The Pathway Illusion in the Imaging Industry

The pattern shows up whenever an entire industry discovers the same aspiration at once. The imaging sector has converged on a shared vocabulary of care pathways, longitudinal patient journeys, and connected ecosystems. The vision is coherent and the direction is correct, but the structural foundation to deliver it does not exist within the incumbent OEM model. The gap is not a technology problem waiting for the right investment cycle. It is the accumulated consequence of how the industry was built, how it is organized, and what it has understood its job to be.

A companion analysis, The Imaging Gap Nobody Built, made the case that the incumbent imaging industry never built the patient-centered floor beneath advanced imaging, the longitudinal surveillance layer that would catch meaningful change before a patient ever reaches a formal diagnostic workup. This piece extends that argument upward. The same commercial choices that left the floor unbuilt now constrain the sector from orchestrating the pathway it has begun to claim. An industry that never made imaging follow the patient cannot credibly position itself as the coordinator of the patient journey.

This is not a critique of any single organization. It is a structural assessment of a business model that evolved to optimize for something other than what it is now promising to deliver. The path forward requires the sector to do something strategically difficult. It must honestly define what imaging is in a care pathway, accept the organizational and commercial implications of that definition, and build toward a role that is genuinely indispensable rather than aspirationally comprehensive.

What a Clinical Pathway Actually Is

A clinical pathway is not an imaging and diagnostics workflow with connections added. It is a multi-domain construct built from laboratory results, patient history, genomics, medications, procedure notes, vital signs, patient-reported outcomes, clinician assessments, payer authorization events, and longitudinal follow-up across months or years. Imaging is a high-value input at specific decision nodes. It is not the connective tissue that holds the pathway together. That role belongs to the EHR, which combines clinical care with the revenue cycle management that governs how care is authorized, documented, and reimbursed.

The same pattern repeats across clinical domains. Imaging and monitoring contribute decisive information at specific nodes, while the pathway itself is coordinated by data and decisions that live outside the imaging stack.

Acknowledging this does not diminish imaging’s clinical importance. It is a prerequisite for honest strategic positioning. Organizations that overstate their role in a pathway will build the wrong products, pursue the wrong partnerships, make the wrong organizational investments, and present cost where they intend to deliver value. The organizations that accurately define their contribution will build something genuinely irreplaceable.

The Infrastructure Gaps That Make Orchestration Impossible

Beyond the conceptual misframing, the imaging sector faces structural infrastructure gaps that make pathway orchestration impossible regardless of strategic intent. These gaps are not primarily technical. They are the downstream product of commercial and organizational decisions made over decades.

The most revealing test of readiness is internal. Across the sector, modality-specific workflows within a single vendor ecosystem frequently fail to exchange data coherently. CT, MRI, ultrasound, and nuclear medicine share a patient identifier but often little else at the data layer. DICOM, the dominant transport standard, was designed in 1993 for device interoperability and pixel-fidelity transmission, not for longitudinal patient-level analysis, cloud-native inference, or integration with systems organized around structured data. A patient moving across modalities and sites is not traveling through a coordinated ecosystem. Reports may follow interoperably, but the images they are based on usually do not, and patients are still asked to carry prior studies on a CD or DVD.

This raises a plain question. If the EHR ecosystem, the system of record for nearly everything else in care, has never adopted DICOM, then DICOM functions less as a true interoperability standard and more as a workaround that hardened into infrastructure. Netflix compresses and streams high-fidelity, full-length film to any device on demand. Healthcare should be able to do better than handing a patient a disc. The persistence of DICOM in its current form is a signal that image sharing needs a modern path, not another decade of accommodation.

The barriers to moving images across settings are well understood and rarely prioritized. Image files dwarf structured clinical data, exchange runs on FHIR and CCDA while imaging sits in DICOM PACS and VNA silos outside it, some organizations earn revenue from repeat imaging, and the metadata embedded in images raises a compliance burden beyond standard exchange. None of this is technically intractable. It has simply never served the installed base to solve it. Radiation dose is the clearest illustration. Cumulative CT exposure is a documented population-scale carcinogen, yet dose data does not reliably follow the patient across systems, or even within one. An industry that cannot make a patient’s own radiation history travel with them is not positioned to orchestrate that patient’s longitudinal pathway.

