Lumen replaces Epic Radiant with a radiology information system that does not merely track images — it understands them. From intelligent protocoling that selects the right study before the technologist touches the scanner, to AI-augmented interpretation that flags findings the human eye might miss, to closed-loop follow-up tracking that ensures no incidental nodule is ever lost to the filing system — Lumen transforms radiology from an image factory into a diagnostic intelligence engine.
The average radiologist interprets one study every three to four minutes across a ten-hour day — a relentless pace that leaves no room for error and no time for the deep clinical correlation that distinguishes a good read from a great one. Meanwhile, the systems they use were designed to move images from a scanner to a screen, not to provide the clinical intelligence that makes those images meaningful. A chest CT arrives without the referring physician's clinical question. A follow-up recommendation is buried in a report that no one tracks. An incidental finding is noted, documented, and forgotten — until the patient returns two years later with advanced disease.
Lumen spans the complete radiology lifecycle — from the moment an imaging order is placed through protocoling, acquisition, interpretation, reporting, communication, follow-up tracking, and quality analytics. Every step is connected to the clinical record through Clarion Scribe and the Sentinel detection engines, ensuring that no image exists in isolation from the patient it serves.
Protocoling is the hidden bottleneck of radiology. Every imaging order must be matched to the correct scanner, the correct acquisition protocol, and the correct contrast administration plan — a decision that depends on the clinical question, the patient's renal function, allergy history, body habitus, and prior imaging. In Epic Radiant, protocoling is a manual radiologist task that consumes hours of physician time on work that does not require diagnostic expertise. Lumen automates 87% of protocol decisions using an AI engine trained on millions of prior studies correlated with clinical indications. The engine reads the ordering diagnosis, checks the patient's GFR, reviews allergy records, identifies prior relevant studies, and selects the optimal protocol — presenting it to the radiologist for one-click approval or modification.
A radiologist reading a chest CT needs more than the image. They need to know why it was ordered, what the clinical question is, what the prior imaging showed, what medications the patient is on, and what the relevant lab values are. Epic Radiant provides a worklist that opens the study; Lumen provides a worklist that opens the patient. When a radiologist selects a study, the workspace loads in under one second with the current images, relevant priors automatically hung in comparison layout, the ordering provider's clinical question, pertinent lab results (including creatinine, tumor markers, and inflammatory markers), active medication list, and any Sentinel AI pre-read findings. The radiologist reads in clinical context from the first glance, not after three clicks into the chart.
Epic Radiant is not a PACS — it is a RIS that connects to a PACS through DICOM interfaces and context-launch URLs. This means the radiology department must maintain two separate systems, two separate databases, and two separate vendor relationships. Lumen takes a different approach: it includes a native DICOM engine that can function as a lightweight enterprise viewer for organizations that want a single platform, while maintaining full integration capability with any third-party PACS — GE Centricity, Philips IntelliSpace, Fujifilm Synapse, Sectra, Intelerad, or any DICOM-compliant archive. The integration uses DICOMweb for modern web-based image retrieval, WADO-RS for zero-footprint viewing, and traditional DICOM C-STORE/C-FIND/C-MOVE for legacy connectivity. Images launch contextually from the patient chart with single sign-on — no separate login, no context switching.
Epic Radiant has no native AI interpretation capability — it displays images and receives reports. Lumen integrates the Clarion Sentinel Visio detection suite directly into the reading workflow. When a chest CT loads on the worklist, the Visio engine has already analyzed it: pulmonary nodules are measured and annotated with Lung-RADS classification, pulmonary emboli are flagged with probability scores, aortic dimensions are automatically measured, and incidental findings in the abdomen are highlighted for secondary review. The radiologist sees these pre-read annotations as an overlay that can be toggled on or off — a second pair of eyes that never fatigues, never rushes, and never misses a 3mm nodule on image 247 of a 400-image dataset.
Critically, the AI does not replace the radiologist. It augments. The pre-read serves as a safety net that catches findings the human eye might miss at the pace of modern clinical practice — particularly on the studies read late at night, at the end of a twelve-hour shift, when cognitive fatigue makes subtle findings invisible.
Technologists are the operational engine of the radiology department — they operate the scanners, position the patients, administer contrast, and document every acquisition parameter. Lumen provides a technologist workspace that streamlines these workflows: modality worklist integration ensures the correct patient and protocol are loaded on the scanner automatically, contrast administration is documented on the patient's medication administration record in real time, and radiation dose is captured from the DICOM dose structured report and tracked longitudinally for every patient. Cumulative dose tracking alerts the ordering physician when a patient's lifetime effective dose approaches guideline thresholds — a safety feature that becomes critical for patients with chronic conditions who undergo repeated imaging.
