DWI-ASPECTS (Diffusion-Weighted Imaging-Alberta Stroke Program Early Computed Tomography Scores) and DWI-FLAIR (Diffusion-Weighted Imaging-Fluid Attenuated Inversion Recovery) Mismatch in Thrombectomy Candidates: An Intrarater and Interrater Agreement Study

Voxel level dense prediction of acute stroke territory in DWI using deep learning segmentation models and image enhancement strategies

PURPOSE

To build a stroke territory classifier model in DWI by designing the problem as a multiclass segmentation task by defining each stroke territory as distinct segmentation targets and leveraging the guidance of voxel wise dense predictions.

MATERIALS AND METHODS

Retrospective analysis of DWI images of 218 consecutive acute anterior or posterior ischemic stroke patients examined between January 2017 to April 2020 in a single center was carried out. Each stroke area was defined as distinct segmentation target with different class labels. U-Net based network was trained followed by majority voting of the voxel wise predictions of the model to transform them into patient level stroke territory classes. Effects of bias field correction and registration to a common space were explored.

RESULTS

Of the 218 patients included in this study, 141 (65%) were anterior stroke, and 77 were posterior stroke (35%) whereas 117 (53%) were male and 101 (47%) were female. The model built with original images reached 0.77 accuracy, while the model built with N4 bias corrected images reached 0.80 and the model built with images which were N4 bias corrected and then registered into a common space reached 0.83 accuracy values.

CONCLUSION

Voxel wise dense prediction coupled with bias field correction to eliminate artificial signal increase and registration to a common space help models for better performance than using original images. Knowing the properties of target domain while designing deep learning models is important for the overall success of these models.

Predictors of Good Functional Outcome in Patients with Tandem Occlusion After Revascularization Treatment: Single-Center Experience with 12-Month Follow-Up

OBJECTIVE

The optimal management and procedural strategy for tandem occlusion (TO) in acute ischemic stroke are still unclear, as is the long-term outcome of these patients. The aim of this study was to evaluate predictors of good functional outcome in patients with TO through the analysis of demographics, clinical, and radiological data with a 1-year follow-up.

METHODS

We collected data on 100 patients with TO who underwent revascularization treatments in our comprehensive stroke center. We divided patients into 2 groups: those with good functional outcome, defined as a modified Rankin Scale 0-2, and those with poor functional outcome, defined as a modified Rankin Scale 3-6 at 3, 6, and 12-month follow-up. Moreover, we investigated which variables were able to influence mortality.

RESULTS

At multivariate analysis, endovascular treatment with mechanical thrombectomy combined with emergent cervical carotid artery stenting was an independent predictor of good functional outcome at 6 and 12 months (adjusted odds ratio [aOR] 4.3, confidence interval [CI] 1.49-12.31, P = 0.007) (aOR 3.5, CI 1.25-9.61, P = 0.017) and was associated with a lower rate of mortality at 3 and 6 months follow-up (aOR 0.14, CI 0.04-0.57, P = 0.006 and aOR 0.296, CI 0.97-0.902, P = 0.032, respectively). Furthermore, smoking habit was associated with a better outcome at 3-month follow-up (aOR 10.7, CI 2.2-51.6, P = 0.003) but not at 6- and 12-month.

CONCLUSIONS

Our research, conducted in a small sample size of patients with acute ischemic stroke due to TO of anterior circulation, found that acute stent placement seems to be safe, improving clinical outcome, and it is associated with low rate of mortality at long-term follow-up.

Does Ischemic Core Volume Modify the Treatment Effect of Endovascular Thrombectomy?

Imaging-based selection has become integral in guiding endovascular thrombectomy for large vessel occlusive stroke, driven by positive trial outcomes incorporating parenchymal and perfusion imaging criteria. While small-moderate core trials transformed acute reperfusion therapy, uncertainties persisted for large ischemic strokes. This was recently addressed in several treatment trials which demonstrated a benefit of endovascular thrombectomy in patients with large ischemic strokes, defined by parenchymal and/or perfusion imaging. Although individual trials suggest treatment benefits regardless of core size, patient-level meta-analyses are essential to clarify this relationship. Our aim was to summarize the imaging ramifications of the major endovascular thrombectomy trials of the past decade focusing on the interaction between the core and the treatment effect, to assist in the design of future meta-analyses. The core-treatment relationship that will be investigated in these meta-anlalyses will likely have major implications in our systems of care designs and in determining the utility of imaging-based selection.

Association of early seizures after ischemic stroke with diffusion-weighted imaging-alberta stroke program early CT score (DWI-ASPECTS) and neutrophil-to-lymphocyte ratio

BACKGROUND

Post-stroke seizure (PSS) is a common considerable complication of acute ischemic stroke (AIS). Early risk assessment can clinical practitioners to plan effective prevention and management. We aimed to determine whether assessing Diffusion-Weighted Imaging-Alberta Stroke Program Early CT Score (DWI-ASPECTS), and neutrophil indices allows for identifying patients at risk of PSS.

METHODS

This prospective study included AIS patients with cortical involvement admitted to a single academic center between January 2020 to October 2023. For all included subjects, DWI-Brain MRI, blood neutrophils, and platelet counts were obtained and the DWI-ASPECTS score was calculated. Then, the patients were followed up for 6 months in terms of PSS occurrence. Based on the occurrence of PSS, patients were divided into two groups of PSS and non-PSS. For analysis, imaging and laboratory data were compared between two groups. Logistic regression was applied to determine the relationship between DWI-ASPECTS and neutrophil indices, with early PSS. Finally, the sensitivity and specificity of these variables for PSS were estimated.

RESULTS

A total of 309 were included in the final statistical analysis. DWI-ASPECT and neutrophil-to-lymphocyte ratio (NLR) were significantly associated with early PSS with OR of 0.74 and OR of 1.13, respectively (P < 0.05). Further analysis showed that, a combination of DWI-ASPECTS, NLR had an area under the curve (AUC) of 0.72 for predicting the occurrence of early PSS.

CONCLUSION

DWI-ASPECTS and NLR are associated with the occurrence of early PSS after cortical ischemic stroke. A combination of these predictors had higher sensitivity and specificity for PSS rather than each factor alone. These findings may be helpful for determining the risk of PSS if validated in future studies.

