Faculty Bibliography

Purpose To compare different methods for generating features from radiology reports and to develop a method to automatically identify findings in these reports. Materials and Methods In this study, 96 303 head computed tomography (CT) reports were obtained. The linguistic complexity of these reports was compared with that of alternative corpora. Head CT reports were preprocessed, and machine-analyzable features were constructed by using bag-of-words (BOW), word embedding, and Latent Dirichlet allocation-based approaches. Ultimately, 1004 head CT reports were manually labeled for findings of interest by physicians, and a subset of these were deemed critical findings. Lasso logistic regression was used to train models for physician-assigned labels on 602 of 1004 head CT reports (60%) using the constructed features, and the performance of these models was validated on a held-out 402 of 1004 reports (40%). Models were scored by area under the receiver operating characteristic curve (AUC), and aggregate AUC statistics were reported for (a) all labels, (b) critical labels, and (c) the presence of any critical finding in a report. Sensitivity, specificity, accuracy, and F1 score were reported for the best performing model's (a) predictions of all labels and (b) identification of reports containing critical findings. Results The best-performing model (BOW with unigrams, bigrams, and trigrams plus average word embeddings vector) had a held-out AUC of 0.966 for identifying the presence of any critical head CT finding and an average 0.957 AUC across all head CT findings. Sensitivity and specificity for identifying the presence of any critical finding were 92.59% (175 of 189) and 89.67% (191 of 213), respectively. Average sensitivity and specificity across all findings were 90.25% (1898 of 2103) and 91.72% (18 351 of 20 007), respectively. Simpler BOW methods achieved results competitive with those of more sophisticated approaches, with an average AUC for presence of any critical finding of 0.951 for unigram BOW versus 0.966 for the best-performing model. The Yule I of the head CT corpus was 34, markedly lower than that of the Reuters corpus (at 103) or I2B2 discharge summaries (at 271), indicating lower linguistic complexity. Conclusion Automated methods can be used to identify findings in radiology reports. The success of this approach benefits from the standardized language of these reports. With this method, a large labeled corpus can be generated for applications such as deep learning. ^© RSNA, 2018 Online supplemental material is available for this article.

Stroke is a leading cause of long-term disability, and outcome is directly related to timely intervention. Not all patients benefit from rapid intervention, however. Thus a significant amount of attention has been paid to using neuroimaging to assess potential benefit by identifying areas of ischemia that have not yet experienced cellular death. The perfusion-diffusion mismatch, is used as a simple metric for potential benefit with timely intervention, yet penumbral patterns provide an inaccurate predictor of clinical outcome. Machine learning research in the form of deep learning (artificial intelligence) techniques using deep neural networks (DNNs) excel at working with complex inputs. The key areas where deep learning may be imminently applied to stroke management are image segmentation, automated featurization (radiomics), and multimodal prognostication. The application of convolutional neural networks, the family of DNN architectures designed to work with images, to stroke imaging data is a perfect match between a mature deep learning technique and a data type that is naturally suited to benefit from deep learning's strengths. These powerful tools have opened up exciting opportunities for data-driven stroke management for acute intervention and for guiding prognosis. Deep learning techniques are useful for the speed and power of results they can deliver and will become an increasingly standard tool in the modern stroke specialist's arsenal for delivering personalized medicine to patients with ischemic stroke.

