Automated Segmentation of Intracranial Thrombus on NCCT and CTA in Patients with Acute Ischemic Stroke Using a Coarse-to-Fine Deep Learning Model.

BACKGROUND AND PURPOSE: Identifying the presence and extent of intracranial thrombi is crucial in selecting patients with acute ischemic stroke for treatment. This article aims to develop an automated approach to quantify thrombus on NCCT and CTA in patients with stroke. MATERIALS AND METHODS: A total of 499 patients with large-vessel occlusion from the Safety and Efficacy of Nerinetide in Subjects Undergoing Endovascular Thrombectomy for Stroke (ESCAPE-NA1) trial were included. All patients had thin-section NCCT and CTA images. Thrombi contoured manually were used as reference standard. A deep learning approach was developed to segment thrombi automatically. Of 499 patients, 263 and 66 patients were randomly selected to train and validate the deep learning model, respectively; the remaining 170 patients were independently used for testing. The deep learning model was quantitatively compared with the reference standard using the Dice coefficient and volumetric error. The proposed deep learning model was externally tested on 83 patients with and without large-vessel occlusion from another independent trial. RESULTS: The developed deep learning approach obtained a Dice coefficient of 70.7% (interquartile range, 58.0%-77.8%) in the internal cohort. The predicted thrombi length and volume were correlated with those of expert-contoured thrombi (r = 0.88 and 0.87, respectively; P < .001). When the derived deep learning model was applied to the external data set, the model obtained similar results in patients with large-vessel occlusion regarding the Dice coefficient (66.8%; interquartile range, 58.5%-74.6%), thrombus length (r = 0.73), and volume (r = 0.80). The model also obtained a sensitivity of 94.12% (32/34) and a specificity of 97.96% (48/49) in classifying large-vessel occlusion versus non-large-vessel occlusion. CONCLUSIONS: The proposed deep learning method can reliably detect and measure thrombi on NCCT and CTA in patients with acute ischemic stroke.

Perceived Limits of Endovascular Treatment for Secondary Medium-Vessel-Occlusion Stroke.

BACKGROUND AND PURPOSE: Thrombus embolization during mechanical thrombectomy occurs in up to 9% of cases, making secondary medium vessel occlusions of particular interest to neurointerventionalists. We sought to gain insight into the current endovascular treatment approaches for secondary medium vessel occlusion stroke in an international case-based survey because there are currently no clear recommendations for endovascular treatment in these patients. MATERIALS AND METHODS: Survey participants were presented with 3 cases involving secondary medium vessel occlusions, each consisting of 3 case vignettes with changes in the patient's neurologic status (improvement, no change, unable to assess). Multivariable logistic regression analyses clustered by the respondent's identity were used to assess factors influencing the decision to treat. RESULTS: In total, 366 physicians (56 women, 308 men, 2 undisclosed) from 44 countries provided 3294 responses to 9 scenarios. Most (54.1%, 1782/3294) were in favor of endovascular treatment. Participants were more likely to treat occlusions in the anterior M2/3 (74.3%; risk ratio = 2.62; 95% CI, 2.27-3.03) or A3 (59.7%; risk ratio = 2.11; 95% CI, 1.83-2.42) segment compared with the M3/4 segment (28.3%; reference). Physicians were less likely to pursue endovascular treatment in patients who showed neurologic improvement than in patients with an unchanged neurologic deficit (49.9% versus 57.0% responses in favor of endovascular treatment, respectively; risk ratio = 0.88, 95% CI, 0.83-0.92). Interventionalists and more experienced physicians were more likely to treat secondary medium vessel occlusions. CONCLUSIONS: Physicians' willingness to treat secondary medium vessel occlusions endovascularly is limited and varies per occlusion location and change in neurologic status. More evidence on the safety and efficacy of endovascular treatment for secondary medium vessel occlusion stroke is needed.

Patient-Relevant Deficits Dictate Endovascular Thrombectomy Decision-Making in Patients with Low NIHSS Scores with Medium-Vessel Occlusion Stroke.

BACKGROUND AND PURPOSE: There is a paucity of evidence regarding the safety of endovascular treatment for patients with acute ischemic stroke due to primary medium-vessel occlusion. The aim of this study was to examine the willingness among stroke physicians to perform endovascular treatment in patients with mild-yet-disabling deficits due to medium-vessel occlusion. MATERIALS AND METHODS: In an international cross-sectional survey consisting of 7 primary medium-vessel occlusion case scenarios, participants were asked whether the presence of personally disabling deficits would influence their decision-making for endovascular treatment despite the patients having low NIHSS scores (<6). Decision rates were calculated on the basis of physician characteristics. Univariable logistic regression clustered by respondent and scenario identity was performed. RESULTS: Three hundred sixty-six participants from 44 countries provided 2562 answers to the 7 medium-vessel occlusion scenarios included in this study. In scenarios in which the deficit was relevant to the patient's profession, 56.9% of respondents opted to perform immediate endovascular treatment compared with 41.0% when no information regarding the patient's profession was provided (risk ratio = 1.39, P < .001). The largest effect sizes were seen for female participants (risk ratio = 1.68; 95% CI, 1.35-2.09), participants older than 60 years of age (risk ratio = 1.61; 95% CI, 1.23-2.10), those with more experience in neurointervention (risk ratio = 1.60; 95% CI, 1.24-2.06), and those who personally performed >100 endovascular treatments per year (risk ratio = 1.63; 95% CI, 1.22-2.17). CONCLUSIONS: The presence of a patient-relevant deficit in low-NIHSS acute ischemic stroke due to medium-vessel occlusion is an important factor for endovascular treatment decision-making. This may have relevance for the conduct and interpretation of low-NIHSS endovascular treatment in randomized trials.

Permeability surface area product analysis in malignant brain edema prediction - A pilot study.

BACKGROUND AND PURPOSE: Using an extended CT perfusion acquisition (150s), we sought to determine the association between perfusion parameters and malignant edema after ischemic stroke. METHODS: Patients (from prospective study PROVE-IT, NCT02184936) with terminal internal carotid artery±proximal middle cerebral occlusion were involved. CTA was assessed for clot location and status of leptomeningeal collaterals. The following CTP parameters were calculated within the ischemic territory and contralaterally: permeability surface area product (PS), cerebral blood flow (CBF) and cerebral blood volume (CBV). PS was calculated using the adiabatic approximation to the Johnson and Wilson model. Outcome was evaluated by midline shift and infarction volume on follow-up imaging. RESULTS: Of 200 patients enrolled, 7 patients (3.5%) had midline shift≥5mm (2 excluded for poor-quality scans). Five patients with midline shift and 5 matched controls were analysed. There was no significant difference in mean PS, CBF and CBV within the ischemic territory between the two groups. A CBV threshold of 1.7ml/100g had the highest AUC=0.72, 95% CI=0.54-0.90 for early midline shift prediction, sensitivity and specificity were 0.83 and 0.67 respectively. CONCLUSION: Our preliminary results did not show significant differences in permeability surface area analysis if analysed for complete ischemic region. CBV parameter had the highest accuracy and there was a trend for the mean PS values for midline shift prediction.