Acute Ischemic Stroke Lesion Measurement on Diffusion- weighted Imaging–Important Considerations in Designing Acute Stroke Trials With Magnetic Resonance Imaging Carly S. Rivers, PhD, MSc,† Joanna M. Wardlaw, FRCR, FmedSci,* Paul A. Armitage, PhD,* Mark E. Bastin, Dphil,‡ Peter J. Hand, MB ChB, MD, MRCP,§ and Martin S. Dennis, MD, FRCP* Background: In acute ischemic stroke, magnetic resonance diffusion-weighted imaging (DWI) is increasingly used to select patients for inclusion or as a surrogate outcome marker in clinical trials, or in routine practice. Little is known of what factors might affect DWI lesion size measurement. We examined morphologic factors that might affect DWI lesion measurement. Methods: On DWI obtained less than 24 hours after stroke, we categorized lesions according to DWI appearance (solitary or multifocal; well-deﬁned or ill-deﬁned edges), lesion size (/5 cm 3 ), and time to imaging (6, 6 –12, and 12–24 hours). Two observers (senior neuroradiologist; less-experienced imaging neuroscientist) measured all lesions. In 4 representative cases we assessed DWI lesion volume using two apparent diffusion coefﬁcient thresholds (0.55 and 0.65  10 -3 mm 2 /s). Results: Among 63 patients (33% imaged  6 hours after stroke), the neuroradiologist measured larger lesion volumes than the imaging neuroscientist (median 4.29 v 3.50 cm 3 , respectively, P  .01). Differences between observers were greatest in patients scanned within 6 hours of stroke, in multifocal ill-deﬁned or large lesions (all P  .01). Both apparent diffusion coefﬁcient thresholds underestimated lesion extent and included remote normal tissue, particularly in multifocal ill-deﬁned large lesions. Conclusion: DWI lesion characteristics inﬂuence lesion volume measure- ment. Large, multifocal, ill-deﬁned DWI lesions obtained in less than 6 hours have the greatest variability. Trials using DWI should account for this in their study design. Key Words: Magnetic resonance imaging— diffusion weighted imaging— observer re- liability—stroke— cerebrovascular disease. © 2007 by National Stroke Association Magnetic resonance (MR) diffusion-weighted (DW) imaging (DWI) is popular in acute stroke because it shows even very small ischemic lesions very early after symptom onset. 1 DWI has been used to select patients for several acute stroke treatment trials, 2-4 and its popularity is increasing. Any method used for clinical decision-making, whether in a clinical trial or in day-to-day practice, should be reliable and reproducible. Previous studies did not assess what factors might inﬂuence DWI lesion mea- surement reliability, either of DWI lesion extent by visual estimation (e.g., / 1/3 middle cerebral artery terri- tory), 5-9 or by volume measurement on a worksta- tion. 10-14 Indeed, as these studies included many patients with subacute, well-deﬁned lesions, they may have over- estimated DWI lesion measurement reliability. Key fac- tors such as lesion size, lesion distribution (solitary, multi- focal), and edge clarity/signal intensity might affect DWI From the *Division of Clinical Neurosciences and ‡Medical Phys- ics, University of Edinburgh, Western General Hospital, United Kingdom; †Clinical Trials Research Unit, University of Leeds, United Kingdom; and §Department of Neurology, Royal Melbourne Hospi- tal, Victoria, Australia. Received July 3, 2006; revision received October 30, 2006; accepted November 9, 2006. Supported by the Scottish Executive’s Chief Scientist Ofﬁce (Ref- erence C2B/4/14), the Row Fogo Charitable Trust, and the Royal Society of Edinburgh/Lloyds TSB Foundation (Dr Rivers). The work was conducted at the SFC Brain Imaging Research Centre for Scot- land (www.dcn.ed.ac.uk/bic). C:\data\papers\DWI other\dwi iov JSCVD 301006. Address correspondence to Joanna M. Wardlaw, FRCR, FmedSci, Division of Clinical Neurosciences, Western General Hospital, Crewe Rd, Edinburgh EH4 2XU United Kingdom. E-mail: joanna.wardlaw@ ed.ac.uk. 1052-3057/$—see front matter © 2007 by National Stroke Association doi:10.1016/j.jstrokecerebrovasdis.2006.11.003 Journal of Stroke and Cerebrovascular Diseases, Vol. 16, No. 2 (March-April), 2007: pp 64-70 64