MRI of cerebral oedema in ischaemic stroke and its current use in routine clinical practice

Currently, with the knowledge of the role of collateral circulation in the development of cerebral ischaemia, traditional therapeutic windows are being prolonged, with time not being the only criterion. Instead, a more personalised approach is applied to select additional patients who might benefit from active treatment. This review briefly describes the current knowledge of the pathophysiology of the development of early ischaemic changes, the capabilities of MRI to depict such changes, and the basics of the routinely used imaging techniques broadly available for the assessment of individual phases of cerebral ischaemia, and summarises the possible clinical use of routine MR imaging, including patient selection for active treatment and assessment of the outcome on the basis of imaging.

Prognostic value of the baseline magnetic resonance score in patients with acute posterior circulation ischaemic stroke after mechanical thrombectomy

AIM

To investigate the prognostic value of the composite posterior circulation Acute Stroke Prognosis Early Computed tomography (CT) Score (ASPECTS)-Collaterals (pcASCO) score, which combines diffusion-weighted imaging (DWI) posterior circulation ASPECTS (pcASPECTS) and the magnetic resonance angiography (MRA)-collateral circulation score at baseline among patients with acute posterior circulation ischaemic stroke after mechanical thrombectomy.

MATERIALS AND METHODS

Patients with acute posterior circulation ischaemic stroke who underwent mechanical thrombectomy were analysed retrospectively. The DWI-pcASPECTS and MRA-collateral circulation score before treatment and the modified Rankin Scale (mRS) at 90 days after treatment were used as the endpoints. An mRS ≤2 was defined as a good prognosis, and an mRS ≥3 was defined as a poor prognosis. Multivariate logistic regression was used to analyse independent predictors of functional outcome 90 days after mechanical thrombectomy.

RESULTS

Mechanical thrombectomy was performed in 57 patients; 38 patients had a good prognosis, 19 patients had a poor prognosis, and 33 patients were successfully recanalised. Univariate logistic regression found that National Institute of Health Stroke Scale (NIHSS) score (OR: 1.18, p<0.001), pcASPECTS (OR: 1.91, p=0.028) and pcASCO score (OR: 0.51, p=0.001) were factors of good functional outcome. Receiver operating characteristic curve (ROC curve) analysis showed that the diagnostic efficiency of the NIHSS and pcASCO was better (AUC = 0.88, 0.83, p<0.05) than that of the pcASPECTS (AUC = 0.65). The prediction model was established by age, NIHSS, and pcASCO, and the diagnostic efficiency of the prediction model was better (AUC = 0.94).

CONCLUSIONS

The composite MR-pcASCO score can be used as an important predictor of the prognosis of patients with acute posterior circulation ischaemic stroke after mechanical thrombectomy.

A deep learning analysis of stroke onset time prediction and comparison to DWI-FLAIR mismatch

INTRODUCTION

When time since stroke onset is unknown, DWI-FLAIR mismatch rating is an established technique for patient stratification. A visible DWI lesion without corresponding parenchymal hyperintensity on FLAIR suggests time since onset of under 4.5 h and thus a potential benefit from intravenous thrombolysis. To improve accuracy and availability of the mismatch concept, deep learning might be able to augment human rating and support decision-making in these cases.

METHODS

We used unprocessed DWI and coregistered FLAIR imaging data to train a deep learning model to predict dichotomized time since ischemic stroke onset. We analyzed the performance of Group Convolutional Neural Networks compared to other deep learning methods. Unlabeled imaging data was used for pre-training. Prediction performance of the best deep learning model was compared to the performance of four independent junior and senior raters. Additionally, in cases deemed indeterminable by human raters, model ratings were used to augment human performance. Post-hoc gradient-based explanations were analyzed to gain insights into model predictions.

RESULTS

Our best predictive model performed comparably to human raters. Using model ratings in cases deemed indeterminable by human raters improved rating accuracy and interrater agreement for junior and senior ratings. Post-hoc explainability analyses showed that the model localized stroke lesions to derive predictions.

DISCUSSION

Our analysis shows that deep learning based clinical decision support has the potential to improve the accessibility of the DWI-FLAIR mismatch concept by supporting patient stratification.

Cost-effectiveness of Endovascular Treatment for Acute Stroke with Large Infarct: A United States Perspective

Background The health economic benefit of endovascular treatment (EVT) in addition to best medical management for acute ischemic stroke with large ischemic core is uncertain. Purpose To assess the cost-effectiveness of EVT plus best medical management versus best medical management alone in treating acute ischemic stroke with large vessel occlusion and a baseline Alberta Stroke Program Early CT Score (ASPECTS) 3-5. Materials and Methods This is a secondary analysis of the randomized RESCUE-Japan LIMIT (Recovery by Endovascular Salvage for Cerebral Ultra-acute Embolism-Japan Large Ischemic Core Trial), with enrollment November 2018 to September 2021, in which the primary outcome was the modified Rankin Scale (mRS) score at 90 days. Participants with a baseline ASPECTS 3-5 (on the basis of noncontrast CT and diffusion-weighted imaging) were randomized 1:1 to receive EVT plus best medical management (n = 100) or best medical management alone (n = 102). The primary outcome of the current study was cost-effectiveness, determined according to the incremental cost-effectiveness ratio (ICER). A decision model consisting of a short-term component (cycle length of 3 months) and a long-term Markov state transition component (cycle length of 1 year) was used to estimate expected lifetime costs and quality-adjusted life-years (QALYs) from health care and societal perspectives in the United States. Upper and lower willingness-to-pay (WTP) thresholds were set at $100 000 and $50 000 per QALY, respectively. A deterministic one-way sensitivity analysis to determine the impact of participant age and a probabilistic sensitivity analysis to assess the impact of parameter uncertainty were conducted. Results A total of 202 participants were included in the study (mean age, 76 years ± 10 [SD]; 112 male). EVT plus best medical management resulted in ICERs of $15 743 (health care perspective) and $19 492 (societal perspective). At the lower and upper WTP thresholds, EVT was cost-effective up to 85 and 90 years (health care perspective) and 84 and 89 years (societal perspective) of age, respectively. When analyzing participants with the largest infarcts (ASPECTS 3) separately, EVT was not cost-effective (ICER, $337 072 [health care perspective] and $383 628 [societal perspective]). Conclusion EVT was cost-effective for participants with an ASPECTS 4-5, but not for those with an ASPECTS 3. ClinicalTrials.gov registration no. NCT03702413 © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Widjaja in this issue.

Persistent perfusion abnormalities at day 1 correspond to different clinical trajectories after stroke

BACKGROUND

Perfusion abnormalities after thrombolysis are frequent within and surrounding ischemic lesions, but their relative frequency is not well known.

OBJECTIVE

To describe the different patterns of perfusion abnormalities observed at 24 hours and compare the characteristics of the patients according to their perfusion pattern.

