Toward patient-tailored perfusion thresholds for prediction of stroke outcome

Reducing False-Positives in CT Perfusion Infarct Core Segmentation Using Contralateral Local Normalization

BACKGROUND AND PURPOSE

The established global threshold of rCBF <30% for infarct core segmentation can lead to false-positives, as it does not account for the differences in blood flow between GM and WM and patient-individual factors, such as microangiopathy. To mitigate this problem, we suggest normalizing each voxel not only with a global reference value (ie, the median value of normally perfused tissue) but also with its local contralateral counterpart.

MATERIALS AND METHODS

We retrospectively enrolled 2830 CTP scans with suspected ischemic stroke, of which 335 showed obvious signs of microangiopathy. In addition to the conventional, global normalization, a local normalization was performed by dividing the rCBF maps with their mirrored and smoothed counterpart, which sets each voxel value in relation to the contralateral counterpart, intrinsically accounting for GM and WM differences and symmetric patient individual microangiopathy. Maps were visually assessed and core volumes were calculated for both methods.

RESULTS

Cases with obvious microangiopathy showed a strong reduction in false-positives by using local normalization (mean 14.7 mL versus mean 3.7 mL in cases with and without microangiopathy). On average, core volumes were slightly smaller, indicating an improved segmentation that was more robust against naturally low blood flow values in the deep WM.

CONCLUSIONS

The proposed method of local normalization can reduce overestimation of the infarct core, especially in the deep WM and in cases with obvious microangiopathy. False-positives in CTP infarct core segmentation might lead to less-than-optimal therapy decisions when not correctly interpreted. The proposed method might help mitigate this problem.

Assessing the diagnostic accuracy of CT perfusion: a systematic review

Background and purpose

Computed tomography perfusion (CTP) has successfully extended the time window for reperfusion therapies in ischemic stroke. However, the published perfusion parameters and thresholds vary between studies. Using Preferred Reporting Items for Systematic Reviews and Meta-Analyses of Diagnostic Test Accuracy Studies (PRISMA-DTA) guidelines, we conducted a systematic review to investigate the accuracy of parameters and thresholds for identifying core and penumbra in adult stroke patients.

Methods

We searched Medline, Embase, the Cochrane Library, and reference lists of manuscripts up to April 2022 using the following terms "computed tomography perfusion," "stroke," "infarct," and "penumbra." Studies were included if they reported perfusion thresholds and undertook co-registration of CTP to reference standards. The quality of studies was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool and Standards for Reporting of Diagnostic Accuracy (STARD) guidelines.

Results

A total of 24 studies were included. A meta-analysis could not be performed due to insufficient data and significant heterogeneity in the study design. When reported, the mean age was 70.2 years (SD+/-3.69), and the median NIHSS on admission was 15 (IQR 13-17). The perfusion parameter identified for the core was relative cerebral blood flow (rCBF), with a median threshold of <30% (IQR 30, 40%). However, later studies reported lower thresholds in the early time window with rapid reperfusion (median 25%, IQR 20, 30%). A total of 15 studies defined a single threshold for all brain regions irrespective of collaterals and the gray and white matter.

Conclusion

A single threshold and parameter may not always accurately differentiate penumbra from core and oligemia. Further refinement of parameters is needed in the current era of reperfusion therapy.

Predicting the tissue outcome of acute ischemic stroke from acute 4D computed tomography perfusion imaging using temporal features and deep learning

Predicting follow-up lesions from baseline CT perfusion (CTP) datasets in acute ischemic stroke patients is important for clinical decision making. Deep convolutional networks (DCNs) are assumed to be the current state-of-the-art for this task. However, many DCN classifiers have not been validated against the methods currently used in research (random decision forests, RDF) and clinical routine (Tmax thresholding). Specialized DCNs have even been designed to extract complex temporal features directly from spatiotemporal CTP data instead of using standard perfusion parameter maps. However, the benefits of applying deep learning to source or deconvolved CTP data compared to perfusion parameter maps have not been formally investigated so far. In this work, a modular UNet-based DCN is proposed that separates temporal feature extraction from tissue outcome prediction, allowing for both model validation using perfusion parameter maps as well as end-to-end learning from spatiotemporal CTP data. 145 retrospective datasets comprising baseline CTP imaging, perfusion parameter maps, and follow-up non-contrast CT with manual lesion segmentations were assembled from acute ischemic stroke patients treated with intravenous thrombolysis alone (IV; n = 43) or intra-arterial mechanical thrombectomy (IA; n = 102) with or without combined IV. Using the perfusion parameter maps as input, the proposed DCN (mean Dice: 0.287) outperformed the RDF (0.262) and simple Tmax-thresholding (0.249). The performance of the proposed DCN was approximately equal using features optimized from the deconvolved residual curves (0.286) compared to perfusion parameter maps (0.287), while using features optimized from the source concentration-time curves (0.296) provided the best tissue outcome predictions.

