Bayesian modeling of Dynamic Contrast Enhanced MRI data in cerebral glioma patients improves the diagnostic quality of hemodynamic parameter maps.

PURPOSE: The purpose of this work is to investigate if the curve-fitting algorithm in Dynamic Contrast Enhanced (DCE) MRI experiments influences the diagnostic quality of calculated parameter maps. MATERIAL AND METHODS: We compared the Levenberg-Marquardt (LM) and a Bayesian method (BM) in DCE data of 42 glioma patients, using two compartmental models (extended Toft's and 2-compartment-exchange model). Logistic regression and an ordinal linear mixed model were used to investigate if the image quality differed between the curve-fitting algorithms and to quantify if image quality was affected for different parameters and algorithms. The diagnostic performance to discriminate between high-grade and low-grade gliomas was compared by applying a Wilcoxon signed-rank test (statistical significance p>0.05). Two neuroradiologists assessed different qualitative imaging features. RESULTS: Parameter maps based on BM, particularly those describing the blood-brain barrier, were superior those based on LM. The image quality was found to be significantly improved (p<0.001) for BM when assessed through independent clinical scores. In addition, given a set of clinical scores, the generating algorithm could be predicted with high accuracy (area under the receiver operating characteristic curve between 0.91 and 1). Using linear mixed models, image quality was found to be improved when applying the 2-compartment-exchange model compared to the extended Toft's model, regardless of the underlying fitting algorithm. Tumor grades were only differentiated reliably on plasma volume maps when applying BM. The curve-fitting algorithm had, however, no influence on grading when using parameter maps describing the blood-brain barrier. CONCLUSION: The Bayesian method has the potential to increase the diagnostic reliability of Dynamic Contrast Enhanced parameter maps in brain tumors. In our data, images based on the 2-compartment-exchange model were superior to those based on the extended Toft's model.

MRI Assessment of Ischemic Lesion Evolution within White and Gray Matter.

BACKGROUND: In acute ischemic stroke (AIS), gray matter (GM) and white matter (WM) have different vulnerabilities to ischemia. Thus, we compared the evolution of ischemic lesions within WM and GM using MRI. METHODS: From a European multicenter prospective database (I-KNOW), available T1-weighted images were identified for 50 patients presenting with an anterior AIS and a perfusion weighted imaging (PWI)/diffusion weighted imaging (DWI) mismatch ratio of 1.2 or more. Six lesion compartments were outlined: initial DWI (b = 1,000 s/mm2) lesion, initial PWI-DWI mismatch (Tmax >4 s and DWI-negative), final infarct mapped on 1-month fluid-attenuated inversion recovery (FLAIR) imaging, lesion growth between acute DWI and 1-month FLAIR, DWI lesion reversal at 1 month and salvaged mismatch. The WM and GM were segmented on T1-weighted images, and all images were co-registered within subjects to the baseline MRI. WM and GM proportions were calculated for each compartment. RESULTS: Fifty patients were eligible for the study. Median delay between symptom onset and baseline MRI was 140 min. The percentage of WM was significantly greater in the following compartments: initial mismatch (52.5 vs. 47.5%, p = 0.003), final infarct (56.7 vs. 43.3%, p < 0.001) and lesion growth (58.9 vs. 41.2%, p < 0.001). No significant difference was found between GM and WM percentages within the initial DWI lesion, DWI reversal and salvaged mismatch compartments. CONCLUSIONS: Ischemic lesions may extend preferentially within the WM. Specific therapeutic strategies targeting WM ischemic processes may deserve further investigation.

Does b1000-b0 Mismatch Challenge Diffusion-Weighted Imaging-Fluid Attenuated Inversion Recovery Mismatch in Stroke?

