Retinal and optic nerve magnetic resonance diffusion-weighted imaging in acute non-arteritic central retinal artery occlusion.

OBJECTIVES: Diffusion weighted imaging hyperintensity (DWI-H) has been described in the retina and optic nerve during acute central retinal artery occlusion (CRAO). We aimed to determine whether DWI-H can be accurately identified on standard brain magnetic resonance imaging (MRI) in non-arteritic CRAO patients at two tertiary academic centers. MATERIALS AND METHODS: Retrospective cross-sectional study that included all consecutive adult patients with confirmed acute non-arteritic CRAO and brain MRI performed within 14 days of CRAO. At each center, two neuroradiologists masked to patient clinical data reviewed each MRI for DWI-H in the retina and optic nerve, first independently then together. Statistical analysis for inter-rater reliability and correlation with clinical data was performed. RESULTS: We included 204 patients [mean age 67.9±14.6 years; 47.5% females; median time from CRAO to MRI 1 day (IQR 1-4.3); 1.5 T in 127/204 (62.3%) and 3.0 T in 77/204 (37.7%)]. Inter-rater reliability varied between centers (κ = 0.27 vs. κ = 0.65) and was better for retinal DWI-H. Miss and error rates significantly differed between neuroradiologists at each center. After consensus review, DWI-H was identified in 87/204 (42.6%) patients [miss rate 117/204 (57.4%) and error rate 11/87 (12.6%)]. Significantly more patients without DWI-H had good visual acuity at follow-up (p = 0.038). CONCLUSIONS: In this real-world case series, differences in agreement and interpretation accuracy among neuroradiologists limited the role of DWI-H in diagnosing acute CRAO on standard MRI. DWI-H was identified in 42.6% of patients and was more accurately detected in the retina than in the optic nerve. Further studies are needed with standardized novel MRI protocols.

Stroke Mortality Outcomes in Uganda.

BACKGROUND AND PURPOSE: Stroke outcome data in Uganda is lacking. The objective of this study was to capture 30-day mortality outcomes in patients presenting with acute and subacute stroke to Mbarara Regional Referral Hospital (MRRH) in Uganda. METHODS: A prospective study enrolling consecutive adults presenting to MRRH with abrupt onset of focal neurologic deficits suspicious for stroke, from August 2014 to March 2015. All patients had head computed tomography (CT) confirmation of ischemic or hemorrhagic stroke. Data was collected on mortality, morbidity, risk factors, and imaging characteristics. RESULTS: Investigators screened 134 potential subjects and enrolled 108 patients. Sixty-two percent had ischemic and 38% hemorrhagic stroke. The mean age of all patients was 62.5 (SD 17.4), and 52% were female. More patients had hypertension in the hemorrhagic stroke group than in the ischemic stroke group (53% vs. 32%, p = 0.0376). Thirty-day mortality was 38.1% (p = 0.0472), and significant risk factors were National Institutes of Health Stroke Scale (NIHSS) score, female sex, anemia, and HIV infection. A one unit increase of the NIHSS on admission increased the risk of death at 30 days by 6%. Patients with hemorrhagic stroke had statistically higher NIHSS scores (p = 0.0408) on admission compared to patients with ischemic stroke, and also had statistically higher Modified Rankin Scale (mRS) scores at discharge (p = 0.0063), and mRS score change from baseline (p = 0.04). CONCLUSIONS: Our study highlights an overall 30-day stroke mortality of 38.1% in southwestern Uganda, and identifies NIHSS at admission, female sex, anemia, and HIV infection as predictors of mortality.

Updated Imaging Features of Dysplastic Cerebellar Gangliocytoma.

