BrainPET Poster Session: Brain Imaging—Stroke

276. Perfusion instability in the penumbra influences infarct probability in acute ischemic stroke patients

J.-M. Lee¹, H. An², A.L. Ford¹, K.D. Vo³, A.M. Nassief¹, C.P. Derdeyn³, W.J. Powers⁴ and W. Lin²

¹Neurology, Washington University School of Medicine, Saint Louis, Missouri; ²Radiology, University of North Carolina, Chapel Hill, North Carolina; ³Radiology, Washington University in St Louis, St Louis, Missouri; ⁴Neurology, University of North Carolina, Chapel Hill, North Carolina, USA

Objectives: It has been postulated that cerebral blood flow (CBF) and metabolism are unstable during the acute phase of cerebral ischemia. However, animal models which rely on fixed occlusions of cerebral vessels, do not allow study of this phenomenon; and human studies have largely relied on single measures during acute ischemia. In this study, we sought to serially examine MR-measured CBF during hyperacute ischemia in a cohort of stroke patients to assess the impact of CBF fluctuations on infarct probability.

Methods: Eight acute ischemic stroke patients were imaged at three time-points: 2.7±0.8 h (tp1) and 6.2±0.2 h (tp2) after symptom onset; and at 1 m (tp3) to delineate the region of the final infarct (FLAIR image). Six patients received intravenous tPA prior to tp1 while two patients did not. Dynamic susceptibility contrast method was used to obtain CBF. Measurements were normalized to the contralateral unaffected hemisphere. Co-registration and tissue segmentation were performed to align the timepoints and separate gray from white matter, respectively. Brain voxels were classified into CBF deciles at tp1, and further subclassified (based on tp2 measures of the same voxel) into voxels with increased CBF versus without increased CBF. Probability of infarction (based on co-registered tp3 FLAIR images) was calculated for each decile and compared between tp1 alone and tp1 with increased CBF at tp2 versus tp1 without increased CBF at tp2. These 3 groups were compared using a repeated-measure ANOVA with post-hoc Newman-Keuls test.

Results: Shown are ΔCBF maps from 2 patients (Figure panel A and B), representing CBF changes between tp1 and tp2 (red increase, green no change, blue decrease). CBF showed significant variability from tp1 to tp2 in all patients. To examine if this CBF variability contributed to infarct probability, 3 probability curves were created and compared. The infarct probability of CBF deciles at tp1 was compared to:

• (1)

  that of the subset of voxels with increased CBF at tp2 versus.

• (2)

  that of the subset of voxels without increased CBF at tp2 (Figure panel C).

Figure

A lower probability of infarction was predicted when tp2 CBF improvement (red triangles) was considered compared to considering tp1 data alone (black circles, P<0.05); likewise, a higher probability of infarction was predicted when tp2 CBF non-improvement (blue squares) was considered compared to considering tp1 data alone (black circles, P<0.05). Moreover, altered probabilities were most notable in the CBF ranges from 20% to 50%, which likely represent the theoretical penumbra.

Conclusion: CBF is highly dynamic during acute ischemia and impacts infarct probability, especially in the ‘penumbral range’ of CBF (20% to 50%). These findings underscore the importance of dynamic changes in CBF during acute ischemia, and caution that single time-point imaging belies the complex physiology during acute ischemia.

332. A cortical region and neuropathway necessary for reading chinese characters revealed by structural and functional MRI

C. Shan^1,2, T. Wang¹, R. Zhu³, X. Zhao³, Z. Lu³, X. Zhou³, X. Weng² and B. Luo³

¹Department of Rehabilitation Medicine, First Affiliated Hospital of Nanjing Medical University, Nanjing; ²Laboratory for Higher Brain Function, Institute of Psychology, Chinese Academy of Sciences, Beijing; ³Department of Neurology, First Affiliated Hospital, College of Medicine, Zhejiang University, Hangzhou, China

Objectives: To examine the crucial region and neural pathway necessary for Chinese logographic characters reading by using structural and functional MRI.

Methods: Three right-handed patients suffering from infarcts in the territory of the left posterior cerebral artery (PCA) were studied. All of them showed difficulties in naming familiar objects and people. Patient JXD was of inability to read aloud and comprehend Chinese characters without any other language deficits, i.e. pure alexia. ¹ Patient CYH only had deficits in naming objects and people, his reading was normal, i.e. anomia. Patient KY could not understand written Chinese material because he made errors in recognizing the left part (i.e. radical) of a Chinese character, e.g. he read (zhi4, order) as (tie3, iron), read (yin2, silver) as (gen1, root), i.e. left hemiparalexia. ² 3D-SPGR MRI scan and specific analysis software were used to overlap the lesions of three patients to distinguish the distinct lesions in pure alexia (JXD) and left hemiparalexia (KY). BOLD fMRI was adopted for JXD and CYH to reveal activations induced by Chinese characters reading. For KY, we used fMRI to demonstrate the activations to the Chinese characters presented in the left and right visual fields and made DTT (diffusion tensor tractography) examination to reveal the interruption of neuropathway (white matter).

