RESUMO
BACKGROUND. An increase in frequency of acute ischemic strokes has been observed among patients presenting with acute neurologic symptoms during the coronavirus disease (COVID-19) pandemic. OBJECTIVE. The purpose of this study was to investigate the association between COVID-19 and stroke subtypes in patients presenting with acute neurologic symptoms. METHODS. This retrospective case-control study included patients for whom a code for stroke was activated from March 16 to April 30, 2020, at any of six New York City hospitals that are part of a single health system. Demographic data (age, sex, and race or ethnicity), COVID-19 status, stroke-related risk factors, and clinical and imaging findings pertaining to stroke were collected. Univariate and multivariate analyses were conducted to evaluate the association between COVID-19 and stroke subtypes. RESULTS. The study sample consisted of 329 patients for whom a code for stroke was activated (175 [53.2%] men, 154 [46.8%] women; mean age, 66.9 ± 14.9 [SD] years). Among the 329 patients, 35.3% (116) had acute ischemic stroke confirmed with imaging; 21.6% (71) had large vessel occlusion (LVO) stroke; and 14.6% (48) had small vessel occlusion (SVO) stroke. Among LVO strokes, the most common location was middle cerebral artery segments M1 and M2 (62.0% [44/71]). Multifocal LVOs were present in 9.9% (7/71) of LVO strokes. COVID-19 was present in 38.3% (126/329) of the patients. The 61.7% (203/329) of patients without COVID-19 formed the negative control group. Among individual stroke-related risk factors, only Hispanic ethnicity was significantly associated with COVID-19 (38.1% of patients with COVID-19 vs 20.7% of patients without COVID-19; p = 0.001). LVO was present in 31.7% of patients with COVID-19 compared with 15.3% of patients without COVID-19 (p = 0.001). SVO was present in 15.9% of patients with COVID-19 and 13.8% of patients without COVID-19 (p = 0.632). In multivariate analysis controlled for race and ethnicity, presence of COVID-19 had a significant independent association with LVO stroke (odds ratio, 2.4) compared with absence of COVID-19 (p = 0.011). CONCLUSION. COVID-19 is associated with LVO strokes but not with SVO strokes. CLINICAL IMPACT. Patients with COVID-19 presenting with acute neurologic symptoms warrant a lower threshold for suspicion of large vessel stroke, and prompt workup for large vessel stroke is recommended.
Assuntos
Arteriopatias Oclusivas/diagnóstico por imagem , Arteriopatias Oclusivas/etiologia , COVID-19/complicações , Neuroimagem/métodos , Acidente Vascular Cerebral/diagnóstico por imagem , Acidente Vascular Cerebral/etiologia , Idoso , Estudos de Casos e Controles , Angiografia Cerebral , Angiografia por Tomografia Computadorizada , Feminino , Humanos , Angiografia por Ressonância Magnética , Masculino , Cidade de Nova Iorque , Estudos Retrospectivos , Fatores de Risco , SARS-CoV-2RESUMO
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.
Assuntos
Registros Eletrônicos de Saúde , Aprendizado de Máquina , Processamento de Linguagem Natural , Radiologia/métodos , Tomografia Computadorizada por Raios X , Área Sob a Curva , Bases de Dados Factuais , Humanos , Sensibilidade e EspecificidadeRESUMO
CAR T-cell therapy (CAR T) for central nervous system lymphoma (CNSL) is a promising strategy, yet responses are frequently not durable. Bridging radiotherapy (BRT) is used for extra-cranial lymphoma where it can improve CAR T outcomes through cytoreduction of high-risk lesions. We hypothesized that BRT would achieve similar, significant cytoreduction prior to CAR T for CNSL (CNS-BRT). We identified CNSL patients with non-Hodgkin B-cell lymphoma who received CNS-BRT prior to commercial CAR T. Cytoreduction from CNS-BRT was calculated as change in lesion size prior to CAR T. Twelve patients received CNS-BRT, and the median follow up among survivors is 11.8 months (IQR: 8.5 - 21.9). Ten patients had CNSL (9 secondary, 1 primary) and 2 patients had epidural disease (evaluable for toxicity). All ten patients with CNSL had progressive disease at the time of CNS-BRT. 1/12 patients experienced grade ≥ 3 cytokine release syndrome (CRS), and 3/12 patients experienced grade ≥ 3 immune effector cell-associated neurotoxicity syndrome (ICANS). CNS-BRT achieved a 74.0% (95% confidence interval: 62.0 - 86.0) mean reduction in lesion size from baseline (p = 0.014) at a median of 12 days from BRT completion and prior to CAR T infusion. Best CNS response included 8 complete responses (CR), 1 partial response (PR), and 1 progressive disease (PD). Three patients experienced CNS relapse outside the BRT field. Preliminary data suggest CNS-BRT achieves rapid cytoreduction and is associated with a favorable CNS response and safety profile. These data support further study of BRT as a bridging modality for CNSL CAR T.