OBJECTIVE: The aim of this study was to perform an updated review of the imaging features of dysplastic cerebellar gangliocytoma (DCG). METHODS: Imaging findings were retrospectively reviewed in 14 patients with DCG. The analysis included size, location, cyst formation, calcification, intralesional hemorrhage, enhancement pattern, and apparent diffusion coefficient (ADC). RESULTS: In addition to revisiting many well-established imaging features of DCG, enhancement was much more common (64.3%) than previously reported, and small enhancing veins were also frequently encountered within or along the periphery of the lesions (50%). Dysplastic cerebellar gangliocytomas had an elevated ADC compared with normal cerebellum (967.8 ± 115.7 vs 770.4 ± 47.3 × 10 mm/s; P < 0.0001). CONCLUSIONS: Enhancement on magnetic resonance imaging within DCG should be an accepted imaging finding rather than being viewed as uncommon or atypical. Dysplastic cerebellar gangliocytomas typically have an elevated ADC compared with normal cerebellum, which may assist in differentiation from other posterior fossa neoplasms.

Computed Tomography-Based Texture Analysis to Determine Human Papillomavirus Status of Oropharyngeal Squamous Cell Carcinoma.

OBJECTIVE: To determine whether machine learning can accurately classify human papillomavirus (HPV) status of oropharyngeal squamous cell carcinoma (OPSCC) using computed tomography (CT)-based texture analysis. METHODS: Texture analyses were retrospectively applied to regions of interest from OPSCC primary tumors on contrast-enhanced neck CT, and machine learning was used to create a model that classified HPV status with the highest accuracy. Results were compared against the blinded review of 2 neuroradiologists. RESULTS: The HPV-positive (n = 92) and -negative (n = 15) cohorts were well matched clinically. Neuroradiologist classification accuracies for HPV status (44.9%, 55.1%) were not significantly different (P = 0.13), and there was a lack of agreement between the 2 neuroradiologists (κ = -0.145). The best machine learning model had an accuracy of 75.7%, which was greater than either neuroradiologist (P < 0.001, P = 0.002). CONCLUSIONS: Useful diagnostic information regarding HPV infection can be extracted from the CT appearance of OPSCC beyond what is apparent to the trained human eye.

Hypoglossal Canal Lesions: Distinctive Imaging Features and Simple Diagnostic Algorithm.

OBJECTIVE: The objective of this study is to compare the prevalence of hypoglossal canal lesions and identify differentiating imaging features. MATERIALS AND METHODS: A 15-year retrospective review of lesions of the hypoglossal nerve and hypoglossal canal, excluding those in patients with metastasis or prior head and neck cancer and those treated with radiation or surgery, was performed. Clinical findings and lesion imaging features were documented. The contrast-enhanced T1-weighted nonenhancing cystic component of hypoglossal schwannomas was compared with size-matched pathologically proven vestibular schwannomas. RESULTS: Our review identified 40 lesions encroaching on the hypoglossal canal, located within the hypoglossal canal, or found in both locations: 16 hypoglossal schwannomas, 15 juxtaarticular cysts, and nine purely cystic hypoglossal canal lesions. Hypoglossal schwannomas, which were intradural when involving the premedullary cistern, most commonly enhanced heterogeneously, with a thick peripheral enhancing component, a central enhancing component, or both enhancing components. Juxtaarticular cysts encroaching on the hypoglossal canal were extradural, were contiguous with a craniocervical junction synovial joint, and showed thin rim enhancement. Hypoglossal canal nonenhancing cystic lesions were confined to the hypoglossal canal, had signal intensity equivalent to CSF, and lacked any contrast enhancement. CONCLUSION: Although hypoglossal schwannomas, juxtaarticular cysts, and hypoglossal canal nonenhancing cystic lesions may all encroach on or lie within the hypoglossal canal, purely cystic lesions are unlikely to be schwannomas. A diagnostic algorithm applying key imaging and clinical findings allows differentiation of these lesions, which have vastly different treatment regimens.

Median Lingual Lymph Nodes: Prevalence on Imaging and Potential Implications for Oral Cavity Cancer Staging.