Controversies in Imaging of Patients with Acute Ischemic Stroke: AJR Expert Panel Narrative Review

The development of reperfusion therapies has profoundly impacted stroke care, initially with the advent of IV thrombolytic (IVT) treatment and, more recently, with the development and refinement of endovascular treatment (EVT). Progress in neuroimaging has supported the paradigm shift of stroke care, and advanced neuroimaging now has a fundamental role in triaging patients for both IVT and EVT. As the standard of care for acute ischemic stroke (AIS) evolves, controversies remain in certain clinical scenarios. This article explores the use of multimodality imaging for treatment selection of AIS in the context of recent guidelines, highlighting controversial topics and providing guidance for clinical practice. Results of major randomized trials supporting EVT are reviewed. Advantages and disadvantages of CT, CTA, MRI, and MRA in stroke diagnosis are summarized, with attention to level 1 evidence supporting the role of vascular imaging and perfusion imaging. Patient selection is compared between approaches based on time thresholds and physiologic approaches based on infarct core measurement using imaging. Moreover, various imaging approaches to core measurement are described. As ongoing studies push treatment boundaries, advanced imaging is expected to help identify a widening range of patients who may benefit from therapy.

JOURNAL CLUB: Evaluation of Diffusion Kurtosis Imaging of Stroke Lesion With Hemodynamic and Metabolic MRI in a Rodent Model of Acute Stroke

OBJECTIVE

Diffusion kurtosis imaging (DKI) has emerged as a new acute stroke imaging approach, augmenting routine DWI. Although it has been shown that a diffusion lesion without kurtosis abnormality is more likely to recover after reperfusion, whereas a kurtosis lesion shows poor response, little is known about the underlying pathophysiologic profile of the kurtosis lesion versus the kurtosis lesion-diffusion lesion mismatch.

MATERIALS AND METHODS

We performed multiparametric MRI, including arterial spin labeling, pH-sensitive amide proton transfer, and DKI, in a rodent model of acute stroke caused by embolic middle cerebral artery occlusion. Diffusion and kurtosis lesions were semiautomatically segmented, and multiparametric MRI indexes were compared among the kurtosis lesion, diffusion lesion, kurtosis lesion-diffusion lesion mismatch, and the contralateral normal tissue area.

RESULTS

We confirmed a significant difference between diffusion lesion and kurtosis lesion volumes (mean [± SD] volume, 151 ± 65 vs 125 ± 47 mm³; p < 0.05). Although ischemic lesions have significantly reduced cerebral blood flow compared with contralateral normal tissue, we did not find a significant difference in cerebral blood flow between the kurtosis lesion and the kurtosis lesion-diffusion lesion mismatch (mean cerebral blood flow, 0.53 ± 0.10 vs 0.47 ± 0.14 mL/g of tissue per minute; p > 0.05). Of importance, the pH of the kurtosis lesion was significantly lower than that of the lesion mismatch (mean pH, 6.81 ± 0.08 vs 6.89 ± 0.09; p < 0.01).

CONCLUSION

The present study confirms that DKI provides an expedient approach for refining the heterogeneous DWI lesion that is associated with graded metabolic derangement, which is promising for improving the infarction core definition and ultimately helping to guide stroke treatment.

Absolute Cerebral Blood Flow Infarction Threshold for 3-Hour Ischemia Time Determined with CT Perfusion and 18F-FFMZ-PET Imaging in a Porcine Model of Cerebral Ischemia

CT Perfusion (CTP) derived cerebral blood flow (CBF) thresholds have been proposed as the optimal parameter for distinguishing the infarct core prior to reperfusion. Previous threshold-derivation studies have been limited by uncertainties introduced by infarct expansion between the acute phase of stroke and follow-up imaging, or DWI lesion reversibility. In this study a model is proposed for determining infarction CBF thresholds at 3hr ischemia time by comparing contemporaneously acquired CTP derived CBF maps to ¹⁸F-FFMZ-PET imaging, with the objective of deriving a CBF threshold for infarction after 3 hours of ischemia. Endothelin-1 (ET-1) was injected into the brain of Duroc-Cross pigs (n = 11) through a burr hole in the skull. CTP images were acquired 10 and 30 minutes post ET-1 injection and then every 30 minutes for 150 minutes. 370 MBq of ¹⁸F-FFMZ was injected ~120 minutes post ET-1 injection and PET images were acquired for 25 minutes starting ~155–180 minutes post ET-1 injection. CBF maps from each CTP acquisition were co-registered and converted into a median CBF map. The median CBF map was co-registered to blood volume maps for vessel exclusion, an average CT image for grey/white matter segmentation, and ¹⁸F-FFMZ-PET images for infarct delineation. Logistic regression and ROC analysis were performed on infarcted and non-infarcted pixel CBF values for each animal that developed infarct. Six of the eleven animals developed infarction. The mean CBF value corresponding to the optimal operating point of the ROC curves for the 6 animals was 12.6 ± 2.8 mL·min^-1·100g^-1 for infarction after 3 hours of ischemia. The porcine ET-1 model of cerebral ischemia is easier to implement then other large animal models of stroke, and performs similarly as long as CBF is monitored using CTP to prevent reperfusion.

