Machine learning for clinical outcome prediction in cerebrovascular and endovascular neurosurgery: systematic review and meta-analysis.

BACKGROUND: Machine learning (ML) may be superior to traditional methods for clinical outcome prediction. We sought to systematically review the literature on ML for clinical outcome prediction in cerebrovascular and endovascular neurosurgery. METHODS: A comprehensive literature search was performed, and original studies of patients undergoing cerebrovascular surgeries or endovascular procedures that developed a supervised ML model to predict a postoperative outcome or complication were included. RESULTS: A total of 60 studies predicting 71 outcomes were included. Most cohorts were derived from single institutions (66.7%). The studies included stroke (32), subarachnoid hemorrhage ((SAH) 16), unruptured aneurysm (7), arteriovenous malformation (4), and cavernous malformation (1). Random forest was the best performing model in 12 studies (20%) followed by XGBoost (13.3%). Among 42 studies in which the ML model was compared with a standard statistical model, ML was superior in 33 (78.6%). Of 10 studies in which the ML model was compared with a non-ML clinical prediction model, ML was superior in nine (90%). External validation was performed in 10 studies (16.7%). In studies predicting functional outcome after mechanical thrombectomy the pooled area under the receiver operator characteristics curve (AUROC) of the test set performances was 0.84 (95% CI 0.79 to 0.88). For studies predicting outcomes after SAH, the pooled AUROCs for functional outcomes and delayed cerebral ischemia were 0.89 (95% CI 0.76 to 0.95) and 0.90 (95% CI 0.66 to 0.98), respectively. CONCLUSION: ML performs favorably for clinical outcome prediction in cerebrovascular and endovascular neurosurgery. However, multicenter studies with external validation are needed to ensure the generalizability of these findings.

Prevalence of unruptured intracranial aneurysms (UIAs) examined in the trauma population.

BACKGROUND: Previous data on the prevalence of unruptured intracranial aneurysms (UIAs) vary widely, and studies based on these data are plagued with unintentional bias. Accurate prevalence data are paramount for any physician who counsels patients with intracranial aneurysms on rupture risk and treatment. We therefore sought to determine a more accurate number for the true prevalence of UIAs. METHODS: A retrospective chart review was conducted at a level 1 trauma center and tertiary care hospital in an urban setting between 2019 and 2020. Inclusion criteria included patients admitted with blunt trauma. Exclusion criteria included not having a head and neck CTA performed and read by an attending radiologist. All head and neck CTA radiology reads were reviewed for incidentally discovered UIAs. Subgroup analysis was performed by age group, race, and gender. RESULTS: A total of 5978 out of 8999 patients met the inclusion criteria, and 54 patients with 58 total aneurysms were identified giving an overall prevalence of 0.9%. Subgroup analysis was performed for all age groups, genders, and racial groups. CONCLUSION: The overall aneurysm prevalence was found to be 0.9% in this sample. This rate is lower than rates previously cited in the literature and those quoted in local practice. This finding has significant implications when attempting to understand average rupture risk. Further studies are needed to power more subgroup analyses to use a more personalized approach to understanding an individual's risk of rupture.

Acute and Long-Term Management of Blunt Cerebrovascular Injury at a Quaternary Referral Level 1 Trauma Center: The Memphis Experience and Comparison of a New Scoring System.