RESUMO
Virtual reality (VR) and augmented Reality (AR) are emerging technologies with the potential to revolutionize Interventional radiology (IR). These innovations offer advantages in patient care, interventional planning, and educational training by improving the visualization and navigation of medical images. Despite progress, several challenges hinder their widespread adoption, including limitations in navigation systems, cost, clinical acceptance, and technical constraints of AR/VR equipment. However, ongoing research holds promise with recent advancements such as shape-sensing needles and improved organ deformation modeling. The development of deep learning techniques, particularly for medical imaging segmentation, presents a promising avenue to address existing accuracy and precision issues. Future applications of AR/VR in IR include simulation-based training, preprocedural planning, intraprocedural guidance, and increased patient engagement. As these technologies advance, they are expected to facilitate telemedicine, enhance operational efficiency, and improve patient outcomes, marking a new frontier in interventional radiology.
Assuntos
Realidade Aumentada , Realidade Virtual , Humanos , Radiologia IntervencionistaRESUMO
OBJECTIVES: We have previously shown that pituitary cysts may affect growth hormone secretion. This study sought to determine cyst evolution during growth hormone treatment in children. METHODS: Forty-nine patients with short stature, a pituitary cyst, and at least two brain MRI scans were included. The percent of the pituitary gland occupied by the cyst (POGO) was calculated, and a cyst with a POGO of ≤15% was considered small, while a POGO >15% was considered large. RESULTS: Thirty-five cysts were small, and 14 were large. Five of the 35 small cysts grew into large cysts, while 6 of the 14 large cysts shrunk into small cysts. Of 4 cysts that fluctuated between large and small, 3 presented as large and 1 as small. Small cysts experienced greater change in cyst volume (CV) (mean=61.5%) than large cysts (mean=-0.4%). However, large cysts had a greater net change in CV (mean=44.2 mm3) than small cysts (mean=21.0 mm3). Older patients had significantly larger mean pituitary volume than younger patients (435.4 mm3 vs. 317.9 mm3) and significantly larger mean CV than younger patients (77.4 mm3 vs. 45.2 mm3), but there was no significant difference in POGO between groups. CONCLUSIONS: Pituitary cyst size can vary greatly over time. Determination of POGO over time is a useful marker for determining the possibility of a pathologic effect on pituitary function since it factors both cyst and gland volume. Large cysts should be monitored closely, given their extreme, erratic behavior.