Results: The analysis of 3D-SPGR MRI data reveals that the common lesions lie in the left medial ventral occipitotemporal cortex. JXD had additional lesions in the left lateral mid-fusiform cortex, where has been labeled as visual word form area (VWFA) by Cohen et al. ³ Compared with CYH, KY had additional lesion in the left splenium of corpus callosum. DTT confirms that the neuropathway through splenium was broken in KY. The fMRI results show Chinese characters activated CYH's left lateral mid-fusiform cortex (corresponding to VWFA) but not JXD's (see Figure 1). For KY, this area was activated only when Chinese characters were presented in his right visual field (RVF) (see Figure 1).

Figure 1.

The activations evoked by Chinese characters. The crosshair correspond to VWFA (for JXD and CYH, TC: x = −43, y = −54, z = −12; for KY, TC: x = −38, y = −65, z = −1).

Conclusions: The left lateral mid-fusiform cortex is a crucial region for Chinese characters reading (recognizing). Therefore, the lesion in this area (JXD) may result in pure alexia. The neuropathway going through splenium of corpus callosum is necessary for transferring of visual information of Chinese characters from left visual field (right occipital cortex) to left VWFA. The infarction in this pathway (KY) can lead to alexia for characters/radicals in left visual field (left hemialexia or left hemiparalexia). Both the cortex and neuropathway crucial for Chinese logographic characters reading are similar to that of alphabetic word reading.

356. Separating recoverable and permanent damage post stroke with CASL FMRI stress testing and 18-FDG PET

G. Deutsch¹, J. O'Malley¹, R. Menon², V. Mark³, B. Corbitt¹, H. Liu¹, J. Den Hollander⁴ and J. Halsey⁵

¹Radiology, Nuclear Medicine Division, University of Alabama at Birmingham; ²Biomedical Engineering; ³Rehabilitation Medicine; ⁴Cardiology; ⁵Neurology, University of Alabama at Birmingham, Birmingham, Alabama, USA

Introduction: Stroke typically results in neurometabolic activity defects in regions distal but functionally connected to the infarct, usually representing deafferentation effects of the lesion, a physiological phenomenon known as diaschisis. We tested a method of continuous arterial spin-labeled (CASL) perfusion MRI, during baseline and cerebrovascular stress, to identify and quantify regional activity reductions outside the lesion that should represent recoverable brain. 18-FDG PET was also performed to assist interpretation.

Materials and methods: We studied 6 patients (mean age 57, range 19 to 72 years) within 2 to 4 weeks post unilateral stroke involving the middle cerebral artery (MCA), with some variation in the extent of cortical and deeper white and gray matter involvement. MR Sequences: Patients and 22 controls were imaged on a 3T, Philips Intera scanner, using a transmit/receive head coil. Supratentorial slices covering cerebrum to cerebellum (single shot spin echo planar imaging; adiabatic-through-fast-passage labeling pulse) were acquired continuously in ascending order during resting-baseline and inhalation of 5% CO2. 18-FDG PET was performed on a GE 8-slice Discovery-LS PET-CT using intravenous injection of 10 mCi F-18 FDG and 3-D acquisition. Data Analysis: CASL data were stored as raw echo amplitudes and transferred for rCBF computations using custom software written in MATLAB™. Forty pairs of labeled and control images were motion corrected and averaged to produce a single set of perfusion images. Changes in rCBF were analyzed using an ROI method, a method for estimating relative functional versus anatomical defect size [Mountz, SemNucMed 2003;33:56–76], and compared to controls with Statistical Parametric Mapping (SPM).

Results: Figure 1 depicts an angulated transverse section from a patient for T1 weighted MRI, CASL at rest and during CO2 stress, and 18-FDG PET. Figure 2 plots rCBF values and PET-18FDG as percent of mean in ROI segments formed by a cortical circumferential profile analysis. All patients showed a significant increase in rCBF between rest and stress and a reduction in the metabolic-to-anatomical ‘defect volume’ (Table 1). The reduced defect volume quantifies an extended penumbra in the affected hemisphere. 18-FDG-PET showed reduction in activity distal to the infarct coinciding with the resting state rCBF reductions. Figure 3 shows SPM analysis of the case above compared to controls. Significant (P<0.01) contralesional as well as ipsilesional hemispheric reductions are seen. Analysis of the CO₂ scan showed differences only within the lesion perimeter. Figure 4 shows absolute rCBF during rest and CO2 in one slice, with ROIs 7,8 improving most.

