The Next Chapter of Radiology Reporting: Untethered, Unified, and AI-Driven
Key takeaways from Sirona Medical’s webinar on the future of radiology reporting
Radiology is at an inflection point. Volumes keep climbing, the radiologist shortage is deepening, and most reading rooms are still held together by a patchwork of legacy PACS, reporting tools, worklists, and EMR tabs that physicians have to stitch together in their heads—case after case, click after click.
In a recent webinar, Sirona Medical’s Chief Growth Officer John Danahy was joined by Senior Product Manager Dr. Trafton Drew, radiologist Dr. Peter Sachs, and VP of Marketing, Andrew Colbert, to unpack what comes next. The conversation centered on one premise: legacy infrastructure is the core problem in radiology today, and the next generation of reporting will look dramatically different once that infrastructure is rebuilt from the ground up.
Here are the biggest takeaways from the discussion.
Legacy infrastructure is quietly taxing every reading room
The panel opened with a sobering look at what physicians actually experience at the start of every shift. Based on Sirona’s observations, radiologists spend an average of 6 to 8 minutes just logging into their systems—and in at least one customer environment, that number stretches to 15 minutes before a single case is opened.
That login time is a symptom of something bigger: fragile integrations between PACS, RIS, reporting, voice recognition, worklist, and EMR that have to re-synchronize every time a radiologist starts work. Danahy pointed out that roughly 70% of radiologists read on two or more systems, and large IDNs can be far worse. One health system the team referenced operates more than 40 separate IT stacks.
Every toggle, every re-login, every workaround is a tax on throughput, quality, and morale—and it compounds the burnout, teleradiology dependence, and retention challenges the industry is already struggling with.
The complex case illustrates the problem
To make this concrete, the panel walked through "Maria"—an older patient with critical limb ischemia, multiple prior interventions, and a CTA plus lower-extremity duplex ultrasound on today’s worklist. To read her case well, a radiologist has to:
- Search PACS for relevant priors
- Open each prior, confirm protocol, and compare
- Analyze the tech sheet
- Transcribe technical details into the report
- Synthesize findings across modalities and time
- Draft an impression and finalize
In a legacy environment, that can mean eight or more toggles between systems before the report is finalized. A Northwestern study cited during the webinar suggested that thoughtful generative-AI integration into this kind of workflow could recover roughly 40% of a radiologist’s time—a number even skeptics in the audience agree is directionally transformative, whether the real figure lands at 15%, 20%, or higher.
"Radiology untethered": what a unified, cloud-native platform unlocks
Sirona’s answer is what the team calls the RadOS—a radiology operating system that unifies the worklist, diagnostic viewer, and reporting into a single cloud-native application that runs entirely in a Chrome browser. No VPN, no local software, active in under 15 seconds.
The philosophy Danahy repeated throughout: if Netflix and Amazon can stream 4K movies, there is no reason a radiologist should wait on images—even for tomosynthesis studies streamed out of AWS.
Three architectural choices make the rest of the vision possible:
1. A unified data model. Because Sirona owns the worklist, viewer, and reporter, it has access to DICOM, HL7, priors, and EMR context in one place. That eliminates the fragile hand-offs between point solutions and gives AI something meaningful to reason over.
2. SironaLex, an ontology layer. This is how the system actually understands the relationships in the data—what a study description really means, how reason-for-study maps to hanging protocols, how priors relate to the current exam. It’s the connective tissue behind features like reliable automated hanging protocols, which has long been one of radiology’s most stubborn problems.
3. FDA-cleared pixel-powered reporting. Sirona is the only player in its category to carry FDA clearance into the reporting layer itself. Because the pixels and the report share one system, measurements, labels, and findings can be bidirectionally linked—click a vertebral label in the report and the viewer navigates to the correct level, across current and prior exams.
The workflow features that matter most to radiologists
Dr. Drew and Dr. Sachs walked through the specific capabilities that actually change the minute-to-minute experience of reading a case:
Prior summaries. Instead of scrolling through every prior report, the system extracts relevant clinical findings from all available priors and surfaces them at the top of the patient jacket.
Automatic comparison insertion. The system knows which priors the radiologist actually hung and examined and automatically inserts the comparison date and description into the report—eliminating a tiny but relentless source of dictation friction.
Anatomic segmentation linked to the report. On an L-spine, for example, vertebral levels are labeled in the images and wired directly into the reporter, with cross-navigation to priors.
Focus mode. Rather than forcing radiologists to toggle between viewer and reporter, Focus Mode lets them dictate freely while watching the images. AI places each finding into the correct anatomical section of the template based on what was said and the patient’s context.
AI-generated impressions. Fine-tuned by individual, clinic, or subspecialty, generated on demand via voice command or a single action.
Agentic AI. This is where the conversation got most forward-looking. The team demonstrated a scenario where a single spoken instruction—"if there’s history of CLI in the tibials, include the extent in the report and hang the associated images"—updates the report, hangs the relevant prior, and flags the finding automatically. Think of it as a resident that never sleeps, running conditional logic across the entire case.
Where radiologists will feel this first
Dr. Sachs, a thoracic radiologist by training, offered two scenarios he’s personally most excited about:
Lung cancer screening. Priors hang correctly regardless of origin. Smoking history and risk factors are surfaced automatically. Pixel AI measures nodules, inserts measurements with image links into the report, and compares to priors. Focus mode lets the radiologist free-dictate incidental findings. Lung-RADS scoring and follow-up recommendations are generated—with agentic AI eventually scheduling referrals and follow-up imaging.
