Colored Nonenhanced Head Computed Tomography Decreases Assessment Time Without Compromising Diagnostic Accuracy from Trainees to Expert Neuro-radiologists in the Setting of Acute Ischemic Stroke

PURPOSE: To improve diagnostic speed and accuracy for the detection of acute ischemic stroke utilizing colored nonenhanced head computed tomography (NECT) images. METHODS: This multi-reader prospective observational study included 100 consecutive adult patients with concern for acute ischemic stroke from 2/1/2018 to 3/17/2018 with grayscale NECT and CT angiogram (CTA) head images and a reference-standard confirmatory brain MRI. Grayscale NECT and CTA head images were collected and colored NECT head images were generated using a custom, fully automated software designed to color the intracranial contents on the head CT images and increase the conspicuity of ischemic strokes. Two randomized imaging sets were generated from each patient's CT exam, including Grayscale and Color+Grayscale images. Two radiology residents and two expert fellowship trained neuroradiologists independently assessed each imaging set, separated by at least two weeks. The mean accuracy, sensitivity, specificity, and time of assessment were compared between individual readers and between the two reader groups, in a multivariate model. RESULTS: Among the 4 readers, the mean accuracy/sensitivity/specificity for correctly diagnosing acute ischemic stroke were 72%/46%/87% using grayscale NECT images only and 69%/36%/86% using Color+Grayscale NECT images (p=0.08/p=0.006/p=0.858). The mean time of assessment with grayscale NECT images only was 59 seconds, which was decreased by an average of 19 seconds when using Color+Grayscale NECT images (p