METHODS

From our thrombolysis registry, we included 226 consecutive patients with an available arterial spin labeling (ASL) perfusion sequence at day 1. We performed a blinded assessment of the perfusion status (hypoperfusion-h, hyperperfusion-H, or normal-N) in the ischemic lesion and in the surrounding tissue. We compared the time course of clinical recovery, the rate of arterial recanalization, and hemorrhagic transformations in the different perfusion profiles.

RESULTS

We identified seven different perfusion profiles at day 1. Four of these (h/h, h/H, H/H, and H/N) represented the majority of the population (84.1%). The H/H profile was the most frequent (34.5%) and associated with 3-month good outcome (modified Rankin Scale (mRS): 63.5%). Patients with persistent hypoperfusion within and outside the lesion (h/h, 12.4%) exhibited worse outcomes after treatment (mRS score 0-2: 23.8%) than other patients, were less frequently recanalized (40.7%), and had more parenchymal hematoma (17.8%). The h/H profile had an intermediate clinical trajectory between the h/h profile and the hyperperfused profiles.

CONCLUSION

ASL hypoperfusion within the infarct and the surrounding tissue was associated with poor outcome. A more comprehensive view of the mechanisms in the hypoperfused surrounding tissue could help to design new therapeutic approaches during and after reperfusion therapies.

CT-Based and CTA-Based Posterior Circulation ASPECTS in Acute Basilar Artery Occlusion: An Agreement Study

BACKGROUND

Posterior circulation Acute Stroke Prognosis Early Computed Tomography Scores (pc-ASPECTS) is a rapid method of measuring early ischemic changes. However, there was no agreement study about pc-ASPECTS.

OBJECTIVES

The purpose of the present study was to evaluate the agreement on pc-ASPECTS based on non-contrast CT (NCCT) and CT angiography (CTA) source images in patients with acute basilar artery occlusion (BAO).

METHODS

We prospectively enrolled consecutive patients with acute BAO from January 2022 to August 2022 at the First Affiliated Hospital of University of Science and Technology of China. The NCCT and CTA were scored independently by 15 raters during 2 different reading sessions at least 3 weeks apart. The pc-ASPECTS based on NCCT and CTA were analyzed on the full scale or were dichotomized (0-6 vs. 7-10, 0-7 vs. 8-10, and 0-8 vs. 9-10). The level of agreement was measured using Fleiss κ statistics.

RESULTS

The median (IQR) CT-based pc-ASPECTS was 8 (6.75-9). The interrater agreement for CT-based pc-ASPECTS (κ = 0.133 [0.132-0.133]) and CTA-based pc-ASPECTS (κ = 0.204 [0.203-0.204]) was slight for all raters. Dichotomizations obtaining the highest concordance for the CT-based pc-ASPECTS (0-6 vs. 7-10) and the CTA-based pc-ASPECTS (0-8 vs. 9-10) failed to increase the interrater agreement to a substantial level (κ = 0.350 [0.348-0.351] and 0.396 [0.395-0.398], respectively). Intrarater agreement for global CT-based pc-ASPECTS was less than substantial for the 14/15 raters and reached the level of substantial for the 3/15 raters with dichotomization.

CONCLUSIONS

Agreement between clinicians assessing CT-based and CTA-based pc-ASPECTS cannot be sufficient to make reproducible clinical decisions and assessments. The dichotomization failed to improve interrater and intrarater agreement to the level of substantial.

Artificial Intelligence for Automated DWI/FLAIR Mismatch Assessment on Magnetic Resonance Imaging in Stroke: A Systematic Review

We conducted this Systematic Review to create an overview of the currently existing Artificial Intelligence (AI) methods for Magnetic Resonance Diffusion-Weighted Imaging (DWI)/Fluid-Attenuated Inversion Recovery (FLAIR)-mismatch assessment and to determine how well DWI/FLAIR mismatch algorithms perform compared to domain experts. We searched PubMed Medline, Ovid Embase, Scopus, Web of Science, Cochrane, and IEEE Xplore literature databases for relevant studies published between 1 January 2017 and 20 November 2022, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. We assessed the included studies using the Quality Assessment of Diagnostic Accuracy Studies 2 tool. Five studies fit the scope of this review. The area under the curve ranged from 0.74 to 0.90. The sensitivity and specificity ranged from 0.70 to 0.85 and 0.74 to 0.84, respectively. Negative predictive value, positive predictive value, and accuracy ranged from 0.55 to 0.82, 0.74 to 0.91, and 0.73 to 0.83, respectively. In a binary classification of ±4.5 h from stroke onset, the surveyed AI methods performed equivalent to or even better than domain experts. However, using the relation between time since stroke onset (TSS) and increasing visibility of FLAIR hyperintensity lesions is not recommended for the determination of TSS within the first 4.5 h. An AI algorithm on DWI/FLAIR mismatch assessment focused on treatment eligibility, outcome prediction, and consideration of patient-specific data could potentially increase the proportion of stroke patients with unknown onset who could be treated with thrombolysis.

FLAIR vascular hyperintensity-DWI mismatch and DWI-FLAIR mismatch ASPECTS for prediction of good outcome after recanalization in anterior circulation stroke; multicenter observational study

INTRODUCTION

Acute ischemic stroke is a medical emergency caused by decreased blood flow to the brain, leading cause of long-term disability. Recanalization, one of the most concerning difficulties linked with intracranial arterial occlusion, has been used to reduce mortality in ischemic stroke treatment. The mismatch concepts MR PWI-DWI or DWI-FLAIR can help identify patients for thrombolysis.

PURPOSE

This paper introduces a novel method of predicting revascularization using the value of fluid-attenuated inversion recovery vascular hyperintensity FVH-DWI mismatch and DWI-FLAIR mismatch, which releases anterior circulation large vessel occlusion (LVO) after endovascular thrombectomy (EVT). Moreover, we present a new scoring system following anatomical region distributed for MCA territory called a DWI-FLAIR MISMATCH ASPECTS.

RESULT

Statistical analysis was performed to predict revascularization and functional outcome with 110 patients with anterior circulation LVO treated with EVT. We found that FVH-DWI mismatch was present in 71 patients (89.9 %) with complete revascularization and present in 8 patients (10.1 %) with no/partial revascularization, which had no significant difference (p = 0.12), and there was no significant difference between good functional outcome and poor functional outcome. Moreover, in 76 patients with DWI-FLAIR mismatch ASPECTS of > 6 point-group, present FVH-DWI mismatch in 57 patients (83.8 %) with complete revascularization had a significant difference as compared to 11 patients (16.2 %) with absent FVH-DWI mismatch (p < 0.05). The clinical outcome in complete revascularization is better than no/partial revascularization, and complete revascularization is independently associated with good functional outcomes (p < 0.05).

