Inova Health System

As artificial intelligence continues to expand across healthcare, imaging has emerged as one of its most active frontiers. From stroke detection to workflow optimization, the number of available tools continues to grow. Yet many health systems still lack a consistent way to evaluate whether these technologies truly deliver value.

The question is no longer whether to invest in AI; it is how to assess its impact in a meaningful and operationally relevant way.

Today, many AI decisions are influenced by vendor demonstrations, isolated use cases, or early performance metrics that do not always translate into real-world improvement. While these inputs are helpful, they often fall short of answering the most important question: does this tool make our system better, for patients, clinicians, and operations?

To address this gap, imaging leaders should adopt a more structured approach, one that evaluates AI not just as a technology, but as an operational asset. In practice, this means assessing impact across five key domains: radiologists, technologists, operations, governance, and financial value.

To operationalize this approach, the Imaging AI Value Framework (IAVF) was developed, a structured model that evaluates AI across these domains. While simple in concept, its strength lies in consistently applying these principles to real-world decision-making.

From a radiologist perspective, AI should enhance clinical performance, not complicate it. The most valuable tools improve detection of easily missed findings, prioritize critical cases, and support faster, more consistent interpretation. Just as important, they reduce cognitive burden. If a tool introduces more noise than signal, its value becomes questionable regardless of technical capability.

In practice, some AI tools already demonstrate clear value. Stroke imaging algorithms, for example, improve detection of critical findings, prioritize urgent cases on the worklist, and accelerate interpretation through automated analysis. More importantly, they align across the full spectrum of radiologist needs, supporting accuracy, efficiency, and workflow prioritization simultaneously. This type of alignment is what defines high-value AI.

For technologists, AI presents an opportunity that is often under-recognized. Tools that improve image quality, guide positioning, assist with protocol selection, and prevent common errors, such as incorrect exam selection or patient mismatches, can significantly strengthen the entire imaging process. When technologists are supported effectively, downstream quality and efficiency improve.

It is also important to distinguish between AI and other advanced imaging technologies. Dual-energy CT, for example, can help differentiate acute from chronic fractures through material-based reconstruction, but its value is rooted in physics rather than algorithmic decision support. Making this distinction is critical when evaluating impact and setting expectations.

“The question is no longer whether to invest in AI; it is how to assess its impact in a meaningful and operationally relevant way.”

Operational impact is where many AI solutions either prove their value or fall short. Improvements in turnaround time, throughput, emergency department flow, and patient access are tangible indicators of success. If an AI tool does not meaningfully improve these metrics, its role should be carefully reconsidered. Technology alone is not the goal, measurable operational improvement is.

Governance and risk are equally important. As AI becomes more embedded in clinical workflows, health systems must ensure appropriate oversight, regulatory alignment, and consistency in deployment. This includes managing bias, false positives, workflow integration, and maintaining clear separation between clinical decision-making and vendor influence.

Financial value must also be grounded in reality. AI should demonstrate clear return on investment, cost avoidance, or measurable efficiency gains. Without this, even technically strong tools may not be sustainable at scale.

Together, these domains form the foundation of a practical evaluation model, one that aligns AI adoption with real-world performance.

One of the challenges facing organizations today is that AI adoption is moving faster than the development of evaluation frameworks. This creates variability, inconsistent expectations, and potential gaps in patient care and operational performance.

A more disciplined approach can address this. By applying consistent criteria across AI tools, health systems can better prioritize investments, scale what works, and avoid unnecessary complexity.

At its core, the goal is simple: AI should make imaging better, better for patients through improved quality and access, better for clinicians through enhanced support and efficiency, and better for the system through measurable operational gains.

As imaging leaders, we have a responsibility to ensure that the technologies we adopt align with these outcomes. Doing so requires moving beyond enthusiasm for innovation and toward a structured, outcomes-driven approach.

As a complement to this perspective, the Imaging AI Value Framework (IAVF) outlines a practical structure for applying these principles in real-world settings.

Imaging AI Value Framework (IAVF) Radiologist Impact

• Improves detection of critical or easily missed findings

• Supports prioritization of urgent cases

• Enhances efficiency and consistency in interpretation

• Reduces cognitive burden

Technologist Impact

• Improves image quality and positioning

• Supports protocol selection

• Reduces errors (e.g., incorrect exam or patient mismatch)

• Improves scan efficiency

Operational Impact

• Improves turnaround time

• Increases throughput

• Supports emergency department flow

• Improves patient access

Governance & Risk

• Ensures regulatory and clinical alignment

• Maintains reliability and workflow integration

• Addresses bias and transparency

• Supports responsible vendor engagement

Financial Value

• Demonstrates measurable return on investment

• Supports cost efficiency

• Enables sustainable scaling

Core Principle

AI should deliver value across the system, not optimize one area while creating unintended burden in another.

The IAVF has been operationalized into a practical model to support AI evaluation in imaging. Continued collaboration will be important as organizations refine how these tools are assessed and implemented in real-world settings.