STUDY DESIGN:Retrospective analysis of prospectively collected data. OBJECTIVE:The aim of this study was to determine the ability of abnormal coagulation profile to predict adverse events following posterior cervical laminectomy and fusion (PCLF). SUMMARY OF BACKGROUND DATA:PCLF is an increasingly common procedure used to treat a variety of traumatic and degenerative spinal conditions. Abnormal coagulation profile is associated with postoperative adverse events, including blood transfusion. There is a paucity of literature that specifically addresses the relationship between coagulation profile and complications following PCLF. METHODS:ACS-NSQIP was utilized to identify patients undergoing PCLF between 2006 and 2013. A total of 3546 patients met inclusion criteria. Multivariate analysis was utilized to identify associations between abnormal coagulation profile and postoperative complications. RESULTS:Membership in the low-platelet cohort was an independent risk factor for myocardial infarction (Odds Ratio (OR) = 5.4 [1.0, 29.1], P = 0.049) and bleeding transfusion (OR = 2.0 [1.2, 3.4], P = 0.011). Membership in the high international normalized ratio group was an independent risk factor for pneumonia (OR = 6.3 [2.5, 16.1], P < 0.001), ventilation >48 hours (OR = 6.5 [2.3, 18.4], P < 0.001), organ space surgical site infection (OR = 11.1 [2.1, 57.3], P = 0.004), urinary tract infection (OR = 3.0 [1.2, 8.0], P = 0.024), bleeding transfusion (OR = 6.0 [3.4, 10.7], P < 0.001), sepsis (OR = 5.1 [1.6, 16.4], P = 0.006), and septic shock (OR = 6.8 [1.7, 27.4], P = 0.007). Membership in the bleeding disorders cohort was an independent predictor of unplanned intubation (OR = 3.2 [1.1, 9.5], P = 0.041), pneumonia (OR = 2.9 [1.2, 7.2], P = 0.023), ventilation >48 hours (OR = 4.8 [1.9, 12.4], P = 0.001), cerebrovascular accident/stroke with neurological deficit (OR = 24.8 [2.9, 210.6], P = 0.003), bleeding transfusion (OR = 2.1 [1.1, 4.1], P = 0.032), reoperation (OR = 3.6 [1.4, 9.3], P = 0.008), and sepsis (OR = 3.4 [1.1, 10.4], P = 0.031). CONCLUSION:This is the first large study to document abnormal coagulation profile as an independent predictor of outcomes following PCLF. Abnormal coagulation profile represents a predictor of complications that can be medically mitigated, and is therefore a valuable parameter to assess preoperatively. Coagulation profile should continue to play a role in targeting patients for risk stratification, preoperative optimization, and quality improvement initiatives. LEVEL OF EVIDENCE:3.

BACKGROUND:The incidence of meningioma has increased drastically recently, particularly in older adults. Surgical intervention has the potential to reduce neurologic symptoms and achieve favorable, long-term outcomes. There is considerable variability in the literature examining the relationship between age and outcomes after meningioma surgery. The objective of this study was to identify the relationship between age and postoperative complications after craniotomy for resection of meningioma. METHODS:The American College of Surgeons National Surgical Quality Improvement Program database was used to identify patients undergoing craniotomy for meningioma resection between 2005 and 2012. Multivariate analysis was used to identify associations between age and postoperative complications. RESULTS:Age >80 years is an independent risk factor for any complication (odds ratio [OR], 2.374; 95% confidence interval [CI], 1.3-4.4; P = 0.015), death within 30 days of surgery (OR, 15.7; 95% CI, 3.0-81.0; P < 0.001), and length of stay >5 days (OR, 3.2; 95% CI, 1.8-5.6; P < 0.001). CONCLUSIONS:Advanced age, particularly >80 years, is an independent predictor of morbidity and mortality in patients undergoing craniotomy for resection of meningioma. As such, it should be considered in preoperative optimization and risk stratification.

BACKGROUND:There is no facile quantitative method for monitoring hydrocephalus (HCP). We propose quantitative computed tomography (CT) ventriculography (qCTV) as a novel computer vision tool for empirically assessing HCP in patients with subarachnoid hemorrhage (SAH). METHODS:Twenty patients with SAH who were evaluated for ventriculoperitoneal shunt (VPS) placement were selected for inclusion. Ten patients with normal head computed tomography (CTH) findings were analyzed as negative controls. CTH scans were segmented both manually and automatically (by qCTV) to generate measures of ventricular volume. RESULTS:identified all patients who went on to require VPS placement (10 of 10; P = 0.011). CONCLUSIONS:qCTV is a reliable means of quantifying ventricular volume and hydrocephalus. This technique offers a new tool for monitoring neurosurgical patients for hydrocephalus, and may be beneficial for use in future research studies, as well as in the routine care of patients with hydrocephalus.