Assuntos
Cistos do Sistema Nervoso Central , Cistos , Hormônio do Crescimento Humano , Doenças da Hipófise , Neoplasias Hipofisárias , Humanos , Criança , Hormônio do Crescimento , Hormônio do Crescimento Humano/uso terapêutico , Cistos/tratamento farmacológico , Cistos/patologia , Hipófise/diagnóstico por imagem , Hipófise/patologia , Doenças da Hipófise/tratamento farmacológico , Doenças da Hipófise/patologia , Neoplasias Hipofisárias/diagnóstico por imagem , Neoplasias Hipofisárias/tratamento farmacológico , Neoplasias Hipofisárias/patologia , Cistos do Sistema Nervoso Central/diagnóstico por imagem , Cistos do Sistema Nervoso Central/tratamento farmacológico , Imageamento por Ressonância Magnética , Estudos RetrospectivosRESUMO
We present a case of sea urchin spine arthritis (SUSA) in a 33-year-old woman who sustained penetrating trauma to the interphalangeal (IP) joint of the hallux while snorkeling in Japan. Serial radiographs and MRI were obtained over a period from 7 weeks to 10 months following injury. At 7 weeks radiographs revealed periarticular osteopenia and subtle marginal erosion, similar to the appearance of tuberculous arthritis. Over the ensuing months, radiographs and MRI documented progressive marginal and periarticular erosions with synovitis, despite preservation of cartilage space and restoration of bone mineral density. Delayed radiographs and imaging features mimic gouty arthropathy. Only the history points to the proper diagnosis, which was confirmed by histopathology, demonstrating necrobiotic granuloma with central fibrinoid necrosis following synovectomy and arthrodesis. The majority of previous case reports affected the hand, with few cases in the feet. In all, radiographic illustrations were limited and demonstrated only minimal osteolysis and periosteal reaction. No other report included MRI or serial radiographs over a long period to illustrate the natural progression of the disease.
Assuntos
Artrite/diagnóstico , Artrite/etiologia , Mordeduras e Picadas/complicações , Mordeduras e Picadas/diagnóstico , Doenças do Pé/diagnóstico , Ouriços-do-Mar , Ferimentos Penetrantes/complicações , Adulto , Animais , Feminino , Doenças do Pé/etiologia , HumanosRESUMO
BACKGROUND: Numerous case reports and case series have described brain Magnetic Resonance Imaging (MRI) findings in Coronavirus disease 2019 (COVID-19) patients with concurrent posterior reversible encephalopathy syndrome (PRES). PURPOSE: We aim to compile and analyze brain MRI findings in patients with COVID-19 disease and PRES. METHODS: PubMed and Embase were searched on April 5th, 2021 using the terms "COVID-19", "PRES", "SARS-CoV-2" for peer-reviewed publications describing brain MRI findings in patients 21 years of age or older with evidence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and PRES. RESULTS: Twenty manuscripts were included in the analysis, which included descriptions of 30 patients. The average age was 57 years old. Twenty-four patients (80%) required mechanical ventilation. On brain MRI examinations, 15 (50%) and 7 (23%) of patients exhibited superimposed foci of hemorrhage and restricted diffusion respectively. CONCLUSIONS: PRES is a potential neurological complication of COVID-19 related disease. COVID-19 patients with PRES may exhibit similar to mildly greater rates of superimposed hemorrhage compared to non-COVID-19 PRES patients.
Assuntos
COVID-19 , Síndrome da Leucoencefalopatia Posterior , Encéfalo/diagnóstico por imagem , Humanos , Imageamento por Ressonância Magnética , Pessoa de Meia-Idade , SARS-CoV-2RESUMO
4D-parathyroid CT scans have become a mainstay in the evaluation and pre-surgical planning for parathyroid adenomas. Most protocols typically rely on non-contrast images, prior to the arterial and delayed phases. Previous reports with dual-energy CT imaging have highlighted the utility of virtual non-contrast images to help reduce radiation dose while maintaining diagnostic accuracy. Herein, we report two cases of surgically proven parathyroid adenomas diagnosed with 4D-parathyroid CT scans performed on dual-layer spectral scanners, and in retrospect highlight the utility of virtual non-contrast images. To our knowledge, this report provides the first description of virtual non-contrast images from dual-layer spectral CT scanners that could aid in the diagnosis of parathyroid adenomas, confirming similar findings described with dual-energy CT scanners.