Table 1

Lesion volumes (cc) from MRI; total rCBF defect (cc) calculated at rest and during CO₁ stress using CASL; and FDG PET defect (cc) calculated at rest for 6 patients

Case	MRI lesion	rCBF Defect		PET
		Rest	CO 2
1	15	31	20	—
2	5	18	9	16.3
3	19	45	26	39.8
4	18	41	30	68.0
5	18	38	26	60.8
6	27	49	31	69.5

Fig 1

Structural MRI, rCBF map at rest with ROTs drawn, rCBF map with CO₂ and FDG PET image at Rest

Fig 2

CASL based rCBF values versus FDG PET at rest for 12 ROIs (% of mean activity)

Fig. 3a

Resting State Controls – Patient

Fig 3b

CO₂. Stress Contois – Patient

Fig 4a

ROI Schematic

Fig 4b

rCBF (ml/100g/min) at rest

Fig 4c

rCBF (ml/100g/min) during CO₂

Discussion: The dramatic improvement in rCBF seen during CO₂ stress outside the perimeter of the infarcted region strongly suggests primary neurometabolic reduction at rest and represents diaschisis or deafferentation effects. These regions also represent viable brain with recovery potential and probably account for some portion of the patients' current cognitive and motor deficits.

565. Hyperintense vessel sign on fluid-attenuated inversion recovery images associates with occluded middle cerebral artery

T. Kono¹, H. Naka², S. Aoki¹, H. Ueno¹, T. Ohshita¹, E. Nomura², T. Ohtsuki¹, T. Kohriyama¹, S. Wakabayashi³ and M. Matsumoto¹

¹Neurology, Hiroshima University Hospital; ²Neurology; ³Neurosurgery, Suiseikai Kajikawa Hospital, Hiroshima, Japan

Objectives: Hyperintense vessel sign (HVS) ¹ that MRI shows as high intensity signal on fluid-attenuated inversion recovery (FLAIR) images is as useful as diffusion-weighted images for early detection of an occluded artery. The aim of this study is to evaluate the frequency of HVS appearance and the following disappearance under whether the occluded vessel being reperfused or not.

Methods: From retrospective case series between January 2004 and March 2007 in Suiseikai Kajikawa Hospital, we selected the patient of ischemic stroke undergoing both 1.5T MRI DWI/ FLAIR images and MR angiography within 24 h of onset. Patients with symptomatic stenosis or occlusion in M1 segment of the middle cerebral artery (MCA) were included and divided into either cardioembolic infarction or large vessel disease according to the Trial of ORG 10172 in Acute Stroke Treatment (TOAST) classification. The extent of stenosis of M1 segment in each patient was classified into the 3 grades as follows: mild (signal reduction), severe (focal signal loss with the presence of distal MCA) and occlusion. The second MRI was performed if medical doctors permitted for necessary. The 2 neurologists assessed every images independently.

Results: A total of 42 patients were enrolled (25 men, 33 to 93 years old; median age 70 years old). The interval from onset to the first MRI was 33 mins to 21 h; median 3 h 35 mins. The pathogenesis of ischemic stroke was either cardiogenic embolism (n = 21) or large vessel disease (n = 21). HVS was detected in 37 patients (88%), and was present in all (100%) of the 21 patients with cardioembolic infarction, that consisted of 19 ‘occlusion’ and 2 ‘severe stenosis’ by the grade. In contrast, 16 (76%) of the 21 patients with large vessel disease had positive HVS; all (100%) of 11 ‘occlusion’ and 4 (44%) of 9 ‘mild to severe stenosis’ showed positive HVS. Not only cardioembolic infarction but also large vessel disease showed positive HVS without statistical difference (P = 0.41). The second MRI was obtained in 22 patients on the median 4 days after the first MRI. HVS disappeared in 11 (85%) of 13 patients under MCA recanalization, while it disappeared in 1 (11.1%) out of 9 if MCA failed to recirculate (OR 44.0, 95% CI: 3.08–655, P = 0.03).

Conclusions: A positive HVS associated with occluded MCA (M1) on cardioembolic infarction or large vessel disease and disappearance of HVS correlated with reperfusion of the once occluded MCA. HVS could be as useful for early detection of MCA occlusion and recanalization well as MRA.