CONCLUSION

FVH-DWI mismatch paired with DWI-FLAIR mismatch ASPECTS > 6 points may be possible to predict revascularization in patients with anterior circulation LVO.

Comparison of semi-quantitative and visual assessment of early MRI signal evolution in acute ischaemic stroke

Background

The evaluation of DWI/FLAIR mismatch in ischaemic stroke patients with unknown, time from onset can determine the treatment strategy. This approach is based on, visual assessment and may be subject to insufficient inter-rater agreement.

Objective

To compare the inter-rater agreement of visual evaluation of FLAIR MRI and proposed region of interest (ROI) semiquantitative method in large vessel occlusion (LVO) strokes.

Methods

Five readers have analysed MRIs of 104 patients obtained within six hours of the onset of stroke symptoms resulting from LVO visually and semi-quantitatively. For the semiquantitative analysis, a ROI method was used to obtain relative signal intensity compared to the unaffected side. Cut-off values of 1.15 and 1.10 were tested. The analysis yielded FLAIR-positive (abnormal) and negative (normal) findings. Percentage agreement and Fleiss kappa coefficients were calculated.

Results

The visual agreement of 5/5 readers and ≥ 4/5 readers occurred in 31% and 59% of cases respectively. Semi-quantitative evaluation using a cut-off value of 1.15 increased the agreements to 67% and 88% respectively. The agreement of visual evaluation was fair. The semi-quantitative method utilising the cut-off of 1.15 had moderate agreement although it increased the number of FLAIR-negative results compared to the visual evaluation. A low cut-off value of 1.10 didn't improve the agreement significantly.

Conclusion

The inter-rater agreement of visual evaluation of FLAIR in patients with short-duration large vessel occlusion stroke was fair. The high cut-off value of semiquantitative evaluation increased the agreement although it changed the proportion of FLAIR positive and negative results.

Deep Learning in Ischemic Stroke Imaging Analysis: A Comprehensive Review

Ischemic stroke is a cerebrovascular disease with a high morbidity and mortality rate, which poses a serious challenge to human health and life. Meanwhile, the management of ischemic stroke remains highly dependent on manual visual analysis of noncontrast computed tomography (CT) or magnetic resonance imaging (MRI). However, artifacts and noise of the equipment as well as the radiologist experience play a significant role on diagnostic accuracy. To overcome these defects, the number of computer-aided diagnostic (CAD) methods for ischemic stroke is increasing substantially during the past decade. Particularly, deep learning models with massive data learning capabilities are recognized as powerful auxiliary tools for the acute intervention and guiding prognosis of ischemic stroke. To select appropriate interventions, facilitate clinical practice, and improve the clinical outcomes of patients, this review firstly surveys the current state-of-the-art deep learning technology. Then, we summarized the major applications in acute ischemic stroke imaging, particularly in exploring the potential function of stroke diagnosis and multimodal prognostication. Finally, we sketched out the current problems and prospects.

Understanding the reliability of trial outcome measures: The example of angiographic results of surgical or endovascular treatments of aneurysms

BACKGROUND

The reliability of outcome measures is of central importance in clinical research. Studies of reliability remain rare in the neurovascular field.

METHODS

A narrative review of the history (1997-2021) of reporting angiographic results of the surgical or endovascular treatments of aneurysms serves to illustrate the importance of precisely defining outcome measures. We also review how the reliability of an angiographic classification system was studied.

DISCUSSION

Outcome measures are commonly used without precise definitions. When definitions or classification systems exist, they are rarely verified for their reliability. Twenty-five years following its introduction, a classification of angiographic results of aneurysm treatments is still being studied and modified.

CONCLUSION

The reliability of outcome measures should be made a research priority if we are to practice outcome-based medical or surgical care.

Intra- and inter-rater consistency of dual assessment by radiologist and neurologist for evaluating DWI-ASPECTS in ischemic stroke

OBJECTIVES

To estimate the intra -and inter-rater consistency of radiologist and neurologist working in pairs attributing DWI-ASPECTS (Diffusion Alberta Stroke Program Early CT Score) in patients with acute middle cerebral artery ischemic stroke referred for mechanical thrombectomy, intravenous thrombolysis or bridging therapy.

METHODS

Five neurologists and 5 radiologists working in pairs and in hour period scored independently and in two reading sessions anonymized DWI-ASPECTS of 80 patients presenting with acute anterior ischaemic stroke in our center. We measured agreement between pairs using intraclass correlation coefficients (ICCs). A Fleiss kappa was used for dichotomized (0-6;7-10) and trichotomized (0-3;4-6;7-10) ASPECTS. The interrater distribution of the score in the trichotomized (0-3;4-6;7-10) ASPECTS was calculated. We determined the interrater (Cohen kappa) and intrarater (Fleiss kappa) agreement on the ASPECTS regions.

RESULTS

The average DWI-ASPECTS was 6.35 (SD±2.44) for the first reading, and 6.47 (SD±2.44) for the second one. The ICC was 0.853 (95%CI, 0.798-0.896) for the interrater, and 0.862 (95%CI, 0.834-0.885) for the intrarater evaluation. Kappa coefficients were high for dichotomized (k=0.75) and trichotomized (k=0.64) ASPECTS. Evaluators agreement on the ASPECTS category (0-3), (4-6) and (7-10) was 88, 76 and 93% respectively. The anatomic region infarcted was well identified (k=0.70-0.77), except for the internal capsula (k=0.57). Interrater agreement was fair for M5 (k=0.37), moderate for internal capsula (0.52) and substantial for the other regions (0.60-0.79).

CONCLUSIONS

Reliability of DWI-ASPECTS is good when determined by radiologist and neurologist working in pairs, which corresponds to our current clinical practice. However, discrepancies are possible for cut-off determination, which may impact the indication of thrombectomy, and for the determination of the exact infarcted region. Agreement to propose category (4-6) is lower than for (0-3) and (8-10) ASPECTS categories.