The cloud transition underway across the imaging sector is real but limited in its current form. Cloud-connected imaging, as broadly implemented, means image storage or AI inference workloads offloaded to hosted infrastructure while the core acquisition and workflow architecture remains premise-bound. What has largely been delivered is a more expensive remote archive with some AI inference layered on top.

True cloud-native architecture is not the relocation of an on-premises service to someone else’s data center. It changes the operating model. Multi-tenant design, multi-region and multi-time-zone availability, data governance, and the privacy and security posture that protects patient data across all of it become core architectural commitments rather than deployment details. It also changes the commercial model. A genuine SaaS offering requires different contracting terms, different service-level and uptime obligations, different data residency and breach-liability provisions, and a recurring-revenue structure the capital equipment sales motion was never built to support. That distinction is where much of the sector’s cloud positioning currently overstates what has actually been delivered.

The EHR dimension is the most important and least acknowledged. The electronic health record is the source of truth for the longitudinal record, the origin of the imaging order, and the system of record for reimbursement. The imaging pathway does not begin at the scanner. It begins at the referral order, which lives in the EHR. EHR platforms carry deep, native RCM competency because billing is foundational to their architecture rather than adjacent to it, including prior authorization, episode cost attribution, and reconciliation between what was ordered, performed, resulted, and reimbursed. As EHR vendors build deeper into radiology information system functionality, they are not entering imaging as outsiders. They are absorbing the workflow layer where imaging sits inside a clinical and financial episode, and the sector’s blind spot around this is one of the more consequential miscalculations in the current landscape.

The Organizational Fault Line

The infrastructure gaps described above are symptoms. The underlying condition is organizational. The imaging sector evolved through acquisition-driven growth and modality-specific business unit structures that were rational for capital equipment sales and remain deeply embedded in how the industry is organized today.

The result is segmentation at every layer. Product teams are organized by modality, with separate roadmaps, architectures, and in many cases separate data schemas under one vendor brand. Sales teams follow the same lines, so a health system with multiple modalities from a single vendor often manages separate contracts and account teams that share no quota, pipeline, or strategy. The experience of a multi-modality OEM relationship is frequently the experience of buying from several companies that happen to share a logo.

The interface and workflow fragmentation that results is not incidental. It is the surface expression of an organization never designed to produce a coherent platform. A technologist moving between modality workflows from the same vendor meets different interface paradigms, terminology, export behaviors, and integration patterns. The promise of a unified imaging partner erodes at every touchpoint when the seller operates as a product shop rather than a solution enterprise.

This fragmentation also undermines the most valuable commercial asset the large imaging OEMs actually possess, a trusted enterprise relationship with health systems at scale. The value of that relationship is proportional to the coherence of what it delivers. Feature competition at the device level, waged by teams that are structurally incentivized to protect modality-specific revenue, converts a potential platform advantage into a series of isolated product transactions. The enterprise relationship becomes a vendor relationship, and a vendor relationship is one competitive bid away from displacement, protected only by switching costs.

Device-by-device feature competition fragments the very brand equity that makes the enterprise relationship worth having. Platform coherence is the differentiator the sector keeps overlooking.

What Emerging AI Architecture Signals

Recent developments in AI research point toward an architectural shift that the imaging sector should read as both a signal and an opportunity, depending on how quickly it reframes its strategic role.

EchoJEPA, developed by researchers at the University of Toronto and the Vector Institute, applies Meta’s Joint Embedding Predictive Architecture to cardiac ultrasound. Conventional imaging AI learns to reconstruct masked pixels, absorbing the noise and artifacts of the signal along with the anatomy. JEPA instead predicts abstract semantic representations of the masked region, learning what is conceptually present rather than what the image looks like at the pixel level. The performance results are significant:

• 79 percent accuracy on view classification using one percent of labeled training data, compared to 42 percent for the best conventional alternative using the full labeled dataset.
• Under simulated image degradation, the JEPA model’s performance dropped by 2.3 percent while competing approaches degraded by up to 16.8 percent.
• Without any training on pediatric cardiac anatomy, the model outperformed baselines that had been explicitly fine-tuned for that population.