CT Perfusion in Acute Stroke: "Black Holes" on Time-to-Peak Image Maps Indicate Unsalvageable Brain

BACKGROUND AND PURPOSE

CT perfusion is becoming important in acute stroke imaging to determine optimal patient-management strategies. The purpose of this study was to examine the predictive value of time-to-peak image maps and, specifically, a phenomenon coined a "black hole" for assessing infarcted brain tissue at the time of scan.

METHODS

Acute stroke patients were screened for the presence of black holes and their follow-up imaging (noncontrast CT or MR) was reviewed to assess for infarcted brain tissue.

RESULTS

Of the 23 patients with signs of acute ischemia on CT perfusion, all had black holes. The black holes corresponded with areas of infarcted brain on follow-up imaging (specificity 100%). Black holes demonstrated significantly lower cerebral blood volumes (P < .001) and cerebral blood flow (P < .001) compared to immediately adjacent tissue.

CONCLUSIONS

Black holes on time-to-peak image maps represent areas of unsalvageable brain.

Acute Stroke Imaging Research Roadmap III Imaging Selection and Outcomes in Acute Stroke Reperfusion Clinical Trials: Consensus Recommendations and Further Research Priorities

BACKGROUND AND PURPOSE

The Stroke Imaging Research (STIR) group, the Imaging Working Group of StrokeNet, the American Society of Neuroradiology, and the Foundation of the American Society of Neuroradiology sponsored an imaging session and workshop during the Stroke Treatment Academy Industry Roundtable (STAIR) IX on October 5 to 6, 2015 in Washington, DC. The purpose of this roadmap was to focus on the role of imaging in future research and clinical trials.

METHODS

This forum brought together stroke neurologists, neuroradiologists, neuroimaging research scientists, members of the National Institute of Neurological Disorders and Stroke (NINDS), industry representatives, and members of the US Food and Drug Administration to discuss STIR priorities in the light of an unprecedented series of positive acute stroke endovascular therapy clinical trials.

RESULTS

The imaging session summarized and compared the imaging components of the recent positive endovascular trials and proposed opportunities for pooled analyses. The imaging workshop developed consensus recommendations for optimal imaging methods for the acquisition and analysis of core, mismatch, and collaterals across multiple modalities, and also a standardized approach for measuring the final infarct volume in prospective clinical trials.

CONCLUSIONS

Recent positive acute stroke endovascular clinical trials have demonstrated the added value of neurovascular imaging. The optimal imaging profile for endovascular treatment includes large vessel occlusion, smaller core, good collaterals, and large penumbra. However, equivalent definitions for the imaging profile parameters across modalities are needed, and a standardization effort is warranted, potentially leveraging the pooled data resulting from the recent positive endovascular trials.

Effect of Collaterals on Clinical Presentation, Baseline Imaging, Complications, and Outcome in Acute Stroke

BACKGROUND AND PURPOSE

Good CTA collaterals independently predict good outcome in acute ischemic stroke. Our aim was to evaluate the role of collateral circulation and its added benefit over CTP-derived total ischemic volume as a predictor of baseline NIHSS score, total ischemic volume, hemorrhagic transformation, final infarct size, and a modified Rankin Scale score >2.

MATERIALS AND METHODS

This was a retrospective study of 395 patients with stroke dichotomized by recanalization (recanalization positive/recanalization negative) and collateral status. Clot burden score was quantified on baseline CTA. Total ischemic volumes were derived from thresholded CTP maps. Final infarct size was assessed on follow-up CT/MRI. We performed uni-/multivariate analyses for each outcome, adjusting for rtPA status, using general linear (continuous variables) and logistic (binary variables) regression. Model comparison with collateral score and total ischemic volume was performed using the F or likelihood ratio test.