586. DTI-parameters of cortico-spinal tract integrity in subacute subcortical stroke

B. Radlinska^1,2, S. Ghinani^1,2, I. Lepert³ and A. Thiel^1,2

¹Neurology & Neurosurgery, McGill University; ²Neurological Sciences, SMBD Jewish General Hospital and Lady Davis Institute for Medical Research; ³MR Imaging Unit, McConnell Brain Imaging Centre, Montreal, QC, Canada

Objectives: Studies investigating motor recovery after stroke have employed functional imaging to describe recovery related changes in functional cortical networks. The role of white matter damage for post-stroke recovery is less well understood. With diffusion tensor imaging (DTI), it became possible to image fiber tracts in the living human brain. The objective of this study is to determine to which extend different DTI parameters like fiber tract volume (FTV) and fractional anisotropy (FA) reflect damage to the corticospinal tract (CST) in subcortical stroke and how they are related to clinical symptoms.

Methods: We investigated 8 right-handed patients with first subcortical ischemic infarct affecting the CST within 10 days after stroke, as well as 5 age matched controls. DTI was performed on a 3T Siemens Trio scanner using a single-shot echo-planar sequence and parallel reconstruction. 64 × 2 diffusion weighted images were acquired using isotropically spaced diffusion encoding directions and a b value of 1000 s/mm² and a total of 2 images with b = 0 s/mm² respectively. In the stroke hemispheres, fibers were traced retrogradely from a standardized seed region in the cerebral peduncle through the infarct using a deterministic algorithm.^1,2 In non-affected hemispheres an identical seed region was used and tracts were traced through the posterior limb of the internal capsule. Mean FA within the tracts and FTV were compared within and between the groups and related to clinical severity (NIH scale). Tract portions above and below the level of the infarct/internal capsule were investigated separately.

Results: CST was successfully traced in both the affected and non-affected hemispheres of patients as well as in all controls. There were no significant differences in mean TVA and FA between left and right CST of controls (TVA_LEFT 11200 +/− 3034.263 mm, TVA_RIGHT 10480 +/− 1885 mm³; FA_LEFT 0.499 +/−0.0298, FA_RIGHT 0.494 +/− 0.0379). The CST volume of patients was significantly reduced on the affected side (TVA_AFFECTED 4726 +/− 1660 mm³, TVA_{NON−AFFECTED} 9896 +/−3144 mm³, P⩽0.001). Mean FA was also reduced in the affected CST (FA_AFFECTED 0.425 +/− 0.0501, FA_{NON−AFFECTED} 0.466 +/− 0.0310, P⩽0.10). Furthermore, affected tracts in patients were significantly diminished in volume (FTV_AFFECTED 4726 +/− 1660 mm³, FTV_CONTROL11200 +/− 3034 mm³, P<0.01) and FA (FA_AFFECTED 0.425 +/− 0.0501, FA_CONTROL 0.499 +/− 0.298, P<0.05) relative to controls. This decrease in tract mean TVA and FA was significant both above (P⩽0.001) and below (P<0.05) the infarct within the patient group. The differences in TVA and FA were the same above and below the infarct. A loose correlation between FA and NIH score was found (r = −0.73, P = 0.1).

Conclusions: Subcortical ischemic infarcts of the CST within 10 days of stroke onset significantly affect both mean CST volume and mean FA. Both parameters are significantly decreased within the stroke patients as well as relative to age matched controls independent of initial clinical symptom severity. The extent of FTV and FA decrease is the same along the entire CST.

711. A aserie of studies on the secondary degeneration in thalamus-cortex tracts after thalamic infarction with diffusion tensor imaging

Z. Liang¹, J. Zheng¹, X. Mo¹, C. Qin¹, Y. Wu¹, S. Jiang¹, Y. Dai², S. Li¹ and J. Zeng³

¹Department of Neurology; ²Department of Radiology, Guangxi Medical University, Nanning; ³Department of Neurology and Stroke Centre, Sun Yat-Sen University, Guangzhou, China

Background and aims: The secondary degeneration in pyramidal tracts after stroke have been demonstrated by animal experiments, atopsy examinations and diffusion tensor imaging. From our previous studies, a subcortical infarct may result in retrograde and anterograde degeneration of pyramidal tracts, moreover the degeneration of pyramidal tracts may hampered the neurological covery. However, the secondary degeneration in thalamus-cortex tracts (TCT) after thalamic infartion and its clinical significance still to be determined.

Methods: Eighteen patients with a recent thalamic infarct underwent three DTIs at first week (W1) and the fourth week (W4) as well as the twelfth week (W12) after onset. Eighteen age- and gender-matched controls underwent DTI one time. Mean diffusivity (MD) and fractional anisotropy (FA) were measured. Neurological deficit, motor deficit and life independence were assessed with NIH Stroke Scale, Fugl-Meyer scale and Barthel index respectively 2 h before each DTI examination. To reveal the secondary degeneration in TCT, Mean diffusivity (MD) and fractional anisotropy (FA) were measured at thalamus, the TCT at corona radiata and centrum semiovale levels.