OBJECTIVE: This study sought to estimate the prevalence of median lingual lymph node (MLLN) metastases from oral cavity squamous cell carcinoma (OCSCC) and determine the frequency with which MLLNs can be identified with magnetic resonance imaging (MRI) in control subjects. METHODS: Pathology reports were used to identify patients with surgically treated OCSCC who underwent preoperative positron emission tomography-computed tomography to define the prevalence of MLLN metastases. As a control group, 500 consecutive face-neck MRIs from noncancer patients were reviewed for structures consistent with MLLNs. RESULTS: In the study group, 1 (0.95%) of 105 OCSCC cases demonstrated a single MLLN metastasis from a lateral tongue tumor (T4aN2c). The MLLN exceeded 1 cm in all planes and was abnormal in morphology. The frequency of suspected MLLNs in controls was 1.0%, with a maximum measurement of 0.9 cm. CONCLUSIONS: Median lingual lymph nodes are infrequently identified with MRI in controls, concordant with the low prevalence of metastases from OCSCC to this inconstant nodal group.

Post-pubertal developmental trajectories of laryngeal shape and size in humans.

Laryngeal morphotypes have been hypothesized related to both phonation and to laryngeal pathologies. Morphotypes have not been validated or demonstrated quantitatively and sources of shape and size variation are incompletely understood but are critical for the explanation of behavioral changes (e.g., changes of physical properties of a voice) and for therapeutic approaches to the larynx. This is the first study to take this crucial step and results are likely to have implications for surgeons and speech language pathologists. A stratified human sample was interrogated for phenotypic variation of the vocal organ. First, computed tomography image stacks were used to generate three-dimensional reconstructions of the thyroid cartilage. Then cartilage shapes were quantified using multivariate statistical analysis of high dimensional shape data from margins and surfaces of the thyroid cartilage. The effects of sex, age, body mass index (BMI) and body height on size and shape differences were analyzed. We found that sex, age, BMI and the age-sex interaction showed significant effects on the mixed sex sample. Among males, only age showed a strong effect. The thyroid cartilage increased in overall size, and the angulation between left and right lamina decreased in older males. Age, BMI and the age-height interaction were statistically significant factors within females. The angulation between left and right lamina increased in older females and was smaller in females with greater BMI. A cluster analysis confirmed the strong age effect on larynx shape in males and a complex interaction between the age, BMI and height variables in the female sample. The investigation demonstrated that age and BMI, two risk factors in a range of clinical conditions, are associated with shape and size variation of the human larynx. The effects influence shape differently in female and male larynges. The male-female shape dichotomy is partly size-dependent but predominantly size-independent.

Classifying the Large Vestibular Aqueduct: Morphometry to Audiometry.

OBJECTIVE: Large vestibular aqueduct (LVA) is the most common inner ear dysplasia identified in patients with hearing loss. Our objective was to systematically quantify LVA morphologies and correlate imaging findings with established audiometric outcomes. STUDY DESIGN: Retrospective review. SETTING: Tertiary referral center. PATIENTS: Patients with large vestibular aqueduct identified radiographically, with or without hearing loss. INTERVENTIONS: Diagnostic only. MAIN OUTCOME MEASURES: Vestibular aqueduct (VA) width at midpoint, width at external aperture, and length were measured on cross-sectional imaging. Morphology was classified as type I (borderline), type II (tubular), or type III (funneled). Audiometric endpoints included air/bone conduction, pure tone averages, and air-bone gaps at 250 and 500 Hz. Statistical associations were evaluated using linear regression models, adjusted for age at first audiogram and sex. RESULTS: One hundred seventeen patients (197 ears) were included, with mean age at first audiogram of 22.2 years (standard deviation, 21.7 yr). Imaging features associated with poor audiometric outcomes were increasing VA width at midpoint and external aperture, decreasing VA length, dilated extraosseous endolymphatic sac, cochleovestibular malformations, and increasing VA type (III > II > I). CONCLUSIONS: Quantitative LVA measurements and a standardized morphologic classification system aid in prediction of early audiometric endpoints.

CT interpretation in a telestroke network: agreement among a spoke radiologist, hub vascular neurologist, and hub neuroradiologist.