A Semi-Supervised Learning Framework to Leverage Proxy Information for Stroke MRI Analysis

Treating acute ischemic stroke (AIS) patients is a time-sensitive endeavor, as therapies target areas experiencing ischemia to prevent irreversible damage to brain tissue. Depending on how an AIS is progressing, thrombolytics such as tissue-plasminogen activator (tPA) may be administered within a short therapeutic window. The underlying conditions for optimal treatment are varied. While previous clinical guidelines only permitted tPA to be administered to patients with a known onset within 4.5 hours, clinical trials demonstrated that patients with signal intensity differences between diffusion-weighted imaging (DWI) and fluid-attenuated inversion recovery (FLAIR) sequences in an MRI study can benefit from thrombolytic therapy. This intensity difference, known as DWI-FLAIR mismatch, is prone to high inter-reader variability. Thus, a paradigm exists where onset time serves as a weak proxy for DWI-FLAIR mismatch. In this study, we sought to detect DWI-FLAIR mismatch in an automated fashion, and we compared this to assessments done by three expert neuroradiologists. Our approach involved training a deep learning model on MRI to classify tissue clock and leveraging time clock as a weak proxy label to supplement training in a semi-supervised learning (SSL) framework. We evaluate our deep learning model by testing it on an unseen dataset from an external institution. In total, our proposed framework was able to improve detection of DWI-FLAIR mismatch, achieving a top ROC-AUC of 74.30%. Our study illustrated that incorporating clinical proxy information into SSL can improve model optimization by increasing the fidelity of unlabeled samples included in the training process.

Basal Ganglia versus Peripheral Infarcts: Predictive Value of Early Fiber Alterations

BACKGROUND AND PURPOSE

Impairment of fiber integrity of the corticospinal tract in the subacute and chronic phases after ischemic stroke has been linked to poor motor outcome. The aim of the study was an assessment of fiber integrity in the acute poststroke phase and an evaluation of its association with the clinical course dependent on the infarction pattern (subtypes: peripheral versus basal ganglia infarction).

MATERIALS AND METHODS

All patients who underwent mechanical recanalization of a large-vessel occlusion in the anterior circulation and postinterventional DTI were included (n = 165). The fractional anisotropy index of the patient-specific corticospinal tract within the posterior limb of the internal capsule was correlated to clinical parameters (NIHSS scores/mRS at 90 days), and the interaction of stroke subtype (peripheral infarcts versus basal ganglia infarction) was tested in a moderation analysis.

RESULTS

The fractional anisotropy index was reduced in the acute poststroke phase with a correlation to clinical presentation, especially in case of peripheral infarcts (eg, with the NIHSS motor subscore: r = -0.4, P < .001). This correlation was absent for basal ganglia infarction (r = -0.008, P > .05). There was a significant association between the fractional anisotropy index and clinical outcome (mRS after 90 days, P < .01), which is moderated by stroke subtype with significant effects only for peripheral infarcts.

CONCLUSIONS

Corticospinal tract abnormalities can be observed in the early stage after mechanical recanalization and have prognostic capacity. This finding increases the clinical value of early DTI imaging parameters. Because the effects observed were limited to peripheral infarcts, further and longitudinal evaluation of fiber integrities within basal ganglia infarction is required.

Clinical feasibility of single-shot fluid-attenuated inversion recovery with wide inversion recovery pulse designed to reduce cerebrospinal fluid and motion artifacts for evaluation of uncooperative patients in acute stroke protocol

Fluid-attenuated inversion recovery (FLAIR) imaging is a key sequence for stroke assessment. Motion artifact reduction with short acquisition time is still challenging, but necessary in the magnetic resonance (MR) stroke protocol, especially for uncooperative patients suspected of stroke. The aim of this study is to investigate the feasibility of modified single-shot FLAIR with wide inversion recovery pulses for use in stroke patients. This is a prospective study, which included 30 patients clinically suspected of stroke who were examined by MR stroke protocol from January 2018 to September 2018. A 1.5 T, multi-shot-turbo spin-echo (TSE) conventional FLAIR, and single-shot-TSE-FLAIR with wide inversion recovery pulse were used. Modified single-shot FLAIR was obtained for 30 patients with suspected stroke who moved during conventional FLAIR scan. Motion artifacts were randomly and independently scored using a 5-grade scale by three radiologists in blinded fashion. Whether the FLAIR vessel hyperintensity sign was present was visually evaluated. Statistical tests included Wilcoxon-signed rank test and weighted Cohen's kappa statistics. The motion artifact score was significantly lower in single-shot FLAIR than in conventional FLAIR (0.37 ± 0.56 vs. 1.83 ± 1.18; p < 0.05. The vessel hyperintensity sign was visualized in 6 and 5 patients on single-shot and conventional FLAIR images, respectively. This study demonstrates the value of single-shot FLAIR for stroke assessment. Single-shot FLAIR reduced motion artifact and visualized vessel hyperintensity sign more than conventional FLAIR. LEVEL OF EVIDENCE: 2. TECHNICAL EFFICACY STAGE: 2.

Predictive value of Alberta stroke program early CT score for perfusion weighted imaging - diffusion weighted imaging mismatch in stroke with middle cerebral artery occlusion

This study aimed to quantitatively assess the consistency and correlation between perfusion weighted imaging (PWI)/ diffusion weighted imaging (DWI) Alberta Stroke Program Early CT Score (ASPECTS) mismatch and PWI/DWI mismatch.Sixty-eight acute ischemic stroke with middle cerebral artery occlusion who underwent magnetic resonance imaging before thrombectomy were eligible. DWI volume, PWI volume and PWI-DWI mismatch were measured. DWI-, PWI-, PWI-DWI ASPECTS were evaluated. Statistical analysis was performed to compare the correlation between volume and ASPECTS of DWI-, PWI- and PWI-DWI mismatch. Receiver operating characteristic curve analysis was used to assess the predictive value of the PWI-DWI ASPECTS mismatch for the occurrence of PWI-DWI mismatch in acute ischemic stroke patients with middle cerebral artery occlusion.Of 68 patients, the DWI volume, PWI volume and PWI-DWI mismatch volume were (27.76 ± 17.53) mL, (167.09 ± 59.64) mL and (139.33 ± 58.18) mL respectively. DWI-ASPECTS was 6.75 ± 1.90 with the interobserver agreement was κ=0.98 (95% CI, 0.95-0.99); PWI-ASPECTS was 3.09 ± 2.11 with the interobserver agreement was κ=0.95 (95% CI, 0.91-0.99); PWI-DWI ASPECTS mismatch was 6.75 ± 1.90. Spearman's rank correlation analysis revealed that PWI-DWI mismatch volume was negatively correlated with PWI-DWI ASPECTS mismatch (r = -0.802; P = .000). Receiver operating characteristic analysis showed that when the PWI-DWI ASPECTS mismatch cut point was ≥ 2, the under curve of PWI-DWI ASPECTS mismatch for predicting PWI-DWI mismatch was 0.954 (95%CI, 0.911-0.998), with the sensitivity and specificity were 84.00% and 100% respectively.PWI-DWI ASPECTS mismatch may represent a convenient surrogate for penumbra in clinical trials.