Results: Fourteen of the patients had some degree of sensory disturbance. From patients' colour encoded images, a deficit signal was observed on thalamic primary lesion, and along the TCT at corona radiata and centrum semiovale levels (Figure 1). FA values of thalamic primary lesion, ipsilateral TCT on corona radiata and centrum semiovale levels decreased significantly from W1 to W12 progressively (P<0.01, respectively). The NIH Stroke Scale decreased, Fugl-Meyer scale and Barthel index increased significantly over the time (P<0.01, respectively). The absolute values of percent reduction of FA value in ipsilateral TCT correlated positively to the degree of sensory disturbance.

Conclusions: The progressive anterograde degeneration in TCT revealed by DTI. may be responsible for persistent sensory disturbance.

720. Prospective investigations on the secondary degeneration of corticospinal tracts in cervical spinal cord following cerebral infarct with diffusion tensor imaging

J. Zeng¹, Z. Liang², Z. Zhang³, F. Wang¹, L. Li¹ and Q. Hou¹

¹Department of Neurology and Stroke Centre, Sun Yat-Sen University, Guangzhou; ²Department of Neurology, Guangxi Medical University, Nanning; ³Department of Radiology, Sun Yat-Sen University, Guangzhou, China

Objective: To investigate the secondary degeneration of corticospinal tracts in cervical spinal cord following a recently cerebral infarct with diffusion tensor imaging (DTI), as well as its potential impact on neurological recovery.

Methods: 26 patients with a focal cerebral infract at unilateral posterior limb of internal capsule, corona radiata or at brainstem underwent DTI and evaluations with the NIH Stroke Scale (NIHSS), the Fugl-Meyer motor scale (FM) and the barthel index (BI) 3 times at the first week (W1), the fourth (W4) and twelfth week (W12) after stroke onset respectively. 26 gender and age match healthy volunteers underwent DTI three time at same time piots. The DTI parameters of Mean diffusivity (MD) and fractional anisotropy (FA value) were measured at cervical spinal cord and initial lesion (posterior limb of internal capsule, c corona radiata or at brainstem). The comparisons of DTI parameters between patient groud and controls were used t test, and the comprision of DTI parameters of patient group from the three time pionts were used competed data analysis of variance. Spearman correlation analysis was used to assess the association between the absolute value of percent change [((W12−W1)/W1) × 100%] of FA values and of clinical scores.

Results: Compared to the controls, the FA values of contralateral side corticospinal tracts in cervical spinal cord in patients significantly decreased at every time point repectively (P<0.01). In patients group, the FA values of of contralateral side corticospinal tracts in cervical spinal cord decreased progressively from W1 to W12 (P<0.01), but no significant modification of MD was observed. The absolute value of percent reduction of FA value of of contralateral side corticospinal tracts in cervical spinal cord in patients associated negatively with the absolute value of percent change of NIHSS and FM (P<0.05), but not with the absolute value of percent change of BI (P>0.05).

Table

The clinical scores of patients in the thr

	n	W1	W4	W12
NIHSS	26	13.26±3.03	9.66±2.45	5.25±2.25
Fugl-Meye	26	28.35±2.25	50.25±2.32	73.38±3.55
BI	26	40±6.43	58.18±3.26	75.25±4.75

Conclusions: The secondary degeneration of corticospinal tracts resulted from cerebral infarction may extend to cervical spinal cord. The secondary degeneration of corticospinal tracts in cervical spinal cord may exist and progress at lest three months, and hamper the process of neurological recovery.

863. Flow heterogeneity MRI reveals compensatory changes in the microcirculation during acute ischemic stroke

D. Liebeskind, J. Alger, B. Buck, T. Schaewe, O.Y. Bang, S. Starkman, D. Kim, L. Ali, B. Ovbiagele and J. Saver

University California Los Angeles, Los Angeles, California, USA

Objectives: Perfusion MRI may demonstrate changes in the flow distributions of the microcirculation as derangements of the normal flow heterogeneity (FH) within a voxel of the ischemic brain. Such FH abnormalities have been used to predict final infarct evolution. We used FH MRI to explore the specific microcirculatory derangements associated with collateral perfusion in acute MCA ischemia.

Methods: FH MRI measures were derived in 75 consecutive cases of acute ischemic stroke with complete occlusion of the MCA on conventional angiography. Separate maps were generated reflecting the width or standard deviation (FHD), skewness (FHS), and kurtosis (FHK) of flow distributions in the microcirculation. In addition standard perfusion MR images were obtained of CBF, CBV, MTT and CPP. Region of interest analyses were performed to compare values of FHD, FHS, and FHK in the ischemic territory with homologous regions of the contralateral hemisphere.

