Radiology worklist management discussions often focus on average throughput and turnaround times. Both matter. But when the conversation turns specifically to intracranial hemorrhage and pulmonary embolism, average metrics obscure the real clinical issue. These two finding categories carry time-to-read stakes that no amount of excellent average performance can compensate for when a specific case is buried in a queue.
Why Time Matters Differently for These Two Findings
Intracranial hemorrhage and pulmonary embolism share a critical characteristic: the clinical management window is tightly time-dependent in a way that other serious findings are not. This is not a statement about which findings are clinically important. It is a statement about the specific relationship between elapsed time and clinical outcome for these two categories.
For intracranial hemorrhage, the concern is hematoma expansion. In spontaneous intracerebral hemorrhage, significant hematoma expansion occurs in a substantial proportion of patients in the first several hours. Neurosurgical consultation, blood pressure management targets, and decisions about reversal of anticoagulation all depend on a radiologist's report being available. The window for certain interventions, particularly in hemorrhagic transformation of ischemic stroke, is even shorter. Time from CT acquisition to radiologist read directly determines when the clinical team can act.
For pulmonary embolism, the relevant time frame is different but similarly compelling. Central and saddle PE findings warrant hemodynamic monitoring and immediate anticoagulation decision-making. The right ventricular strain findings on CT can indicate submassive PE risk stratification that influences whether a patient goes to the ICU, gets catheter-directed thrombolysis, or is managed on the floor. These are clinical decisions that cannot be made until the imaging report exists. A CTPA that sits in a queue behind 18 routine studies is a study whose clinical management is on hold for the duration of that wait.
How These Cases Arrive on the Worklist
The most straightforward scenario is a study ordered stat in an emergency context: the patient in the ED with sudden-onset headache and maximal-severity presentation, or the post-surgical patient with acute respiratory deterioration. These studies should arrive stat-flagged and surface quickly in standard worklist priority tiers. This scenario works reasonably well in most departments.
The scenario that works less well is the study ordered outside an obvious acute context. Consider a patient presenting to an urgent care center with two days of worsening dyspnea and mild pleuritic chest pain. The clinician orders a CTPA, but the clinical picture does not scream PE; the patient is stable, has no tachycardia at triage, and the order goes in at routine priority. The study arrives in the worklist as routine, takes its place in the sequence, and gets read in arrival order. If the scan shows a bilateral pulmonary embolism with moderate right heart strain, the clinical team is waiting for a report that has been queue-ordered behind cases that arrived before it.
A similar pattern plays out with hemorrhage. A patient with a mild traumatic brain injury, hemodynamically stable, CT ordered for standard evaluation. Clinically, nothing signals an urgent read at order time. The scan shows a small but expanding subdural hematoma. Arrival-order worklist position means that finding waits behind whatever was ordered before it.
The Triage Architecture for Acute Findings
The priority hierarchy for worklist triage needs to account for the difference between known urgency at order time and discoverable urgency post-acquisition. Known urgency is handled by stat ordering; it works when it is used appropriately. Discoverable urgency requires a different mechanism: evaluation of scan content after acquisition to identify findings that were not predictable from the clinical presentation.
For intracranial hemorrhage and PE, the imaging characteristics that distinguish these findings from negative studies are well-defined and consistent enough for reliable triage model training. Parenchymal density changes in hemorrhage, filling defects in contrast-enhanced CTPA for PE, and associated secondary findings (sulcal effacement, right ventricular enlargement) are features that a trained model can evaluate to generate a priority signal before the radiologist begins the session.
ImageAssist operates this way: DICOM study received, model evaluation, worklist priority update via RIS API, all within under 60 seconds under normal conditions in our pilot deployments. The radiologist opening their session sees the flagged study at position 1. Everything after that is standard clinical radiology. The radiologist reads, confirms or does not confirm the finding, generates the report, and initiates the critical findings communication workflow.
The Role of Critical Findings Protocols
Critical findings protocols are the standard mechanism for rapid communication once a finding has been read. ACR guidelines and The Joint Commission standards establish expectations for radiologist-to-referring-clinician communication timelines for critical findings. These protocols are important and they function well in most departments. They address the communication phase of the workflow.
What they do not address is the pre-read phase: the time between when a study is available for read and when a radiologist actually opens it. Critical findings protocols assume the radiologist has already reached the study. Triage prioritization addresses the upstream problem: getting the radiologist to the study that needs the protocol invoked as quickly as possible.
We want to be clear that worklist triage is not a substitute for critical findings communication protocols. Both layers are necessary. Triage shortens the time to read; critical findings protocols ensure the read generates a rapid clinical response. They operate sequentially and should both be present in a well-functioning department.
Thinking About False Negatives
No triage model has perfect sensitivity. A triage system for hemorrhage and PE will occasionally miss findings: a subtle subarachnoid hemorrhage in a study where the blood is isodense to brain on a particular acquisition, or a peripheral subsegmental PE that is near the edge of spatial resolution. These false negatives mean a study that should have been flagged was not.
The honest way to think about this is that false negatives in a triage system represent cases that were handled exactly as they would have been before the system was deployed: arrival-order position, normal workflow. The system did not make those cases worse; it failed to improve them. The goal of triage deployment is to reduce the proportion of acute findings that reach the radiologist with unnecessary delay, not to guarantee zero delay for every acute finding. A system that flags 85 to 90 percent of acute findings for priority position is substantially better than a system that flags none, even if it misses the remaining fraction.
The false positive rate matters in the opposite direction: false positives mean studies that were flagged as potentially urgent turn out on radiologist read to be negative. The radiologist spends a few extra minutes on a priority read that was normal. This is a real cost, and threshold tuning for a given department's volume and case mix can manage it. The clinical evidence page describes how we approach this calibration question.
What Priority Position Actually Changes
For a department running a 12-hour shift with 150 to 200 chest CTs on the worklist, moving a hemorrhage or PE study from position 40 to position 1 changes the expected time-to-read from several hours to minutes. In the scenario above where a bilateral PE with right heart strain is at position 18 in a routine-priority queue, repositioning it to position 1 may compress a 90-minute wait to under 5 minutes. The radiologist reads the study, reports the findings, and the clinical team has the information they need to act.
This is the specific operational improvement that ImageAssist is designed to deliver for these two finding categories. If your department has had adverse events or near-misses connected to delayed reads of hemorrhage or PE, or if your worklist audit shows longer-than-expected time-to-read for critical findings, the clinical team can discuss how triage deployment would work at your site.
