Performance of Automated RAPID Intracranial Hemorrhage Detection in Real-World Practice: A Single-Institution Experience.

BACKGROUND AND PURPOSE: Intracranial hemorrhage (ICH) is a common finding in patients presenting to the emergency department with acute neurological symptoms. Noncontrast head computed tomography (NCCT) is the primary modality for assessment and detection of ICH in the acute setting. RAPID ICH software aims to automatically detect ICH on NCCT and was previously shown to have high accuracy when applied to a curated test data set. Here, we measured the test performance characteristics of RAPID ICH software in detecting ICH on NCCT performed in patients undergoing emergency stroke evaluation at a tertiary academic comprehensive stroke center. MATERIALS AND METHODS: This retrospective study assessed consecutive patients over a 6-month period who presented with acute neurological symptoms suspicious for stroke and underwent NCCT with RAPID ICH postprocessing. RAPID ICH detection was compared with the interpretation of a reference standard comprising a board-certified or board-eligible neuroradiologist, or in cases of discrepancy, adjudicated by a consensus panel of 3 neuroradiologists. Accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of RAPID ICH for ICH detection were determined. RESULTS: Three hundred seven NCCT scans were included in the study. RAPID ICH correctly identified 34 of 37 cases with ICH and 228 of 270 without ICH. RAPID ICH had a sensitivity of 91.9% (78.1%-98.3%), specificity of 84.4% (79.6%-88.6%), NPV of 98.7% (96.3%-99.6%), PPV of 44.7% (37.6%-52.1%), and overall accuracy of 85.3% (80.9%-89.1%). CONCLUSIONS: In a real-world scenario, RAPID ICH software demonstrated high NPV but low PPV for the presence of ICH when evaluating possible stroke patients.

Central Triage of Acute Stroke Patients Across a Distributive Stroke Network Is Safe and Reduces Transfer Denials.

Background and Purpose: Mechanical thrombectomy has dramatically increased patient volumes transferred to comprehensive stroke centers (CSCs), resulting in transfer denials for patients who need higher level of care only available at a CSC. We hypothesized that a distributive stroke network (DSN), triaging low severity acute stroke patients to a primary stroke center (PSC) upon initial telestroke consultation, would safely reduce transfer denials, thereby providing additional volume to treat severe strokes at a CSC. Methods: In 2017, a DSN was implemented, in which mild stroke patients were centrally triaged, via telestroke consultation, to a PSC based upon a simple clinical severity algorithm, while higher acuity/severity strokes were triaged to the CSC. In an observational cohort study, data on acute ischemic stroke patients presenting to regional community hospitals were collected pre- versus post-DSN implementation. Safety outcomes and rate of CSC transfer denials were compared pre-DSN versus post-DSN. Results: The pre-DSN cohort (n=150), triaged to the CSC, had a similar rate of symptomatic intracerebral hemorrhage and discharge location compared with the post-DSN cohort (n=150), triaged to the PSC. Time to stroke unit admission was faster post-DSN (2 hours 40 minutes) versus pre-DSN (3 hours 29 minutes; P<0.001). Transfer denials were reduced post-DSN (3.8%) versus pre-DSN (1.8%; P=0.02), despite an increase in telestroke consultation volume over the same period (median, 3 calls per day pre-DSN versus 5 calls per day post-DSN; P=0.001). No patients who were triaged to the PSC required subsequent transfer to the CSC. Conclusions: A DSN, triaging mild ischemic stroke patients from community hospitals to a PSC, safely reduced transfer denials to the CSC, allowing greater capacity at the CSC to treat higher acuity stroke patients.

Changes in events alter how people remember recent information.

Observers spontaneously segment larger activities into smaller events. For example, "washing a car" might be segmented into "scrubbing," "rinsing," and "drying" the car. This process, called event segmentation, separates "what is happening now? from "what just happened." In this study, we show that event segmentation predicts activity in the hippocampus when people access recent information. Participants watched narrative film and occasionally attempted to retrieve from memory objects that recently appeared in the film. The delay between object presentation and test was always 5 sec. Critically, for some of the objects, the event changed during the delay whereas for others the event continued. Using fMRI, we examined whether retrieval-related brain activity differed when the event changed during the delay. Brain regions involved in remembering past experiences over long periods, including the hippocampus, were more active during retrieval when the event changed during the delay. Thus, the way an object encountered just 5 sec ago is retrieved from memory appears to depend in part on what happened in those 5 sec. These data strongly suggest that the segmentation of ongoing activity into events is a control process that regulates when memory for events is updated.