Assessment of Small Vessel Function Using 7T MRI in Patients With Sporadic Cerebral Small Vessel Disease: The ZOOM@SVDs Study.

BACKGROUND AND OBJECTIVES: Cerebral small vessel disease (cSVD) is a major cause of stroke and dementia, but little is known about disease mechanisms at the level of the small vessels. 7T-MRI allows assessing small vessel function in vivo in different vessel populations. We hypothesized that multiple aspects of small vessel function are altered in patients with cSVD and that these abnormalities relate to disease burden. METHODS: Patients and controls participated in a prospective observational cohort study, the ZOOM@SVDs study. Small vessel function measures on 7T-MRI included perforating artery blood flow velocity and pulsatility index in the basal ganglia and centrum semiovale, vascular reactivity to visual stimulation in the occipital cortex, and reactivity to hypercapnia in the gray and white matter. Lesion load on 3T-MRI and cognitive function were used to assess disease burden. RESULTS: Forty-six patients with sporadic cSVD (mean age ± SD 65 ± 9 years) and 22 matched controls (64 ± 7 years) participated in the ZOOM@SVDs study. Compared with controls, patients had increased pulsatility index (mean difference 0.09, p = 0.01) but similar blood flow velocity in basal ganglia perforating arteries and similar flow velocity and pulsatility index in centrum semiovale perforating arteries. The duration of the vascular response to brief visual stimulation in the occipital cortex was shorter in patients than in controls (mean difference -0.63 seconds, p = 0.02), whereas reactivity to hypercapnia was not significantly affected in the gray and total white matter. Among patients, reactivity to hypercapnia was lower in white matter hyperintensities compared with normal-appearing white matter (blood-oxygen-level dependent mean difference 0.35%, p = 0.001). Blood flow velocity and pulsatility index in basal ganglia perforating arteries and reactivity to brief visual stimulation correlated with disease burden. DISCUSSION: We observed abnormalities in several aspects of small vessel function in patients with cSVD indicative of regionally increased arteriolar stiffness and decreased reactivity. Worse small vessel function also correlated with increased disease burden. These functional measures provide new mechanistic markers of sporadic cSVD.

Reliability of velocity pulsatility in small vessels on 3Tesla MRI in the basal ganglia: a test-retest study.

OBJECTIVE: Recent work showed the feasibility of measuring velocity pulsatility in the perforating arteries at the level of the BG using 3T MRI. However, test-retest measurements have not been performed, yet. This study assessed the test-retest reliability of 3T MRI blood flow velocity measurements in perforating arteries in the BG. MATERIALS AND METHODS: Two-dimensional phase-contrast cardiac gated (2D-PC) images were acquired for 35 healthy controls and repeated with and without repositioning. 2D-PC images were processed and analyzed, to assess the number of detected perforating arteries (Ndetected), mean blood flow velocity (Vmean), and velocity pulsatility index (vPI). Paired t-tests and Bland-Altman plots were used to compare variance in outcome parameters with and without repositioning, and limits of agreement (LoA) were calculated. RESULTS: The LoA was smallest for Vmean (35%) and highest for vPI (79%). Test-retest reliability was similar with and without repositioning of the subject. DISCUSSION: We found similar LoA with and without repositioning indicating that the measurement uncertainty is dominated by scanner and physiological noise, rather than by planning. This enables to study hemodynamic parameters in perforating arteries at clinically available scanners, provided sufficiently large sample sizes are used to mitigate the contribution of scanner- and physiological noise.

Outcome Prediction of Postanoxic Coma: A Comparison of Automated Electroencephalography Analysis Methods.

BACKGROUND: To compare three computer-assisted quantitative electroencephalography (EEG) prediction models for the outcome prediction of comatose patients after cardiac arrest regarding predictive performance and robustness to artifacts. METHODS: A total of 871 continuous EEGs recorded up to 3 days after cardiac arrest in intensive care units of five teaching hospitals in the Netherlands were retrospectively analyzed. Outcome at 6 months was dichotomized as "good" (Cerebral Performance Category 1-2) or "poor" (Cerebral Performance Category 3-5). Three prediction models were implemented: a logistic regression model using two quantitative features, a random forest model with nine features, and a deep learning model based on a convolutional neural network. Data from two centers were used for training and fivefold cross-validation (n = 663), and data from three other centers were used for external validation (n = 208). Model output was the probability of good outcome. Predictive performances were evaluated by using receiver operating characteristic analysis and the calculation of predictive values. Robustness to artifacts was evaluated by using an artifact rejection algorithm, manually added noise, and randomly flattened channels in the EEG. RESULTS: The deep learning network showed the best overall predictive performance. On the external test set, poor outcome could be predicted by the deep learning network at 24 h with a sensitivity of 54% (95% confidence interval [CI] 44-64%) at a false positive rate (FPR) of 0% (95% CI 0-2%), significantly higher than the logistic regression (sensitivity 33%, FPR 0%) and random forest models (sensitivity 13%, FPR, 0%) (p < 0.05). Good outcome at 12 h could be predicted by the deep learning network with a sensitivity of 78% (95% CI 52-100%) at a FPR of 12% (95% CI 0-24%) and by the logistic regression model with a sensitivity of 83% (95% CI 83-83%) at a FPR of 3% (95% CI 3-3%), both significantly higher than the random forest model (sensitivity 1%, FPR 0%) (p < 0.05). The results of the deep learning network were the least affected by the presence of artifacts, added white noise, and flat EEG channels. CONCLUSIONS: A deep learning model outperformed logistic regression and random forest models for reliable, robust, EEG-based outcome prediction of comatose patients after cardiac arrest.