BACKGROUND AND OBJECTIVES: The management of blunt cerebrovascular injuries (BCVIs) remains an important topic within trauma and neurosurgery today. There remains a lack of consensus within the literature and significant variation across institutions. The purpose of this study was to evaluate management of BCVI at a large, tertiary referral trauma center. METHODS: Institutional Review Board approval was obtained to conduct a retrospective review of patients with BCVI at our Level 1 Trauma Center. Computed tomography angiography was used to identify BCVI for each patient. Patient information was collected, and statistical analysis was performed. With the included risk factors for ischemic complications, a novel scoring system based on ischemic risk, the "Memphis Score," was developed and evaluated to grade BCVI. RESULTS: Two hundred seventeen patients with BCVI from July 2020 to August 2022 were identified. The most common mechanism of injury was motor vehicle collision (141, 65.0%). Vertebral arteries were the most common vessel injured (136, 51.1%) with most injuries occurring at a high cervical location (101, 38.0%). Denver Grade 1 injuries (89, 33.5%) and a Memphis Score of 1 were most frequent (172, 64.6%), and initial anticoagulation with heparin drip was initiated 56.7% of the time (123). Endovascular treatment was required in 24 patients (11.1%) and was usually performed in the first 48 hours (15, 62.5%). While Denver Grade (P = .019) and Memphis Score (P < .00001) were significantly higher in those patients undergoing endovascular treatment, only the Memphis Score demonstrated a significant difference between those patients who had stroke or worsening on follow-up imaging and those who did not (P = .0009). CONCLUSION: Although BCVI management has improved since early investigative efforts, institutions must evaluate and share their data to help clarify outcomes. The novel "Memphis Score" presents a standardized framework to communicate ischemic risk and guide management of BCVI.

Automated emergent large vessel occlusion detection by artificial intelligence improves stroke workflow in a hub and spoke stroke system of care.

BACKGROUND: Emergent large vessel occlusion (ELVO) acute ischemic stroke is a time-sensitive disease. OBJECTIVE: To describe our experience with artificial intelligence (AI) for automated ELVO detection and its impact on stroke workflow. METHODS: We conducted a retrospective chart review of code stroke cases in which VizAI was used for automated ELVO detection. Patients with ELVO identified by VizAI were compared with patients with ELVO identified by usual care. Details of treatment, CT angiography (CTA) interpretation by blinded neuroradiologists, and stroke workflow metrics were collected. Univariate statistical comparisons and linear regression analysis were performed to quantify time savings for stroke metrics. RESULTS: Six hundred and eighty consecutive code strokes were evaluated by AI; 104 patients were diagnosed with ELVO during the study period. Forty-five patients with ELVO were identified by AI and 59 by usual care. Sixty-nine mechanical thrombectomies were performed.Median time from CTA to team notification was shorter for AI ELVOs (7 vs 26 min; p<0.001). Door to arterial puncture was faster for transfer patients with ELVO detected by AI versus usual care transfer patients (141 vs 185 min; p=0.027). AI yielded a time savings of 22 min for team notification and a 23 min reduction in door to arterial puncture for transfer patients. CONCLUSIONS: AI automated alerts can be incorporated into a comprehensive stroke center hub and spoke system of care. The use of AI to detect ELVO improves clinically meaningful stroke workflow metrics, resulting in faster treatment times for mechanical thrombectomy.

Predicting the degree of difficulty of the trans-radial approach in cerebral angiography.

BACKGROUND: To evaluate anatomical and clinical factors that make trans-radial cerebral angiography more difficult. METHODS: A total of 52 trans-radial diagnostic angiograms were evaluated in a tertiary care stroke center from December 2019 until March 2020. We analyzed a number of anatomical variables to evaluate for correlation to outcome measures of angiography difficulty. RESULTS: The presence of a proximal radial loop had a higher conversion to femoral access (p<0.03). The presence of a large diameter aortic arch (p<0.01), double subclavian innominate curve (p<0.01), left proximal common carotid artery (CCA) loop (p<0.001), acute subclavian vertebral angle (p<0.01), and absence of bovine aortic arch anatomy (p=0.03) were associated with more difficult trans-radial cerebral angiography and increased fluoroscopy time-per-vessel. CONCLUSION: The presence of a proximal radial loop, large diameter aortic arch, double subclavian innominate curve, proximal left CCA loop, acute subclavian vertebral angle, and absence of bovine aortic arch anatomy were associated with more difficult trans-radial cerebral angiography. We also introduce a novel grading scale for diagnostic trans-radial angiography.