INTRODUCTION/BACKGROUND:ADAPT (a direct aspiration first pass technique) has been shown to be fast, cost-effective, and associated with excellent angiographic and clinical outcomes in the treatment of acute ischemic stroke (AIS). OBJECTIVE:To identify any and all preoperative factors that are associated with successful revascularization using aspiration alone. METHODS:A retrospective review of 76 patients with AIS treated with thrombectomy was carried out. Cohort 1 included cases in which aspiration alone was successful (Thrombolysis in Cerebral Infarction 2b or 3). Cohort 2 included cases in which aspiration was unsuccessful or could not be performed despite an attempt. RESULTS:There was no difference between cohorts in gender, race, medications, National Institute of Health Stroke Scale score, IV tissue plasminogen activator, site or side of the occlusion, dense vessel sign, aortic arch type, severe stenosis, clot length, operator years of experience, and guide/aspiration catheters used. Patients in cohort 1 were on average younger (66.5 vs 74.1 years, p=0.025). There was a trend for more patients in cohort 2 to have atrial fibrillation/arrhythmias (62.5% vs 45.5%, p=0.168) and have a cardiogenic stroke etiology (78.1% vs 56.8%, p=0.086). There was also a trend for more reverse curves (2.3 vs 1.7, p=0.107), larger vessel diameter (3.26 mm vs 2.88 mm, p=0.184), larger vessel-to-catheter ratio (2.09 vs 1.87, p=0.192), and worse clot burden score (5.38 vs 6.68, p=0.104) in cohort 2. CONCLUSIONS:Aspiration success was associated with younger age. Our findings suggest that ADAPT can be used for the vast majority of patients but it may be beneficial to use a different method first in the elderly.

STUDY DESIGN/METHODS:A retrospective cohort study. OBJECTIVE:The aim of this study was to identify associations between abnormal coagulation profile and postoperative morbidity and mortality in patients undergoing posterior lumbar fusion (PLF). SUMMARY OF BACKGROUND DATA/BACKGROUND:The literature suggests that abnormal coagulation profile is associated with postoperative complications, notably the need for blood transfusion. However, there is little research that directly addresses the influence of coagulation profile on postoperative complications following PLF. METHODS:The American College of Surgeons National Surgical Quality Improvement Program database (ACS-NSQIP) was utilized to identify patients undergoing PLF between 2006 and 2013. Nine thousand two hundred ninety-five patients met inclusion criteria. Multivariate analysis was utilized to identify associations between abnormal coagulation profile and postoperative complications. RESULTS:Low platelet count was an independent risk factor for organ space surgical site infections (SSIs) [odds ratio (OR) = 6.0, P < 0.001], ventilation >48 hours (OR = 4.5, P = 0.002), Acute renal failure (OR = 5.8, P = 0.007), transfusion (OR = 1.6, P < 0.001), sepsis (OR = 2.2, P = 0.037), reoperation (OR = 2.5, P = 0.001), and death (OR = 3.7, P = 0.049). High partial thromboplastin time (PTT) was an independent risk factor for ventilation >48 hours (OR = 5.6, P = 0.002), cerebrovascular accident (CVA)/stroke with neurological deficit (OR = 5.1, P = 0.011), cardiac arrest (OR = 5.4, P = 0.030), transfusion (OR = 1.5, P = 0.020), and death (OR = 4.5, P = 0.050). High International Normalized Ration (INR) was an independent risk factor for pneumonia (OR = 8.7, P = 0.001), pulmonary embolism (OR = 5.6, P = 0.021), deep venous thrombosis/Thrombophlebitis (OR = 4.8, P = 0.011), septic shock (OR = 8.4, P = 0.048), and death (OR = 9.8, P = 0.034). Bleeding disorder was an independent risk factor for organ space SSI (OR = 5.4, P = 0.01), pneumonia (OR = 3.0, P = 0.023), and sepsis (OR = 4.4, P < 0.001). CONCLUSION/CONCLUSIONS:Abnormal coagulation profile was an independent predictor of morbidity and mortality in patients undergoing PLF. As such, it should be considered in preoperative optimization and risk stratification.