Comparison of automated and manual DWI-ASPECTS in acute ischemic stroke: total and region-specific assessment

OBJECTIVE

To compare the DWI-Alberta Stroke Program Early Computed Tomography Score calculated by a deep learning-based automatic software tool (eDWI-ASPECTS) with the neuroradiologists' evaluation for the acute stroke, with emphasis on its performance on 10 individual ASPECTS regions, and to determine the reasons for inconsistencies between eDWI-ASPECTS and neuroradiologists' evaluation.

METHODS

This retrospective study included patients with middle cerebral artery stroke who underwent MRI from 2010 to 2019. All scans were evaluated by eDWI-ASPECTS and two independent neuroradiologists (with 15 and 5 years of experience in stroke study). Inter-rater agreement and agreement between manual vs. automated methods for total and each region were evaluated by calculating Kendall's tau-b, intraclass correlation coefficient (ICC), and kappa coefficient.

RESULTS

In total, 309 patients met our study criteria. For total ASPECTS, eDWI-ASPECTS and manual raters had a strong positive correlation (Kendall's tau-b = 0.827 for junior raters vs. eDWI-ASPECTS; Kendall's tau-b = 0.870 for inter-raters; Kendall's tau-b = 0.848 for senior raters vs. eDWI-ASPECTS) and excellent agreement (ICC = 0.923 for junior raters and automated scores; ICC = 0.954 for inter-raters; ICC = 0.939 for senior raters and automated scores). Agreement was different for individual ASPECTS regions. All regions except for M5 region (κ = 0.216 for junior raters and automated scores), internal capsule (κ = 0.525 for junior raters and automated scores), and caudate (κ = 0.586 for senior raters and automated scores) showed good to excellent concordance.

CONCLUSION

The eDWI-ASPECTS performed equally well as senior neuroradiologists' evaluation, although interference by uncertain scoring rules and midline shift resulted in poor to moderate consistency in the M5, internal capsule, and caudate nucleus regions.

KEY POINTS

• The eDWI-ASPECTS based on deep learning perform equally well as senior neuroradiologists' evaluations. • Among the individual ASPECTS regions, the M5, internal capsule, and caudate regions mainly affected the overall consistency. • Uncertain scoring rules and midline shift are the main reasons for regional inconsistency.

Endovascular Thrombectomy for Low ASPECTS Large Vessel Occlusion Ischemic Stroke: A Systematic Review and Meta-Analysis

Background:

The current American Heart Association guidelines for acute ischemic stroke reserve Grade 1A recommendation for the use of endovascular thrombectomy (EVT) for patients with an Alberta Stroke Program Early Computed Tomography Score (ASPECTS) of ≥6.

Objective:

We aim to determine the safety and efficacy of EVT for large vessel occlusion ischemic stroke patients with low ASPECTS (5 or less).

Methods:

Medline, Cochrane Central Register of Controlled Trials, and ClinicalTrials.gov were searched for studies appraising the outcomes of EVT for low ASPECTS ischemic stroke. A meta-analysis of proportions compared the clinical outcomes of patients undergoing EVT and those receiving best medical therapy only.

Results:

Nine studies (1,196 patients) were included. There was a trend (p = 0.11) toward a higher rate of symptomatic intracranial hemorrhage (sICH) in the EVT group (9.2%; 95% CI 6.1–13.6; I2 53.37%) compared to the medical group (5.5%; 95% CI 3.7–8.1; I2 0%). There was no difference (p = 0.41) in the pooled 90-day mortality of EVT patients (30.7%; 95% CI 21.7–41.5; I2 84.23%) and medical patients (36.6%; 95% CI 26.4–48.1; I2 76.2%). EVT patients had better (p = 0.001) 90-day outcomes, with 27.7% (95% CI 21.8–34.5; I2 62.08%) of patients attaining a modified Rankin Scale of 0–2 compared to only 3.7% (95% CI 2.3–5.9; I2 87.21%) in the medical group.

Conclusions:

This meta-analysis demonstrates a trend in higher sICH among low ASPECTS patients undergoing EVT. Despite this, a significant proportion of this subset of patients still achieved good functional outcomes at 90 days. Randomized trials are necessary to substantiate this result as significant bias is inherent in the observational studies included in this review.

Effect of hyperglycemia on stroke outcome is not homogeneous to all patients treated with mechanical thrombectomy

OBJECTIVES

Admission hyperglycemia is a penumbra-modifying factor that is associated with poor functional outcome in acute ischemic stroke (AIS) patients treated with intravenous rt-PA and/or mechanical thrombectomy (MT). Insulin therapy has failed to demonstrate a clinical benefit and the question of the patient selection remains under debate. We assessed the relationship between admission glycemia (AG) and functional outcome in AIS patients treated by MT according to both penumbra characteristics and reperfusion status.

PATIENTS AND METHODS

We performed a retrospective analysis of a multi-center registry of consecutive AIS (NIHSS ≥ 10) due to middle cerebral artery occlusion treated by MT (± tissue Plasminogen Activator (tPA)). To evaluate the association between AG and the 3-month functional outcome (modified Rankin Scale (mRS) ≤2), univariable and multivariable analyses were used. Subgroup analyses were performed according to both clinical-ASPECTS Mismatch (CAM2) and the complete recanalization (CR) status defined by a mTICI scale (modified Thrombolysis in Cerebral Infarction) 2b/3.

RESULTS

216 AIS patients were included (Median Age: 68.43[58.12-77.95], median NIHSS: 18[15-21]). 104/216 (48.15%) patients had mRS≤2 at 3 months. AG was an independent predictor of functional outcome (/1 g/L OR: 0.10[0.03-0.37]) after adjusting for potential cofounders. Among subgroups formed by combining CAM2 and CR, AG was found to be predictor of functional outcome only in CAM2⁺/CR⁺ and specifically when recanalization was early.

CONCLUSION

This study highlights the fact that the relationship between AG and prognosis is not homogeneous for all patients and indicates that AG has a deleterious effect on the ischemic penumbra, thus explaining its statistical association with functional outcome. Stroke neuroprotection by targeting hyperglycemia should be considered in acute stroke patients with mismatch and early complete recanalization. More prospective randomized trials are needed to generalize the conclusions.

Mechanical Thrombectomy in Ischemic Stroke Patients With Alberta Stroke Program Early Computed Tomography Score 0-5

Supplemental Digital Content is available in the text.