Collateral circulation evident on catheter angiography was graded with the ASITN/SIR scale.

Results: Pretreatment FH MRI revealed microcirculatory flow abnormalities in all 75 cases. FHD or the width of the flow distribution was increased in all cases (P<0.001). FHS images revealed redistribution of flow within slower routes of the microcirculation (P<0.001). FHK images demonstrated the predominance of particular flow routes in all cases (P<0.001). Topographic features of the FH abnormalities correlated with the degree of collateral flow noted at angiography.

Conclusions: Multiparametric FH MRI reveals consistent changes in the microcirculation associated with collateral perfusion in acute ischemic stroke due to MCA occlusion. Flow redistribution may involve recruitment of specific preferential routes with slower flow to enhance oxygen extraction. FH images can be rapidly generated and provide unique insights into tissue state that may help guide management.

976. Collateral perfusion and cerebral blood volume diminish infarct core and penumbra in acute ischemic stroke due to intracranial atherosclerosis

D. Liebeskind¹, J. Alger¹, G. Kim², N. Sanossian³, C. Chung², K. Lee² and O.Y. Bang²

¹University California Los Angeles, Los Angeles, California, USA; ²Samsung Medical Center, Seoul, South Korea; ³University of Southern California, Los Angeles, California, USA

Objectives: Etiology is rarely considered when evaluating perfusion abnormalities in acute ischemic stroke. Arterial occlusion due to progressive luminal stenosis with intracranial atherosclerosis (IA) may allow collateral perfusion to develop compared with abrupt cardioembolic (CE) occlusion. Diversion of cerebral blood volume (CBV) from core to periphery, leading to CBV collapse may also be offset by enhance collateral perfusion in the setting of IA. We analyzed diffusion and perfusion MRI in acute ischemic stroke based on etiology.

Methods: A consecutive series of acute stroke cases with MCA ischemia and NIHSS score>4 had MRI< 6 h from symptom onset. Post-processing generated Tmax>2secs (Tmax2), Tmax>8 secs (Tmax8), and CBV maps. Volumetric measures of lesions were extracted for Tmax and CBV (ischemic core (IC)<.50 ml/100g; penumbra to infarct (PI) = 0.51–0.75; penumbra to reverse (PR) = 0.76–1.00; normal (N) = 1.01–6.99; hyperemia (H) = 7.00–15.00). Lesion volumes were compared based on SSS TOAST classification (IA versus CE) and the subset with isolated M1 occlusion.

Results: Age and time to MRI did not differ between groups (20 IA and 46 CE). NIHSS scores were much lower with IA (mean 8.6 versus 13.8, P = 0.002) and diffusion volumes were 10-fold smaller with IA (P = 0.004). Tmax2 volumes were equivalent, yet Tmax8 volumes were far smaller in IA (P = 0.016) suggesting greater sensitivity to Tmax thresholds. CBV IC lesions were much smaller than CE cases (P = 0.02) and far more variability in IC lesions was noted amongst CE cases likely reflecting a broad range of collateral perfusion. PI and PR lesions trended smaller in IA with no difference in H volumes. The extent of abnormal tissue (IC, PI, PR, H) was far greater in CE reflecting larger gradients of CBV (P = 0.015). Similar patterns with significant discrepancies of CBV based on etiology were noted in the isolated M1 occlusions (12 IA, 12 CE). Isolated occlusion of M1 resulted in dramatically lower NIHSS scores (mean 10.4 versus 16.0, P = 0.005) and revealed strikingly smaller areas of ischemic core and penumbra. Gradients of CBV from core to penumbra and hyperemia downstream from isolated M1 occlusion were considerably smaller in IA (P = 0.039) likely reflecting well-developed arterial collateral inflow to balance cerebral venous diversion from core to periphery.

Conclusions: Ischemic stroke due to IA results in dramatically less sever neurological deficits and diminished lesion volumes. Time-domain perfusion parameters (e.g. Tmax) may not accurately reflect ischemic severity unless stroke etiology is considered. Collateral perfusion, reflected by preserved CBV, is more robust in IA with less variability across cases likely due to arteriogenesis. Smaller CBV gradients across the territory may make IA cases less vulnerable to collateral failure. Future predictive imaging models and treatment strategies must account for stroke etiologies such as IA.