Engagement in research and academic productivity are crucial components in the training of a neurosurgeon. This process typically begins in residency training. In this study, we analyzed individual resident productivity as it correlated to publications across all Accreditation Council for Graduate Medical Education (ACGME)-accredited neurosurgery training programs in an attempt to identify how programs have developed and fostered a research culture and environment. We obtained a list of current neurosurgery residents in ACGME-accredited programs from the American Association of Neurological Surgeons database. An expanded PubMed and Scopus search was conducted for each resident through the present time. We tabulated all articles attributed to each resident. We then categorized the publications based on each neurosurgical subspecialty while in residency. A spreadsheet-based statistical analysis was performed. This formulated the average number of resident articles, h-indices, and most common subspecialty categories by training program. We analyzed 1352 current neurosurgery residents in 105 programs. There were a total of 10 645 publications, of which 3985 were resident first-author publications during the period of study. The most common subspecialties among all resident publications were vascular (24.9%), spine (16.9%), oncology (16.1%), pediatric (5.6%), functional (4.9%), and trauma (3.8%). The average resident published 2.9 first-author papers with average of 38.0 first-author publications by total residents at each program (range 0-241). The average h-index per resident is 2.47 ± 3.25. When comparing previously published faculty h-index program rankings against our resident h-index rankings, there is a strong correlation between the 2 datasets with a clear delineation between Top-20 productivity and that of other programs (average h-index 4.2 vs 1.7, respectively, P < .001). Increasing program size leads to a clear increase in academic productivity on both the resident and faculty level (average h-index 1.6, 1.9, 3.9 for 1, 2, and 3 resident per year programs, respectively, P < .001). Resident first-author publications correlated with recently described academic departmental productivity. Subspecialty resident publications are highest in cerebrovascular surgery. Resident research and publication is a key metric for assessing the productivity of academic neurosurgery programs and is consistent with one of the core foci of neurosurgical training.

BACKGROUND: Melanoma has high propensity to metastasize to the brain. With recent gains in improving patient survival, stereotactic radiosurgery may offer an effective and less neurotoxic alternative to whole brain radiation. In the study, we report on the safety and efficacy of stereotactic radiosurgery (SRS) in treating melanoma brain metastases in 87 patients. METHODS: This retrospective, multicenter study examined 87 patients with 309 metastases underwent single-dose or multifractionated SRS for treatment of intracranial metastases from malignant melanoma. RESULTS: A total of eighty-seven patients with a median age of 62 years (26 to 85) were treated from 2007 to 2014. Eighty patients (92%) also had extracranial metastases at time of treatment, and sixty-nine patients (79%) had uncontrolled systemic disease. 79 patients (91%) underwent single-dose, 7 (8%) underwent 3 fractions, and 1 (1%) were treated in 5 fractions. Mean tumor volume (GTV) treated was 1.92 cc. Five patients developed symptoms of acute phase toxicity, and four developed late phase toxicity. None had radionecrosis. The median survival was 6 months. The Kaplan-Meier 1-year survival rate was 31%, and 1-year local control rate was 91%. 1-year survival rate for BRAF positive patients was 42%, and BRAF negative patients was 27%. Forty-two patients (48%) had distant intracranial recurrences, and 1-year distant control rate was 32%. CONCLUSIONS: SRS is a safe and effective treatment option for intracranial metastases from malignant melanoma. This paper serves as a reference for what is achievable in the absence of highly effective systemic therapy.

Meningiomas are common intracranial tumors that can have a complex arterial supply. In a patient with a recurrent large left clinoid aypical grade 2 meningioma, diagnostic angiography revealed left proximal middle cerebral artery (MCA) occlusion and reconstitution of the MCA circulation from the left middle meningeal artery. Consequently, the planned surgical approach was adjusted to preserve the left external carotid artery circulation in a difficult meningioma resection. This case illustrates a role for preoperative angiography in surgical planning of patients with complex skull base tumors involving intracranial vascular structures and highlights the unique neovascularization that often occurs as meningiomas parasitize the exracranial blood supply.