 2025. Accessed May 24, 2025. https://cardiovascularbusiness.com/topics/artificial-intelligence/fda-has-cleared-more-1000-ai-algorithms-many-cardiology
4. European Society of Radiology (ESR), Dos Santos DP, Kotter E, Mildenberger P, Martí-Bonmatí L. ESR paper on structured reporting in radiology—update 2023. Insights Imaging. 2023;14(1):199. doi:10.1186/s13244-023-01560-0
5. Casey A, Davidson E, Poon M, et al. A systematic review of natural language processing applied to radiology reports. BMC Med Inform Decis Mak. 2021;21(1):179. doi:10.1186/s12911-021-01533-7
6. Unlocking the Power of AI: The Vital Role of Structured Reporting. June 5, 2024. Accessed May 19, 2025. https://mint-medical.com/news/unlocking-the-power-of-ai-the-vital-role-of-structured-reporting
7. Woo S, Andrieu PC, Abu-Rustum NR, et al. Bridging Communication Gaps Between Radiologists, Referring Physicians, and Patients Through Standardized Structured Cancer Imaging Reporting: The Experience with Female Pelvic MRI Assessment Using O-RADS and a Simulated Cohort Patient Group. Acad Radiol. 2024;31(4):1388-1397. doi:10.1016/j.acra.2023.08.005
8. Quality RY| |. The Future of Radiology Reports: How Structured Reporting Is Rewriting the Rules. April 14, 2017. Accessed May 20, 2025. https://radiologybusiness.com/topics/care-delivery/healthcare-quality/future-radiology-reports-how-structured-reporting-rewriting
9. Brady AP, Allen B, Chong J, et al. Developing, purchasing, implementing and monitoring AI tools in radiology: practical considerations. A multi-society statement from the ACR, CAR, ESR, RANZCR & RSNA. Insights Imaging. 2024;15(1):16. doi:10.1186/s13244-023-01541-3

--Issue 06--