Background and Purpose—

If anterior circulation large vessel occlusion acute ischemic stroke patients presenting with ASPECTS 0–5 (Alberta Stroke Program Early CT Score) should be treated with mechanical thrombectomy remains unclear. Purpose of this study was to report on the outcome of patients with ASPECTS 0–5 treated with mechanical thrombectomy and to provide data regarding the effect of successful reperfusion on clinical outcomes and safety measures in these patients.

Methods—

Multicenter, pooled analysis of 7 institutional prospective registries: Bernese-European Registry for Ischemic Stroke Patients Treated Outside Current Guidelines With Neurothrombectomy Devices Using the SOLITAIRE FR With the Intention for Thrombectomy (Clinical Trial Registration—URL: https://www.clinicaltrials.gov. Unique identifier: NCT03496064). Primary outcome was defined as modified Rankin Scale 0–3 at day 90 (favorable outcome). Secondary outcomes included rates of day 90 modified Rankin Scale 0–2 (functional independence), day 90 mortality and occurrence of symptomatic intracerebral hemorrhage. Multivariable logistic regression analyses were performed to assess the association of successful reperfusion with clinical outcomes. Outputs are displayed as adjusted Odds Ratios (aOR) and 95% CI.

Results—

Two hundred thirty-seven of 2046 patients included in this registry presented with anterior circulation large vessel occlusion and ASPECTS 0–5. In this subgroup, the overall rates of favorable outcome and mortality at day 90 were 40.1% and 40.9%. Achieving successful reperfusion was independently associated with favorable outcome (aOR, 5.534; 95% CI, 2.363–12.961), functional independence (aOR, 5.583; 95% CI, 1.964–15.873), reduced mortality (aOR, 0.180; 95% CI, 0.083–0.390), and lower rates of symptomatic intracerebral hemorrhage (aOR, 0.235; 95% CI, 0.062–0.887). The mortality-reducing effect remained in patients with ASPECTS 0–4 (aOR, 0.167; 95% CI, 0.056–0.499). Sensitivity analyses did not change the primary results.

Conclusions—

In patients presenting with ASPECTS 0–5, who were treated with mechanical thrombectomy, successful reperfusion was beneficial without increasing the risk of symptomatic intracerebral hemorrhage. Although the results do not allow for general treatment recommendations, formal testing of mechanical thrombectomy versus best medical treatment in these patients in a randomized controlled trial is warranted.

MT-DRAGON score for outcome prediction in acute ischemic stroke treated by mechanical thrombectomy within 8 hours

OBJECTIVES

The MRI-DRAGON score includes clinical and MRI parameters and demonstrates a high specificity in predicting 3 month outcome in patients with acute ischemic stroke (AIS) treated with intravenous tissue plasminogen activator (IV tPA). The aim of this study was to adapt this score to mechanical thrombectomy (MT) in a large multicenter cohort.

METHODS

Consecutive cases of AIS treated by MT between January 2015 and December 2017 from three stroke centers were reviewed (n=1077). We derived the MT-DRAGON score by keeping all variables of the MRI-DRAGON score (age, initial National Institutes of Health Stroke Scale score, glucose level, pre-stroke modified Rankin Scale (mRS) score, diffusion weighted imaging-Alberta Stroke Program Early CT score ≤5) and considering the following variables: time to groin puncture instead of onset to IV tPA time and occlusion site. Unfavorable 3 month outcome was defined as a mRS score >2. Score performance was evaluated by c statistics and an external validation was performed.

RESULTS

Among 679 included patients (derivation and validation cohorts, n=431 and 248, respectively), an unfavorable outcome was similar between the derivation (51.5%) and validation (58.1%, P=0.7) cohorts, and was significantly associated with all MT-DRAGON parameters in the multivariable analysis. The c statistics for unfavorable outcome prediction was 0.83 (95%CI 0.79 to 0.88) in the derivation and 0.8 (95%CI 0.75 to 0.86) in the validation cohort. All patients (n=55) with an MT-DRAGONscore ≥11 had an unfavorable outcome and 60/63 (95%) patients with an MT-DRAGON score ≤2 points had a favorable outcome.

CONCLUSION

The MT-DRAGON score is a simple tool, combining admission clinical and radiological parameters that can reliably predict 3 month outcome after MT.

Intravenous thrombolysis and thrombectomy decisions in acute ischemic stroke: An interrater and intrarater agreement study

PURPOSE

We aimed to assess agreement on intravenous tissue-plasminogen activator (IV tPA) and mechanical thrombectomy (MT) management decisions in acute ischemic stroke (AIS) patients. Secondary objectives were to assess agreement on Diffusion-Weighted-Imaging-Alberta-Stroke-Program-EArly-CT-Score (DWI-ASPECTS), and clinicians' willingness to recruit patients in a randomized controlled trial (RCT) comparing medical management with or without MT.

MATERIALS AND METHODS

Studies assessing agreement of IV tPA and MT were systematically reviewed. An electronic portfolio of 41 AIS patients was sent to randomly selected providers at French stroke centers. Raters were asked 4 questions for each case: (1) What is the DWI-ASPECTS? (2) Would you perform IV tPA? (3) Would you perform MT? (4) Would you include the patient in a RCT comparing standard medical therapy with or without MT? Twenty responders were randomly selected to study intrarater agreement. Agreement was assessed using Fleiss' Kappa statistics.

RESULTS

The review yielded two single center studies involving 2-5 raters, with various results. The electronic survey was answered by 86 physicians (60 vascular neurologists and 26 interventional neuroradiologists). The interrater agreement was moderate for IV tPA treatment decisions (κ=0.565 [0.420-0.680]), but only fair for MT (κ=0.383 [0.289-0.491]) and for combined treatment decisions (κ=0.399 [0.320-0.486]). The intrarater agreement was at least substantial for the majority of raters. The interrater agreement for DWI-ASPECTS was fair (κ=0.325 [0.276-0.387]). Physicians were willing to include a mean of 14±9 patients (33.1%±21.7%) in a RCT.

CONCLUSION

Disagreements regarding the use of IVtPA or MT in the management of AIS patients remain frequent. Further trials are needed to resolve the numerous areas of uncertainty.

Impact of infarct location on functional outcome following endovascular therapy for stroke

OBJECTIVES

The relationship between stroke topography (ie, the regions damaged by the infarct) and functional outcome can aid clinicians in their decision-making at the acute and later stages. However, the side (left or right) of the stroke may also influence the identification of clinically relevant regions. We sought to determine which brain regions are associated with good functional outcome at 3 months in patients with left-sided and right-sided stroke treated by endovascular treatment using the diffusion-weighted imaging-Alberta Stroke Program Early CT Score (DWI-ASPECTS).