Assuntos
Neoplasias das Paratireoides , Redução da Medicação , Tomografia Computadorizada Quadridimensional/métodos , Humanos , Neoplasias das Paratireoides/diagnóstico por imagem , Neoplasias das Paratireoides/cirurgiaRESUMO
Coronavirus disease 2019 (COVID-19), a clinical manifestation of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), was declared a global pandemic by the World Health Organization on March 11, 2020. Hypercoagulable state has been described as one of the hallmarks of SARS-CoV-2 infection and has been reported to manifest as pulmonary embolisms, deep vein thrombosis, and arterial thrombosis of the abdominal small vessels. Here we present cases of arterial and venous thrombosis pertaining to the head and neck in COVID-19 patients.
Assuntos
Betacoronavirus , COVID-19 , Infecções por Coronavirus , Pneumonia Viral , Trombose Venosa , COVID-19/complicações , COVID-19/diagnóstico , Infecções por Coronavirus/epidemiologia , Humanos , Pandemias , Pneumonia Viral/epidemiologia , SARS-CoV-2 , Trombose Venosa/virologiaRESUMO
STUDY DESIGN: Cross-sectional database study. OBJECTIVE: The objective of this study was to develop an algorithm for the automated measurement of spinopelvic parameters on lateral lumbar radiographs with comparable accuracy to surgeons. SUMMARY OF BACKGROUND DATA: Sagittal alignment measurements are important for the evaluation of spinal disorders. Manual measurement methods are time-consuming and subject to rater-dependent error. Thus, a need exists to develop automated methods for obtaining sagittal measurements. Previous studies of automated measurement have been limited in accuracy, inapplicable to common plain films, or unable to measure pelvic parameters. METHODS: Images from 816 patients receiving lateral lumbar radiographs were collected sequentially and used to develop a convolutional neural network (CNN) segmentation algorithm. A total of 653 (80%) of these radiographs were used to train and validate the CNN. This CNN was combined with a computer vision algorithm to create a pipeline for the fully automated measurement of spinopelvic parameters from lateral lumbar radiographs. The remaining 163 (20%) of radiographs were used to test this pipeline. Forty radiographs were selected from the test set and manually measured by three surgeons for comparison. RESULTS: The CNN achieved an area under the receiver-operating curve of 0.956. Algorithm measurements of L1-S1 cobb angle, pelvic incidence, pelvic tilt, and sacral slope were not significantly different from surgeon measurement. In comparison to criterion standard measurement, the algorithm performed with a similar mean absolute difference to spine surgeons for L1-S1 Cobb angle (4.30°â±â4.14° vs. 4.99°â±â5.34°), pelvic tilt (2.14°â±â6.29° vs. 1.58°â±â5.97°), pelvic incidence (4.56°â±â5.40° vs. 3.74°â±â2.89°), and sacral slope (4.76°â±â6.93° vs. 4.75°â±â5.71°). CONCLUSION: This algorithm measures spinopelvic parameters on lateral lumbar radiographs with comparable accuracy to surgeons. The algorithm could be used to streamline clinical workflow or perform large scale studies of spinopelvic parameters.Level of Evidence: 3.
Assuntos
Aprendizado Profundo , Vértebras Lombares/diagnóstico por imagem , Região Lombossacral/diagnóstico por imagem , Interpretação de Imagem Radiográfica Assistida por Computador/métodos , Radiografia/métodos , Algoritmos , HumanosRESUMO
BACKGROUND: Authors have noticed an increase in lung apex abnormalities on CT angiography (CTA) of the head and neck performed for stroke workup during the coronavirus disease 2019 (COVID-19) pandemic. OBJECTIVE: To evaluate the incidence of these CTA findings and their relation to COVID-19 infection. METHODS: In this retrospective multicenter institutional review board-approved study, assessment was made of CTA findings of code patients who had a stroke between March 16 and April 5, 2020 at six hospitals across New York City. Demographic data, comorbidities, COVID-19 status, and neurological findings were collected. Assessment of COVID-19 related lung findings on CTA was made blinded to COVID-19 status. Incidence rates of COVID-19 related apical findings were assessed in all code patients who had a stroke and in patients with a stroke confirmed by imaging. RESULTS: The cohort consisted of a total of 118 patients with mean±SD age of 64.9±15.7 years and 57.6% (68/118) were male. Among all code patients who had a stroke, 28% (33/118) had COVID-19 related lung findings. RT-PCR was positive for COVID-19 in 93.9% (31/33) of these patients with apical CTA findings.Among patients who had a stroke confirmed by imaging, 37.5% (18/48) had COVID-19 related apical findings. RT-PCR was positive for COVID-19 in all (18/18) of these patients with apical findings. CONCLUSION: The incidence of COVID-19 related lung findings in stroke CTA scans was 28% in all code patients who had a stroke and 37.5% in patients with a stroke confirmed by imaging. Stroke teams should closely assess the lung apices during this COVID-19 pandemic as CTA findings may be the first indicator of COVID-19 infection.