1036. Brain tissue survival probability is a function of early MTT and subsequent evolution

H. An¹, A.L. Ford², K.D. Vo², A.M. Nassief², C.P. Derdeyn², W.J. Powers¹, J.-M. Lee² and W. Lin¹

¹University of North Carolina at Chapel Hill, Chapel Hill, North Carolina; ²Washington University in St Louis, St Louis, Missouri, USA

Objectives: Dynamic susceptibility contrast (DSC) perfusion method is widely used to study acute stroke patients, because it provides assessments of cerebral blood volume (CBV), mean transit time (MTT), and cerebral blood flow (CBF). Quantitative measures of CBV and CBF are difficult to obtain and their values differ between gray and white matter. On the other hand, MTT values in normal brain are quite uniform across the entire brain. Thus, MTT is commonly used to discern regions of ischemia. It has been suggested that MTT overestimates the size of the ischemic lesions, but it is unclear if this overestimation is caused by an inappropriate MTT threshold value. In this study, we aimed to examine brain tissue survival probability using the temporal evolution of MTT at 3 h and 6 h after symptom onset.

Methods: Eight acute ischemic stroke patients were studied at three sequential time points: 2.7±0.8 hours (tp1) and 6.2±0.2 h (tp2) and 1mo (tp3). MTT values were obtained at both tp1 and tp2. In addition, FLAIR images were acquired one month (tp3) to delineate the final infarct. Six patients received intravenous tPA, while the remaining two patients did not due to contraindications. Changes in MTT (ΔMTT = MTT(tp2) MTT(tp1)) were calculated, and negative values of ΔMTT were considered reperfusion from tp1 and tp2. Conditional probability density function of tissue survival (1-infarction) was computed at given MTT and ΔMTT values. To minimize the effects of variation induced by noise, small isolated volumes (<2cc) were excluded from this analysis.

Results: Mean MTT in the unaffected hemisphere was 6.2±0.1secs. The survival probability for tp1 MTT and tp2 MTT were very similar for MTT shorter than 15 secs. However, the survival probability was higher at tp1 than at tp2 for MTT values longer than 15 secs, and this deviation reached its maximum for values of 20 to 25 secs (Figure, Panel A). These data suggest that the survival rate is only time dependent for MTT values greater than a certain threshold (15 secs). Furthermore, very long MTT delays (as long as 25–30secs) still had substantial survival probabilities especially at tp1, suggesting these regions may still include both tissue at risk and core. Moreover, brain tissue survival probability was a function of both MTT change as well as initial MTT delay at tp1; shorter initial MTT delays and greater ΔMTT's improvement resulted in higher survival probability (Figure, panel B). Conversely, worsening MTT at tp2 (positive ΔMTT) decreased the chance of tissue survival.

Figure

(A) Survival probability at given MTT from tp1 and tp2 alone. (B) Survival probability as a function of tp1 MTT and DMTT. Different levels of survival probability are represented by colors. Blue color indicates that no region was detected within that particular MTT and DMTT range. The colorbar shows the tissue survival probability from 0 to 100%.

Conclusion: Our results suggest that survival of ischemic tissue depends on both initial MTT values and subsequent evolution during the hyperacute phase of stroke. The survival rate was only time dependent for MTT values greater than 15 secs.

1059. Evaluation of automated-partial-volume-correction software (PVEout) by means of un-biased random sampling-based volumetric assessment for two sets of clinical MRI data

A. Yamamoto^1,2, C. Svarer³, K. Ishida¹, C. Yokota⁴, Y. Tomii⁴, M. Yamauchi¹, K. Iihara⁵, S. Miyamoto⁵, K. Minematsu⁴ and H. Iida^1,2,3

¹Department of Investigative Radiology, National Cardiovascular Center Research Institute; ²Department Medical Physics and Engineering, Graduate School of Medicine, Osaka University, Osaka, Japan; ³Neurobiology Research Unit, Copenhagen University Hospital, Copenhagen, Denmark; ⁴Cerebrovascular Division, Department of Medicine; ⁵Neurosurgery Division, Department of Medicine, National Cardiovascular Center, Osaka, Japan

Introduction: Partial-volume-effect (PVE) is an essential source of errors, which degrades the accuracy in quantitative assessment of bio-physiologic functional parameters in PET and SPECT. A novel software, PVElab (running with SPM5), has been developed to automatically correct for PVE using clinical MRI images.^1,2 The program is based on automatic registration of MRI to PET/SPECT, followed by automatic segmentation (SPM5) ³ and correction of PET/SPECT counts for fractional volume as determined from the segmentation. This study was intended to evaluate the accuracy and inter-subject consistency of this automatic segmentation procedure in two typical sets of clinical MRI.