The architectural implication extends beyond benchmark performance. If high-performing cardiac imaging AI operates on compressed semantic representations rather than pixel-fidelity DICOM payloads, the data transport problem that has historically separated imaging systems from clinical integration changes fundamentally. The system does not need to move a multi-gigabyte DICOM file to a downstream clinical system. It moves a structured clinical inference, an ejection fraction estimate, a stenosis grade, a nodule characterization with a confidence interval. That output is not an image. It is a structured clinical finding, and structured clinical findings are the currency that EHR systems, clinical decision support engines, quality registries, and payer audit trails can actually consume.

This opportunity is available only to organizations that accept the platform participant framing rather than the pathway orchestrator framing. Imaging outputs delivered in structured form participate in the pathway by exposing data the rest of it can act on, which is the contribution the OEMs can honestly make.

The Platform Participant Model

The constructive alternative to pathway overreach is not a smaller ambition. It is a more precisely targeted one. The imaging sector can build toward something genuinely irreplaceable by accepting a specific strategic identity. Imaging is a platform participant that contributes structured, high-fidelity clinical data to care pathways coordinated at layers it does not and should not control.

This reorientation has concrete implications for product strategy, organizational design, and commercial model.

At the product level, the organizing principle shifts from device features to platform coherence. Data ingress and egress standards, defined explicitly and held consistently across modalities, become the primary differentiator rather than the modality-specific capabilities that drive roadmap investment today. The platform team, with client input, defines the output requirements of the device. Metadata schemas, output formats, and integration specifications are not afterthoughts appended to releases. They are the architecture that determines whether imaging participates in a clinical workflow or sits adjacent to it.

At the data output level, the shift is from unstructured radiology reports to coded, structured findings that decision support engines can consume without NLP intermediation. Population health, quality registries including the ACC and NCDR, CMS quality measurement, and value-based analytics all require structured imaging data at scale, and the sector produces almost none of it in a usable form. Closing that gap is a platform value proposition, not a hardware feature, and the few organizations that commercialize the modality and the platform together hold the advantage.

This is also where the market access gap described in The Imaging Gap Nobody Built reappears. Structured, payer-consumable output is the raw material of a health economics argument, and the OEM commercial model was never organized to build one. The same absence that has kept outpatient point-of-care ultrasound and standalone imaging AI underreimbursed will keep pathway claims rhetorical until the sector produces data that payers and value-based programs can act on.

At the organizational level, the modality-specific P&L and sales team structure that produces internal fragmentation needs to be examined against the platform coherence objective. Enterprise account strategy cannot be coherent when the teams executing it are structurally incentivized to compete for the same budget across modality lines. Platform-level value creation requires platform-level commercial accountability.

At the AI layer, the emerging architecture signals suggest that the most valuable imaging AI outputs will be structured inferences rather than annotated images. Investment in AI that produces pathway-actionable structured data, calibrated for integration with EHR and decision support systems, is investment in the platform participant role. Investment in AI that produces better images for radiologist review is investment in the existing model.

At the commercial level, the shift is from selling a device to enabling a program. Standing up an imaging capability is far more than placing hardware, and most of the work, from workforce and integration to reporting, governance, reimbursement, and program management, sits inside the health system and has little to do with the device. The interactive supplement below maps the full set of delivery requirements against their real owners and shows how thin the vendor footprint is. The players that understand this operational reality, and the internal alignment a health system has to achieve, will pivot from vendor to partner. That pivot is the practical form of the platform participant role, and it is where durable enterprise relationships are earned.

The imaging sector does not need to be smaller in ambition. It needs to be more precise about where its contribution creates irreplaceable value. Platform coherence, structured data output, and workflow participation are those targets.

The Imaging Value Journey and Delivery Matrix

Knowing What You Are

Strategic clarity about identity is not a concession. Knowing who you are, and who you are not, is a competitive advantage in markets where overreach has opened a credibility gap between what incumbents promise and what health systems experience. The sector has earned deep trust with health systems over decades. Pathway overreach puts that trust at risk, and platform coherence would compound it.

Health systems that invested in multi-modality relationships expected a coherent enterprise partner and often received a collection of capable devices from teams that share neither a platform nor an account strategy. The pathway narrative, continued without the structural changes to support it, widens that gap rather than closing it.

Frequently Asked Questions  ·  Erik’s Hot Take