BACKGROUND AND PURPOSE: Our aim was to explore whether the mismatch in lesion visibility between b1000 and b0 images is an alternative to mismatch between diffusion-weighted imaging and fluid-attenuated inversion recovery imaging as a surrogate marker of stroke age. METHODS: We analyzed patients from the European multicenter I-KNOW database. Independent readers assessed the visibility of ischemic lesions of the anterior circulation on b0 and fluid-attenuated inversion recovery imaging images. The signal-intensity ratio for b0 and fluid-attenuated inversion recovery imaging images was also measured from the segmented stroke lesion volume on b1000 images. RESULTS: This study included 112 patients (68 men; mean age, 67.4 years) with stroke onset within (n=85) or longer than (n=27) 4.5 hours. b1000-b0 mismatch identified patients within 4.5 hours of stroke onset with moderate sensitivity (72.9%; 95% confidence interval [CI], 63.5-82.4) and specificity (70.4%; 95% CI, 53.2-87.6), high positive predictive value (88.6%; 95% CI, 81.1-96.0), and low negative predictive value (45.2%; 95% CI, 30.2-60.3). Global comparison of b1000-b0 mismatch with diffusion-weighted imaging-fluid-attenuated inversion recovery imaging mismatch (considered the imaging gold standard) indicated high sensitivity (85.9%; 95% CI, 78.2-93.6), specificity (91.2%; 95% CI, 76.3-98.1), and positive predictive value (96.7%; 95% CI, 88.0-99.1) and moderate negative predictive value (73.8%; 95% CI, 60.5-87.1) of this new approach. b0 signal-intensity ratio (r=0.251; 95% CI, 0.069-0.417; P=0.008) was significantly although weakly correlated with delay between stroke onset and magnetic resonance imaging. CONCLUSIONS: b1000-b0 mismatch may identify patients with ischemic stroke of the within 4.5 hours of onset with high positive predictive value, perhaps constituting an alternative imaging tissue clock.

Sequential MR Assessment of the Susceptibility Vessel Sign and Arterial Occlusion in Acute Stroke.

PURPOSE: Susceptibility vessel sign (SVS) may likely influence recanalization after thrombolysis. We assessed, through the European sequential MRI database "I-KNOW," the relationship between the presence of SVS on T2-weighted gradient echo imaging, its angiographic counterpart on magnetic resonance angiography and its subsequent impact on recanalization after thrombolysis. MATERIALS AND METHODS: Initial clinical and MRI characteristics and early follow up were analyzed in acute ischemic stroke patients treated with rt-Pa within 4.5 hours. Patients underwent multimodal MRI at admission. Sequential imaging performed 3 hours, 2 days and 1 month later allowed the analysis of SVS changes and recanalization. RESULTS: Fifty patients were included in the study. SVS was observed in 54% of cases at admission. SVS was still present in 46% patients at 3 hours, 16% at 2 days, and 0% at 1 month. It was an independent predictor of no recanalization after thrombolysis (P = .04). After 3 hours, SVS disappeared in only 4 cases, and was not linked with recanalization on MRA. Conversely, when SVS persisted, a partial or complete recanalization was observed in 9 and 6 cases, respectively. CONCLUSIONS: SVS is a predictor of lower recanalization rate. Its disappearance is not necessarily correlated with recanalization.

Predicting infarction within the diffusion-weighted imaging lesion: does the mean transit time have added value?

BACKGROUND AND PURPOSE: There is ample evidence that in anterior circulation stroke, the diffusion-weighted imaging (DWI) lesion may escape infarction and thus is not a reliable infarct predictor. In this study, we assessed the predictive value of the mean transit time (MTT) for final infarction within the DWI lesion, first in patients scanned back-to-back with 15O-positron emission tomography and MR (DWI and perfusion-weighted imaging; "Cambridge sample") within 7 to 21 hours of clinical onset, then in a large sample of patients with anterior circulation stroke receiving DWI and perfusion-weighted imaging within 12 hours (85% within 6 hours; "I-KNOW sample"). METHODS: Both samples underwent structural MRI at approximately 1 month to map final infarcts. For both imaging modalities, MTT was calculated as cerebral blood volume/cerebral blood flow. After image coregistration and matrix resampling, the MTT values between voxels of interest that later infarcted or not were compared separately within and outside DWI lesions (DWI+ and DWI-, respectively) both within and across patients. In the I-KNOW sample, receiver operating characteristic curves were calculated for these voxel of interest populations and areas under the curve and optimal thresholds calculated. RESULTS: In the Cambridge data set (n=4), there was good concordance between predictive values of MTT (positron emission tomography) and MTT (perfusion-weighted imaging) for both DWI+ and DWI- voxels of interest indicating adequate reliability of MTT (perfusion-weighted imaging) for this purpose. In the I-KNOW data set (N=42), the MTT significantly added to the DWI lesion to predict infarction in both DWI- and DWI+ voxels of interest with areas under the curve approximately 0.78 and 0.64 (both P<0.001) and optimal thresholds approximately 8 seconds and 11 seconds, respectively. CONCLUSIONS: Despite the relatively small samples, this study suggests that adding MTT (perfusion-weighted imaging) may improve infarct prediction not only as already known outside, but also within, DWI lesions.