Data-heavy exams: DEXA, OB ultrasound, MFM. These are the studies that torture radiologists not because the interpretation is hard, but because the data entry is. Scanned PDFs, tech worksheets, dozens of measurements—all of it is ripe for agentic automation. As Dr. Drew noted, dictating dozens of numbers isn’t just slow; it’s a major source of error. Agentic import cuts both time and mistakes.
Tumor boards. Getting outside images normalized, hung properly, and stitched together with pathology, radiomics, genomics, and prior treatment means radiologists can walk in ready to contribute clinical expertise—rather than spending the day before the conference wrangling CDs.
Asked to estimate the efficiency gain, Dr. Sachs was deliberate: even a 10 to 20% time savings per case, multiplied across today’s volumes and today’s staffing shortages, is enormous.
The business case: efficiency, profitability, and cost disruption
The panel closed on a point that often gets lost in discussions of AI: this isn’t just a clinical story. Because Sirona is multi-tenant and cloud-native, the underlying cost structure is fundamentally different from legacy on-prem PACS—what Danahy described as "disruptively low cost" at a fraction of traditional PACS and bolt-on reporting solutions.
Combined with the efficiency gains, that creates room for practices to scale profitably, supplement existing systems during migrations, or support new business models—teleradiology, hospital PACS provided by rad groups, specialist groups splintering off to serve new markets.
The big idea
The future of radiology reporting isn’t another bolt-on to a legacy PACS. It’s a unified, cloud-native, AI-infused platform where the pixels, the report, the priors, and the EMR all live under one roof—so the radiologist can stop being a data integrator and go back to being a physician.
"Unloading the radiologist of the tedious components of managing data means we can actually bring our expertise to the clinical team—rather than being grunts managing data elements on our own." — Dr. Peter Sachs
Watch the full on-demand webinar
Frequently Asked Questions
What is the future of radiology reporting? The future of radiology reporting is a unified, cloud-native platform that combines the worklist, diagnostic viewer, and reporter into a single application, with AI integrated throughout the workflow. Instead of toggling between PACS, RIS, reporting, and EMR, radiologists work from one system where priors, HL7 data, pixel AI, and generative AI are orchestrated to reduce clicks, surface patient context automatically, and let physicians focus on interpretation.
Why is legacy PACS infrastructure a problem for radiology today? Legacy PACS environments stitch together many separate systems through fragile integrations. Radiologists commonly spend 6 to 8 minutes—sometimes 15 or more—just logging in each day, and roughly 70% read on two or more systems. Every toggle, synchronization, and workaround slows down interpretation, increases error risk, and contributes to burnout.
How does AI improve radiology reporting workflow? AI improves the reporting workflow in several ways: automating hanging protocols using ontology-driven understanding of study context, summarizing prior reports, inserting comparison dates and measurements automatically, labeling anatomy, routing free-form dictation into the correct template sections (focus mode), generating impressions, and running agentic conditional logic to pull in relevant priors or data on demand. Independent research cited in the webinar suggests generative-AI integration could recover up to 40% of a radiologist’s time.
What is agentic AI in radiology? Agentic AI refers to AI systems that can take multi-step actions on behalf of the radiologist based on natural-language instructions or conditional logic. In a radiology reporting context, agentic AI can, for example, detect a finding, pull the associated prior exam, hang the correct images, update the comparison field, and insert relevant text into the report—without requiring the radiologist to perform each step manually.
What is a cloud-native radiology platform? A cloud-native radiology platform is one that was built for the cloud from the ground up rather than a traditional on-premise system lifted into a hosted environment. It typically runs in a browser with no local software or VPN, streams images similar to how streaming services deliver video, scales elastically, and unifies data across the worklist, viewer, and reporter. This architecture enables faster login, easier integration of AI, and significantly lower total cost of ownership.
How much time can a radiologist save with a unified AI-driven reporting system? Estimates vary, but the panel cited industry research suggesting gains as high as 40%, while Dr. Peter Sachs offered a more conservative real-world estimate of 10 to 20% time savings per case from eliminating toggling, automating measurement insertion, and surfacing patient context automatically. At current imaging volumes and staffing levels, even the lower end of that range is highly material.
What is Focus Mode in radiology reporting? Focus Mode is a dictation experience that lets the radiologist keep their eyes on the images and speak freely without manually navigating to specific sections of the report template. AI interprets the dictation and places each finding into the correct anatomical section based on the patient’s context and the template structure.
Does Sirona Medical integrate with third-party AI tools? Yes. Because Sirona has unified access to the pixels, the report, and the patient data, third-party AI vendors—such as those focused on lung nodule detection or other specialized pixel AI—can integrate directly with both the viewer and the reporter. This positions Sirona as a platform for an ecosystem of specialized AI rather than a competitor to it.
Is Sirona’s platform only for teleradiology? No. While the cloud-native architecture makes it an excellent fit for teleradiology and distributed reading, Sirona is designed as a universal PACS and reporting platform for any environment, including hospital systems, imaging centers, and radiology groups providing services across multiple facilities.
Can Sirona be deployed as a reporting-only solution? Yes. Sirona can be implemented as a modular reporting solution, though the platform still benefits from access to pixel data to deliver its full pixel-powered reporting experience—including automated measurements, anatomic segmentation, and bidirectional links between the viewer and the report.
Is Sirona’s reporting solution FDA cleared? Yes. Sirona extends FDA clearance through its reporting application, which the company describes as a differentiator in the category and reflects the clinical rigor required when pixels and reports are tightly integrated into a single workflow.
Ready to see the platform in action? Schedule a demo or watch the on-demand webinar to explore how a unified, AI-driven reporting experience can transform your practice.