Assuntos
Betacoronavirus , Infecções por Coronavirus/epidemiologia , Pneumopatias/epidemiologia , Pandemias , Pneumonia Viral/epidemiologia , Acidente Vascular Cerebral/epidemiologia , Adulto , Idoso , Idoso de 80 Anos ou mais , COVID-19 , Estudos de Coortes , Infecções por Coronavirus/diagnóstico por imagem , Feminino , Humanos , Incidência , Pneumopatias/diagnóstico por imagem , Masculino , Pessoa de Meia-Idade , Pneumonia Viral/diagnóstico por imagem , Estudos Retrospectivos , SARS-CoV-2 , Acidente Vascular Cerebral/diagnóstico por imagem , Tomografia Computadorizada por Raios X/métodosRESUMO
The spine and spinal cord are composed of multiple segments initiated by different embryologic mechanisms and advanced under different systems of control. In humans, the upper central nervous system is formed by primary neurulation, the lower by secondary neurulation, and the intervening segment by junctional neurulation. This article focuses on the distal spine and spinal cord to address their embryogenesis and the molecular derangements that lead to some distal spinal malformations.
Assuntos
Desenvolvimento Embrionário , Coluna Vertebral/anormalidades , Coluna Vertebral/anatomia & histologia , Humanos , Coluna Vertebral/embriologiaRESUMO
BACKGROUND: Differentiating glioblastoma, brain metastasis, and central nervous system lymphoma (CNSL) on conventional magnetic resonance imaging (MRI) can present a diagnostic dilemma due to the potential for overlapping imaging features. We investigate whether machine learning evaluation of multimodal MRI can reliably differentiate these entities. METHODS: Preoperative brain MRI including diffusion weighted imaging (DWI), dynamic contrast enhanced (DCE), and dynamic susceptibility contrast (DSC) perfusion in patients with glioblastoma, lymphoma, or metastasis were retrospectively reviewed. Perfusion maps (rCBV, rCBF), permeability maps (K-trans, Kep, Vp, Ve), ADC, T1C+ and T2/FLAIR images were coregistered and two separate volumes of interest (VOIs) were obtained from the enhancing tumor and non-enhancing T2 hyperintense (NET2) regions. The tumor volumes obtained from these VOIs were utilized for supervised training of support vector classifier (SVC) and multilayer perceptron (MLP) models. Validation of the trained models was performed on unlabeled cases using the leave-one-subject-out method. Head-to-head and multiclass models were created. Accuracies of the multiclass models were compared against two human interpreters reviewing conventional and diffusion-weighted MR images. RESULTS: Twenty-six patients enrolled with histopathologically-proven glioblastoma (n=9), metastasis (n=9), and CNS lymphoma (n=8) were included. The trained multiclass ML models discriminated the three pathologic classes with a maximum accuracy of 69.2% accuracy (18 out of 26; kappa 0.540, P=0.01) using an MLP trained with the VpNET2 tumor volumes. Human readers achieved 65.4% (17 out of 26) and 80.8% (21 out of 26) accuracies, respectively. Using the MLP VpNET2 model as a computer-aided diagnosis (CADx) for cases in which the human reviewers disagreed with each other on the diagnosis resulted in correct diagnoses in 5 (19.2%) additional cases. CONCLUSIONS: Our trained multiclass MLP using VpNET2 can differentiate glioblastoma, brain metastasis, and CNS lymphoma with modest diagnostic accuracy and provides approximately 19% increase in diagnostic yield when added to routine human interpretation.