Intravenous thrombolysis pretreatment and other predictors of infarct in a new previously unaffected territory (INT) in ELVO strokes treated with mechanical thrombectomy.

INTRODUCTION: One uncommon complication of mechanical thrombectomy (MT) is an infarct in a new previously unaffected territory (infarct in new territory (INT)). OBJECTIVE: To evaluate the predictors of INT with special focus on intravenous thrombolysis(IVT)pretreatmentbefore MT. METHODS: Consecutive patients with emergent large vessel occlusion (ELVO) treated with MT during a 5-year period were evaluated. INT was defined using standardized methodology proposed by ESCAPE investigators. The predictors of INT and its impact on outcomes were investigated. RESULTS: A total of 419 consecutive patients with ELVO received MT (mean age 64±15 years, 50% men, median baseline National Institutes of Health Stroke Scale score 16 points (IQR 11-20), 69% pretreated with IVT). The incidence of INT was lower in patients treated with combination therapy (IVTandMT) than in patients treated with MT alone, respectively (10% vs 20%; p=0.011). The INT group had more patients with posterior circulation occlusions than the group without INT (28% vs 10%, respectively; p<0.001). The rates of 3-month functional independence were lower in patients with INT (30% vs 50%; p=0.007). IVT pretreatment was not independently related to INT (OR=0.75; 95% CI 0.32 to 1.76), and INT did not emerge as an independent predictor of 3-month functional independence (OR=0.69; 95% CI 0.29 to 1.62) on multivariable logistic regression models. Location of posterior circulation occlusion was independently associated with a higher odds of INT (OR=3.33; 95% CI 1.43 to 7.69; p=0.005). CONCLUSIONS: IVT pretreatment is not independently associated with a lower likelihood of INT in patients with ELVO treated with MT. Patients with ELVO with posterior circulation occlusion are more likely to have INT after MT.

Impact of pretreatment with intravenous thrombolysis on reperfusion status in acute strokes treated with mechanical thrombectomy.

INTRODUCTION: We sought to evaluate the impact of pretreatment with intravenous thrombolysis (IVT) on the rate and speed of successful reperfusion (SR) in patients with emergent large vessel occlusion (ELVO) treated with mechanical thrombectomy (MT) in a high-volume tertiary care stroke center. METHODS: Consecutive patients with ELVO treated with MT were evaluated. Outcomes were compared between patients who underwent combined IVT and MT (IVT+MT) and those treated with direct MT (dMT). The elapsed time between groin puncture to beginning of reperfusion (GPTBRT) and the numbers of device passes required to achieve SR were also documented. RESULTS: A total of 287 and 132 patients were treated with IVT+MT and dMT, respectively. The IVT+MT group had higher SR (73.8% vs 62.9%; p=0.023) and 3-month functional independence (modified Rankin Scale score 0-2;51.6% vs 38.2%; p=0.008) rates. The median GPTBRT was shorter in the IVT+MT group (48 (IQR 33-70) vs 70 (IQR 44-98) min; p<0.001). Among patients who achieved SR (n=292), the median number of required device passes was lower in the IVT+MT subgroup (1 (IQR 1-1) vs 2 (IQR 1-2); p<0.001), while the rate of patients requiring ≤2 device passes was higher (98% vs 77%; p<0.001). IVT+MT was independently related to higher odds of SR (OR 1.64; 95% CI 1.03 to 2.61; p=0.036) and shorter GPTBRT (unstandardized linear regression coefficient -20.39; 95% CI -27.56 to -13.22; p<0.001) on multivariable analyses adjusting for potential confounders. Among patients with SR, IVT+MT was independently associated with a higher likelihood of ≤2 device passes (OR 14.63; 95% CI 4.46 to 48.00; p<0.001). CONCLUSIONS: IVT pretreatment appears to increase the rates of SR and shortens the duration of the endovascular procedure by requiring fewer device passes in patients with ELVO treated with MT.