RESULTS

Collateral presence independently and inversely predicted all outcomes except hemorrhagic transformation in patients who were recanalization negative and mRS >2 in patients who were recanalization positive. The greatest collateral benefit occurred in patients who were recanalization negative, contributing 16.5% and 19.2% of the variability for final infarct size and mRS >2. The collateral score model is superior to the total ischemic volume for mRS >2 prediction, but a combination of total ischemic volume and collateral score is superior for mRS >2 and final infarct prediction (24% and 28% variability, respectively). In patients who were recanalization positive, a model including collateral score and total ischemic volume was superior to that of total ischemic volume for hemorrhagic transformation and final infarct prediction but was muted compared with patients who were recanalization negative (11.3% and 16.9% variability).

CONCLUSIONS

Collateral circulation is an independent predictor of all outcomes, but the magnitude of significance varies, greater in patients who were recanalization negative versus recanalization positive. Total ischemic volume assessment is complementary to collateral score in many cases.

Contribution and additional impact of imaging to the SPAN-100 score

BACKGROUND AND PURPOSE

Stroke Prognostication by Using Age and NIHSS score (SPAN-100 index) facilitates stroke outcomes. We assessed imaging markers associated with the SPAN-100 index and their additional impact on outcome determination.

MATERIALS AND METHODS

Of 273 consecutive patients with acute ischemic stroke (<4.5 hours), 55 were characterized as SPAN-100-positive (age +NIHSS score ≥ 100). A comprehensive imaging review evaluated differences, using the presence of the hyperattenuated vessel sign, ASPECTS, clot burden score, collateral score, CBV, CBF, and MTT. The primary outcome assessed was favorable outcome (mRS ≤ 2). Secondary outcomes included recanalization, lack of neurologic improvement, and hemorrhagic transformation. Uni- and multivariate analyses assessed factors associated with favorable outcome. Area under the curve evaluated predictors of favorable clinical outcome.

RESULTS

Compared with the SPAN-100-negative group, the SPAN-100-positive group (55/273; 20%) demonstrated larger CBVs (<0.001), poorer collaterals (P < .001), and increased hemorrhagic transformation rates (56.0% versus 36%, P = .02) despite earlier time to rtPA (P = .03). Favorable outcome was less common among patients with SPAN-100-positive compared with SPAN-100-negative (10.9% versus 42.2%; P < .001). Multivariate regression revealed poorer outcome for SPAN-100-positive (OR = 0.17; 95% CI, 0.06-0.38; P = .001), clot burden score (OR = 1.14; 95% CI, 1.05-1.25; P < .001), and CBV (OR = 0.58; 95% CI, 0.46-0.72; P = .001). The addition of the clot burden score and CBV improved the predictive value of SPAN-100 alone for favorable outcome from 60% to 68% and 74%, respectively.

CONCLUSIONS

SPAN-100-positivity predicts a lower likelihood of favorable outcome and increased hemorrhagic transformation. CBV and clot burden score contribute to poorer outcomes among high-risk patients and improve stroke-outcome prediction.

Final infarct volume estimation on 1-week follow-up MR imaging is feasible and is dependent on recanalization status

PURPOSE

We aim to characterize infarct volume evolution within the first month post-ischemic stroke and to determine the effect of recanalization status on early infarct volume estimation.

METHODS

Ischemic stroke patients recruited for the MONITOR and VISION studies were retrospectively screened and patients who had infarcts on diffusion-weighted imaging (DWI) at baseline and had at least two follow-up MR scans (n = 56) were included. Pre-defined target imaging time points, obtained on a 3-T MR scanner, were 12 hours (h), 24 h, 7 days, and ≥30 days post-stroke. Infarct tissue was manually traced blinded to the images at the other time points. Infarct expansion index was calculated by dividing infarct volume at each follow-up time point by the baseline DWI infarct volume. Recanalization was assessed within 24 h post-stroke. Correlation and statistical comparison analysis were done using the Spearman, Mann-Whitney, and Kruskal-Wallis tests.

RESULTS

Follow-up infarct volumes were positively correlated with the baseline infarct volume (ρ > 0.81; p < 0.001) where the strongest correlation existed between baseline and 7-day post-stroke infarct volumes (ρ = 0.92; p < 0.001). The strongest correlation among the follow-up imaging was found between infarct volumes 7-day post-stroke and ≥30-day time points (ρ = 0.93; p < 0.001). Linear regression showed a close-to unity slope between 7-day and final infarct volumes (slope = 1.043; p < 0.001). Infarct expansion was higher in the non-recanalized group than the recanalized group at the 7-day (p = 0.001) and ≥30-day (p = 0.038) time points.

CONCLUSIONS

Final infarct volume can be approximated as early as 7 days post-stroke. Final infarct volume approximation is significantly associated with recanalization status.