Page 235

Karger_ESC London_2013

London, United Kingdom 2013 12 Brain imaging Voxel-based mapping of hemodynamic impairment in symptomatic carotid disease: an oxy-gen- E-Poster Session Red Cerebrovasc Dis 2013; 35 (suppl 3)1-854 235 15 PET study R.R. Moustafa1, P.S. Jones2, I. Momjian-Mayor3, F.I. Aigbirhio4, T.D. Fryer5, E.A. Warburton6, J.- C. Baron7 Ain Shams University, University of Cambridge, UK, Cairo, EGYPT1, University of Cambridge, Cambridge, UNITED KINGDOM2, University of Cambridge, Cambridge, UNITED KINGDOM3, University of Cambridge, Cambridge, UNITED KINGDOM4, University of Cambridge, Cambridge, UNITED KINGDOM5, University of Cambridge, Cambridge, UNITED KINGDOM6, University of Cambridge, Cambridge, UNITED KINGDOM7 Background Hemodynamic impairment distal to severe internal carotid artery (ICA) stenosis/occlusion can be demonstrated using quantitative 15O-PET as increased mean transit time (MTT), cerebral blood volume (CBV) and oxygen extraction fraction (OEF). The spatial distribution and extent of such ab-normalities is traditionally defined using the region-of-interest (ROI) approach, which is restricted by the choice of ROI size, shape and location. Here we tested for the first time an observer-indepen-dent voxel-based method to objectively map the 3D topography of 15O-PET defined hemodynamic changes in a previously reported sample (Moustafa, Stroke, 2011). Methods Triple-O15 PET was performed in 16 pts with non-cardioembolic stroke/TIA and ≥70% ICA stenosis or occlusion, and 10 age-matched controls. Transcranial Doppler (TCD) was used to clas-sify patients according to presence or not of microembolic signals (MES). After normalization of quantitative perfusion and OEF maps and appropriate masking of venous pools, voxel-based analy-sis using SPM8 was done separately for each variable. The symptomatic hemisphere was compared between patients and controls using an independent sample t-test analysis, and between MES+ and MES- patients using a multi-group analysis. Based on strong a priori hypotheseis the p<0.001 de-fault cut-off was used. Results Significant delay in MTT was found in patients in confluent frontal, parietal and temporal regions, extending into the centrum semiovale (CSO). Increased CBV was found in the central CSO and frontal areas, and elevated OEF in small frontal areas. Hemodynamic impairment (MTT and CBV) was significantly greater and more extensive in MES- than in MES+ patients. Conclusion Voxel-based analysis accurately depicted the expected occurrence and topography of hemodynam-ic impairment in patients vs. controls and in MES- vs MES+ subgroups. This study documents that voxel-based analysis of 15O-PET datasets is feasible and applicable in future studies. 11 Brain imaging A relative signal intensity cut off can improve interrater agreement when judging FLAIR posi-tivity in acute ischemic stroke patients I. Galinovic1, J. Puig2, L. Neeb3, J. Guibernau4, A. Kemmling5, S. Siemonsen6, S. Pedraza7, B. Cheng8, G. Thomalla9, J.B. Fiebach10 Center for Stroke Research Berlin, Berlin, GERMANY1, Hospital Universitari Dr Josep Trueta, Girona, SPAIN2, Center for Stroke Research Berlin, Berlin, GERMANY3, Hospital Universitari Dr Josep Trueta, Girona, SPAIN4, University Medical Center Hamburg-Eppendorf, Hamburg, GERMA-NY5, University Medical Center Hamburg-Eppendorf, Hamburg, GERMANY6, Hospital Universi-tari Dr Josep Trueta, Girona, SPAIN7, University Medical Center Hamburg-Eppendorf, Hamburg, GERMANY8, University Medical Center Hamburg-Eppendorf, Hamburg, GERMANY9, Center for Stroke Research Berlin, Berlin, GERMANY10 Background The match or mismatch between visibility of an ischemic lesion in diffusion-weighted imaging and its visibility on fluid-attenuated inversion recovery (FLAIR) images is considered to stand for an estimate of time from stroke onset. A truly accurate relative signal intensity (rSI) cut off for judging time based on such FLAIR hyperintensities has however thus far eluded discovery. Therefore qual-itative assessment of the presence of a FLAIR hyperintensity remains the standard, but suffers from rater bias and low interrater agreement. In preparing for a multicenter, randomised, placebo-con-trolled trial of efficacy and safety of MRI-based thrombolysis in wake-up stroke (WAKE-UP) we sought to assess and improve interrater agreement stemming from our study’s specific imaging crite-ria. Methods A sample of images from 143 acute stroke patients was used for training 6 raters to qualitatively evaluate images based on the WAKE-UP study criteria. In a subsample of 45 cases a quantitative analysis was also carried out, with 5 raters placing regions of interest on FLAIR images, inde-pendently from one another and using the hot-spot technique (which delineates the area of brightest signal), to produce rSI values. These values were compared against a consensus decision (regarding FLAIR positivity) reached by two expert neuroradiologists. Results For the entire sample, interrater agreement between the 6 raters regarding patient inclusion or ex-clusion was 73% with a free-marginal kappa of 0.455. In the subsample analysis, a rSI cut off value of 1.20 had a high true positive rate (between 0.870 and 0.957) and low false positive rate (between 0.045 and 0.158) for agreeing with the expert consensus decision regarding FLAIR positivity. Conclusion Interrater agreement for qualitative assessment based on WAKE-UP imaging criteria was moderate. Introducing a fixed cut off value of rSI, as an aid to visual assessment of FLAIR positivity, stands to improve interrater agreement.


Karger_ESC London_2013
To see the actual publication please follow the link above