Two failures define radiology's safety gaps: critical results that are not communicated to the ordering provider in time, and follow-up recommendations that are documented in the report but never acted on. Lumen addresses both through closed-loop tracking systems. When a radiologist identifies a critical finding, Lumen triggers an escalation workflow that documents the communication attempt, the provider reached, the read-back confirmation, and the timestamp — creating the medicolegal-compliant record that every malpractice attorney looks for. For follow-up recommendations, Lumen uses natural language processing to extract recommended follow-up imaging from the report narrative, creates a tracked recommendation linked to the patient's record, and generates automated reminders to the ordering provider and the patient at the appropriate interval. Thirty-five percent of follow-up recommendations are currently lost. In Lumen, zero are.
Breast imaging operates under a unique regulatory and workflow framework that generic RIS systems handle poorly. Lumen includes a dedicated mammography module with MQSA-compliant tracking, BI-RADS structured reporting with outcome audit capability, screening recall management with automated patient letter generation, and AI-assisted breast density assessment. The density assessment engine evaluates each mammogram and assigns a density category that informs supplemental screening recommendations — a requirement in 40 states that notify patients of dense breast tissue. Screening history is tracked longitudinally with automated comparison to prior mammograms, and the system generates the letters, tracks the responses, and escalates non-responders through configurable recall workflows.
Radiology leadership needs visibility into operational performance, quality metrics, and financial productivity. Lumen provides real-time dashboards covering report turnaround time by modality, subspecialty, and radiologist; critical result communication compliance; discrepancy and amendment rates with peer comparison; appropriate use scoring for advanced imaging orders; scanner utilization and throughput; and revenue-per-study analytics. The ACR Appropriateness Criteria integration feeds back into Clarion Mandate at the point of order, creating a closed loop where imaging utilization data informs ordering behavior — not through punitive alerts, but through evidence-based guidance that helps referring physicians order the right study for the clinical question.
A 42-radiologist academic department performing 420,000 studies annually deployed Lumen to replace Epic Radiant and a third-party PACS. The intelligent protocoling engine auto-selected protocols for 87% of studies, eliminating 3.2 hours of daily radiologist protocoling time across the group. The Sentinel Visio pre-read engine identified 1,247 clinically significant findings in the first year that were not initially noted in the preliminary resident interpretation — including 34 pulmonary emboli, 89 incidental renal masses, and 214 pulmonary nodules requiring follow-up. The NLP follow-up tracking system captured 100% of imaging recommendations and achieved an 89% completion rate, compared to the department's prior rate of 62%. Average report turnaround time decreased from 4.2 hours to 3.0 hours across all modalities.
An eight-site community imaging network operating three different PACS platforms consolidated its radiology workflow onto Lumen. The universal DICOM engine federated all three archives into a single reading worklist, eliminating the need for radiologists to log into separate systems for studies performed at different locations. Prior studies from any site were automatically available for comparison regardless of which PACS stored them. Patient dose tracking was unified across all sites for the first time, revealing that 340 patients in the first year had cumulative CT doses exceeding recommended lifetime thresholds — patients who would not have been identified when dose data was siloed across three separate systems. The network saved $420,000 annually in eliminated PACS licensing overlap.
I read 80 to 100 studies a day. At that pace, I will miss things. Every radiologist will. We are human. Lumen's AI pre-read does not replace my judgment — it catches the things I might miss at image 347 of a 400-slice CT when I have been reading for nine hours. In the first year, it flagged 34 pulmonary emboli that were not in the preliminary resident read. Thirty-four. Each one of those is a patient who could have been discharged without treatment. I no longer read alone. I read with a system that never gets tired, never gets distracted, and never looks away.
For years, we documented follow-up recommendations in our reports and hoped someone would act on them. We knew the data — 35% of our recommendations were never completed. But we had no system to track them, no way to close the loop, no process to ensure that the 8mm pulmonary nodule we recommended for 3-month follow-up actually received that follow-up. Lumen changed that equation entirely. Every recommendation is extracted, tracked, and escalated. Our completion rate went from 62% to 89% in the first year. The eleven percent we have not yet reached are the ones I think about. But the twenty-seven percent we recovered — those are the patients whose cancers we will catch early instead of late.
See Lumen configured for your modality mix, your PACS environment, and your reading workflow.