Early stroke detection and extent: impact of experience and the role of computed tomography angiography source images

AIM: To test the performance of computed tomography angiography "source images" (CTA-SI) versus unenhanced CT (NCCT) for stroke detection and extent using the Alberta Stroke Programme Early CT Score (ASPECTS), and examine the effect of experience and clinical history. MATERIALS AND METHODS: Studies of 23 consecutive patients presenting within 4.5 h were analysed by three reviewers of varying experience. Each reviewer, blinded to clinical information reviewed a random order of NCCT and CTA-SI and documented side of infarct and the ASPECTS. The readings were repeated for CTA-Sl with and without clinical information. Performance measures and observer agreement were calculated. Applying an ASPECTS threshold of <= 7, the number of patients misclassified was determined. R ESULTS: CTA-SI improved trainee accuracy by 9%, but had little impact on more experienced readers. The accuracy and sensitivity of stroke extent assessment was increased for all readers, but was greatest for the trainee (17% and 12%, respectively). Clinical history contributed little to CTA-SI accuracy. Observer agreement was higher for CTA-SI. NCCT could have resulted in the misclassification of more patients than CTA-SI. CONCLUSION: CTA-SI are an important adjunct in acute stroke assessment, improving stroke extent determination for all readers irrespective of level of experience. In addition less experienced readers may benefit from CTA-SI for detection of presence of strokes. CTA-SI performance appears independent of clinical history. CTA-SI resulted in fewer misclassified patients if an ASPECTS threshold of <= 7 is considered. (c) 2007 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.