Materials and methods: One data set was obtained from a 1.5-T MRI scanner (Siemens) performed on 13 clinical patients with lacunar infarction in white matter areas. The acquisition sequence was a spin-echo, with TR 630 msec, and TE 14 msecs. Another was from a 3-T MRI scanner (GE) with 3D T₁ sequence on 8 healthy volunteers (1 male, and 7 female) whose age ranging from 20 to 35 year old. The acquisition sequence was a fast-spoiled-gradient-echo prepared with inversion-recovery (IR-FSPGR, TR/TE = 8.6/1.8 msecs, FA 8°, TI 600 msecs). Eight regions-of-interest (ROI) were delineated using the individual MR images for each subject, and grey-matter (GM) volume was obtained using PVElab. The un-biased random-sampling (RS)^4,5 was also applied to obtain the GM volume as a reference. A sophisticated software program was developed for this purpose, with which the operator define points that hits the structure of interest within the ROI defined for PVElab. GM volume values were compared for each ROIs, and intra-subject and inter-subject consistency of GM volumes were evaluated by PVElab and RS.

Results and discussion: PVElab (SPM5) and RS provided GM volumes which were significantly correlated to each other in each subject for both 1.5-T and 3-T data sets. The averaged difference for all subjects was 0.93%±3.18% for 1.5-T and 1.17%±6.35% for 3-T, in which the subject-dependent variation was attributed to the fact that the MRI intensity varies, dependent on subject. The reason for greater variation with 3-T MRI is probably due to the fact that 3-T MRI is more sensitive to error sources. The variation of differences for ROIs within each subject was in most of cases 1% to 2%. However, large variations were observed in two cases for 1.5-T and in one case for 3-T MRI. The Exact reason for this is unknown, but probably both MRI acquisition and segmentation algorithm are responsible to this.

Conclusion: The automatic segmentation tool in PVElab for PVE correction is feasible for clinical use. The inter-subject consistency and intra-subject variations are reasonably small, but further studies and careful evaluation are still needed.

1077. Cerebral venous flow physiology and impact of cerebral venous thrombosis: a phase-contrast MRI study

S. Stoquart-El Sankari^1,2, P. Lehmann³, A. Villette³, H. Deramond³, M.-E. Meyer^1,4 and O. Balédent^1,4

¹Department of Imaging and Biophysics; ²Department of Neurology; ³Department of Radiology; ⁴Department of Nuclear Medicine, Amiens University Hospital, Amiens Cedex, France

Objectives: Although it is supposed to play a crucial role in regulating intracranial hydrodynamics, a few studies have been devoted to the cerebral venous system physiology and pathology. Phase-Contrast MRI (PC-MRI) enables non invasive and rapid measurements of cerebrospinal fluid (CSF) and blood flows, and has been used for venous evaluation in jugular veins, and intracerebral pathways, but physiological quantitative flow parameters are still lacking.

Material: PC-MRI sequences were added to brain MRI conventional protocol in 19 patients with suspected Cerebral Venous Thrombosis (CVT), among whom 6 patients had a positive diagnosis on MR venography. Results were compared to those of 18 age-matched healthy volunteers (HV). In each HV, we calculated the venous pulsatility index (VPI) at both cervical and cerebral levels. The VPI was calculated as follows: VPI = ((Fmax-Fmin/Fmax)) × 100 where Fmax and Fmin represent respectively maximum and minimum venous flow amplitude.

Results: In HV, we found heterogeneous individual venous flows, as well as variable side dominance in paired veins and sinuses. PC-MRI enabled reproducible measurements of venous flows in Superior sagittal, straight and transverse sinuses. Arterial and CSF flows were normal. The VPI comparison showed a significant difference (P<0.05) between the internal jugular veins (VPI = 38±15), and the SSS (VPI = 21±10). In CVT patients, PCMRI did not detect any venous flow in the veins and/or sinuses with thrombosis whereas arterial flows were preserved. Interestingly, aqueductal and cervical CSF flows were hyperdynamic in one patient with signs of increased intracranial pressure, and decreased in 3 patients with extended thrombosis in the Superior Sagittal and Transverse sinuses (Figure).

Figure

CSF stroke volumes in patients with cerebral venous Thrombosis. Stroke volumes are represented in ml per cardiac cycle. The dashed rectangle corresponds to the average ± 1 standard deviation observed in the HV population.

Conclusion: We found heterogeneous types of venous pathways in HV, suggesting a variable role of accessory (epidural and vertebromedullar) drainage in physiological conditions. Besides, our study is the first to enable venous flow quantification in the major intracranial sinuses with excellent detection rates. VPI study showed higher pulsatility in the extracranial venous compartment. Heterogeneous results in patients with CVT show the complex role of veins in the regulation of intracranial hydrodynamics. PC-MRI enables reliable, safe and rapid measurements of CSF and blood flows. It helps understanding the complexity of the venous system and its role in regulating the intracranial pressure in physiological conditions and in pathologies like CVT.