METHODS

Patients with ischaemic stroke (n = 405) were included from the ASTER trial and Pitié-Salpêtrière registry. Blinded readers rated ASPECTS on day 1 DWI. Stepwise logistic regression analyses were performed to identify the regions related to 3-month outcome in left (n = 190) and right (n = 215) sided strokes with the modified Rankin scale (0-2) as a binary independent variable and with the 10 regions-of-interest of the DWI-ASPECTS as independent variables.

RESULTS

Median National Institute of Health Stroke Scale (NIHSS) at baseline was 17 (IQR: 12-20), median age was 70 years (IQR: 58-80) and median day-one NIHSS 9 (IQR: 4-18). Not all brain regions have the same weight in predicting good outcome at 3 months; moreover, these regions depend on the affected hemisphere. In left-sided strokes, the multivariate analysis revealed that preservation of the caudate nucleus, the internal capsule and the cortical M5 region were independent predictors of good outcome. In right-sided strokes, the cortical M3 and M6 regions were found to be clinically relevant.

CONCLUSION

Cortical non-motors areas related to outcome differed between left-sided and right-sided strokes. This difference might reflect the specialisation of the dominant and non-dominant hemispheres for language and attention, respectively. These results may influence decision-making at the acute and later stages.

TRIAL REGISTRATION NUMBER

NCT02523261.

Thrombectomy for Stroke at 6-24 hours without Perfusion CT Software for Patient Selection

BACKGROUND

Although, thrombectomy for stroke more than 6 hours after onset supported by automated perfusion computed tomography (CT) software (RAPID, iSchemaView) is effective, this software is not available in Japan. This study aimed to elucidate the efficacy of thrombectomy 6-24 hours after onset in our patient cohort using conventional imaging mismatch.

METHODS

Of 586 ischemic stroke patients who underwent thrombectomy registered from January 2015 to December 2017, patients with occlusion of the intracranial internal carotid artery or middle cerebral artery, who had last been known to be well 6-24 hours earlier and who had a prestroke modified Rankin scale (mRS) score 0 or 1 were enrolled. Clinical outcomes were the scores of the utility-weighted (UW) mRS, which ranges from 0 (death) to 10 (no symptom or disability), and the rate of functional independence (mRS score of 0-2) at 90 days.

RESULTS

This study sample included 31 patients. The median baseline National Institutes of Health Stroke Scale score was 17 (interquartile range [IQR], 13-20), and the median Diffusion-Weighted Imaging-Alberta Stroke Program Early CT Score was 7 (IQR, 5-8). The median interval between the time that the patient was last known well and revascularization was 741 (IQR, 641-818) minutes. The mean UW mRS score at 90 days was 5.3, the rate of functional independence was 32%, and the 90-day mortality rate was 13%.

CONCLUSIONS

Thrombectomy 6-24 hours after onset which can be performed with conventional imaging mismatch might be secured for improving functional independence in stroke patients.

Agreement between core laboratory and study investigators for imaging scores in a thrombectomy trial

PURPOSE

We aimed to assess the agreement between study investigators and the core laboratory (core lab) of a thrombectomy trial for imaging scores.

METHODS

The Alberta Stroke Program Early CT Score (ASPECTS), the European Collaborative Acute Stroke Study (ECASS) hemorrhagic transformation (HT) classification, and the Thrombolysis In Cerebral Infarction (TICI) scores as recorded by study investigators were compared with the core lab scores in order to assess interrater agreement, using Cohen's unweighted and weighted kappa statistics.

RESULTS

There were frequent discrepancies between study sites and core lab for all the scores. Agreement for ASPECTS and ECASS HT classification was less than substantial, with disagreement occurring in more than one-third of cases. Agreement was higher on MRI-based scores than on CT, and was improved after dichotomization on both CT and MRI. Agreement for TICI scores was moderate (with disagreement occurring in more than 25% of patients), and went above the substantial level (less than 10% disagreement) after dichotomization (TICI 0/1/2a vs 2b/3).

CONCLUSION

Discrepancies between scores assessed by the imaging core lab and those reported by study sites occurred in a significant proportion of patients. Disagreement in the assessment of ASPECTS and day 1 HT scores was more frequent on CT than on MRI. The agreement for the dichotomized TICI score (the trial's primary outcome) was substantial, with less than 10% of disagreement between study sites and core lab.

TRIAL REGISTRATION NUMBER

NCT02523261, Post-results.

Impact of Lesion Load Thresholds on Alberta Stroke Program Early Computed Tomographic Score in Diffusion-Weighted Imaging

Background and aims

Assessment of ischemic lesions on computed tomography or MRI diffusion-weighted imaging (DWI) using the Alberta Stroke Program Early Computed Tomography Score (ASPECTS) is widely used to guide acute stroke treatment. However, it has never been defined how many voxels need to be affected to label a DWI-ASPECTS region ischemic. We aimed to assess the effect of various lesion load thresholds on DWI-ASPECTS and compare this automated analysis with visual rating.

Materials and methods

We analyzed overlap of individual DWI lesions of 315 patients from the previously published predictive value of fluid-attenuated inversion recovery study with a probabilistic ASPECTS template derived from 221 CT images. We applied multiple lesion load thresholds per DWI-ASPECTS region (>0, >1, >10, and >20% in each DWI-ASPECTS region) to compute DWI-ASPECTS for each patient and compared the results to visual reading by an experienced stroke neurologist.

Results

By visual rating, median ASPECTS was 9, 84 patients had a DWI-ASPECTS score ≤7. Mean DWI lesion volume was 22.1 (±35) ml. In contrast, by use of >0, >1-, >10-, and >20%-thresholds, median DWI-ASPECTS was 1, 5, 8, and 10; 97.1% (306), 72.7% (229), 41% (129), and 25.7% (81) had DWI-ASPECTS ≤7, respectively. Overall agreement between automated assessment and visual rating was low for every threshold used (>0%: κ_w = 0.020 1%: κ_w = 0.151; 10%: κ_w = 0.386; 20% κ_w = 0.381). Agreement for dichotomized DWI-ASPECTS ranged from fair to substantial (≤7: >10% κ = 0.48; >20% κ = 0.45; ≤5: >10% κ = 0.528; and >20% κ = 0.695).

Conclusion

Overall agreement between automated and the standard used visual scoring is low regardless of the lesion load threshold used. However, dichotomized scoring achieved more comparable results. Varying lesion load thresholds had a critical impact on patient selection by ASPECTS. Of note, the relatively low lesion volume and lack of patients with large artery occlusion in our cohort may limit generalizability of these findings.