RESUMO
OBJECTIVE: Predicting vision recovery following surgical decompression of the optic chiasm in pituitary adenoma patients remains a clinical challenge, as there is significant variability in postoperative visual function that remains unreliably explained by current prognostic factors. Available literature inadequately characterizes alterations in adenoma patients involving the lateral geniculate nucleus (LGN). This study examined the association of LGN degeneration with chiasmatic compression as well as with the retinal nerve fiber layer (RNFL), pattern standard deviation (PSD), mean deviation (MD), and postoperative vision recovery. PSD is the degree of difference between the measured visual field pattern and the normal pattern ("hill") of vision, and MD is the average of the difference from the age-adjusted normal value. METHODS: A prospective study of 27 pituitary adenoma patients and 27 matched healthy controls was conducted. Participants were scanned on a 7T ultra-high field MRI scanner, and 3 independent readers measured the LGN at its maximum cross-sectional area on coronal T1-weighted MPRAGE imaging. Readers were blinded to diagnosis and to each other's measurements. Neuro-ophthalmological data, including RNFL thickness, MD, and PSD, were acquired for 12 patients, and postoperative visual function data were collected on patients who underwent surgical chiasmal decompression. LGN areas were compared using two-tailed t-tests. RESULTS: The average LGN cross-sectional area of adenoma patients was significantly smaller than that of controls (13.8 vs 19.2 mm2, p < 0.0001). The average LGN cross-sectional area correlated with MD (r = 0.67, p = 0.04), PSD (r = -0.62, p = 0.02), and RNFL thickness (r = 0.75, p = 0.02). The LGN cross-sectional area in adenoma patients with chiasm compression was 26.6% smaller than in patients without compression (p = 0.009). The average tumor volume was 7902.7 mm3. Patients with preoperative vision impairment showed 29.4% smaller LGN cross-sectional areas than patients without deficits (p = 0.003). Patients who experienced improved postoperative vision had LGN cross-sectional areas that were 40.8% larger than those of patients without postoperative improvement (p = 0.007). CONCLUSIONS: The authors demonstrate novel in vivo evidence of LGN volume loss in pituitary adenoma patients and correlate imaging results with neuro-ophthalmology findings and postoperative vision recovery. Morphometric changes to the LGN may reflect anterograde transsynaptic degeneration. These findings indicate that LGN degeneration may be a marker of optic apparatus injury from chiasm compression, and measurement of LGN volume loss may be useful in predicting vision recovery following adenoma resection.
RESUMO
Rapid diagnosis and treatment of acute neurological illnesses such as stroke, hemorrhage, and hydrocephalus are critical to achieving positive outcomes and preserving neurologic function-'time is brain'1-5. Although these disorders are often recognizable by their symptoms, the critical means of their diagnosis is rapid imaging6-10. Computer-aided surveillance of acute neurologic events in cranial imaging has the potential to triage radiology workflow, thus decreasing time to treatment and improving outcomes. Substantial clinical work has focused on computer-assisted diagnosis (CAD), whereas technical work in volumetric image analysis has focused primarily on segmentation. 3D convolutional neural networks (3D-CNNs) have primarily been used for supervised classification on 3D modeling and light detection and ranging (LiDAR) data11-15. Here, we demonstrate a 3D-CNN architecture that performs weakly supervised classification to screen head CT images for acute neurologic events. Features were automatically learned from a clinical radiology dataset comprising 37,236 head CTs and were annotated with a semisupervised natural-language processing (NLP) framework16. We demonstrate the effectiveness of our approach to triage radiology workflow and accelerate the time to diagnosis from minutes to seconds through a randomized, double-blinded, prospective trial in a simulated clinical environment.