BACKGROUND AND PURPOSE: The American Stroke Association guidelines emphasized the need for further high-quality studies that assess agreement by radiologists and nonradiologists engaged in emergency telestroke assessments and decision-making. Therefore, the objective of this study was to determine the level of agreement of baseline brain CT scan interpretations of patients with acute stroke presenting to telestroke spoke hospitals between central reading committee neuroradiologists and each of 2 groups, spoke hospital radiologists and hub hospital vascular neurologists (telestrokologists). METHODS: The Stroke Team Remote Evaluation Using a Digital Observation Camera Arizona trial was a prospective, urban single-hub, rural 2-spoke, randomized, blinded, controlled trial of a 2-way, site-independent, audiovisual telemedicine and teleradiology system designed for remote evaluation of adult patients with acute stroke versus telephone consultation to assess eligibility for treatment with intravenous thrombolysis. In the telemedicine arm, the subjects' CT scans were interpreted by the hub telestrokologist and in the telephone arm by the spoke radiologist. All subjects' CT scans were subsequently interpreted centrally, independently, and blindly by 2 hub neuroradiologists. The primary CT outcome was determination of a CT-based contraindication to thrombolytic treatment. Kappa statistics and exact agreement rates were used to analyze interobserver agreement. RESULTS: Fifty-four subjects underwent random assignment. The overall agreement for the presence of radiological contraindications to thrombolysis was excellent (0.91) and did not differ substantially between the hub telestrokologist to neuroradiologist and spoke radiologist to neuroradiologist (0.92 and 0.89, respectively). CONCLUSIONS: In the context of a telestroke network designed to assess patients with acute stroke syndromes, agreement over the presence or absence of radiological contraindications to thrombolysis was excellent whether the comparisons were between a telestrokologist and neuroradiologist or between spoke radiologist and neuroradiologist. CLINICAL TRIAL REGISTRATION: URL: http://www.clinicaltrials.gov. Unique identifier: NCT00623350.

The role of digital subtraction myelography in the diagnosis and localization of spontaneous spinal CSF leaks.

OBJECTIVE: The objective of our study was to review the clinical utility of digital subtraction myelography for the diagnosis of spinal CSF leaks in patients with spontaneous intracranial hypotension (SIH) and those with superficial siderosis. MATERIALS AND METHODS: Procedure logs from 2007 to 2011 were reviewed to identify cases in which digital subtraction myelography was performed to diagnose spinal CSF leaks. Electronic medical records were reviewed to obtain information regarding diagnosis and outcome. For patients to be included in the study, preprocedural spinal MRI had to show an extradural fluid collection spanning more than one vertebral level and postmyelographic CT had to confirm the presence of an active CSF leak. If digital subtraction myelography successfully showed the site of the CSF leak, the location was documented. RESULTS: Eleven patients (seven men and four women; mean age, 49.0 years) underwent digital subtraction myelography during the study period. Six patients had SIH and five patients had superficial siderosis. The extradural fluid collection on spinal MRI averaged a length of 15.5 vertebral levels. Digital subtraction myelography successfully showed the site of the CSF leak in nine of the 11 patients, and all of the dural tears were located in the thoracic spine between T3 and T11. CONCLUSION: Digital subtraction myelography is a valuable diagnostic tool for the localization of rapid spinal CSF leaks and should be considered in patients who are clinically suspected to have a dural tear that is accompanied by a longitudinally extensive extradural fluid collection on spinal MRI.

Reversible Cerebral Vasoconstriction Syndrome in the Setting of COVID-19 and Pleomorphic Sarcoma: A Case Report.

INTRODUCTION: Reversible cerebral vasonstriction syndrome (RCVS) is an increasingly recognized clinical and radiologic syndrome. However, it has been rarely reported in the setting of the novel coronavirus disease-2019 (COVID-19) infection or sarcomatous tumors. RCVS might be the initial manifestations of COVID-19 infection or noncatecholamine producing masses including sarcoma. CASE REPORT: A 44-year-old male who developed COVID-19-related symptoms followed by rapid onset of severe headaches in the setting of persistently elevated blood pressure (BP). Brain imaging showed multifocal arterial narrowing in the anterior and posterior circulation consistent with RCVS. Serial imaging demonstrated resolution of the arterial narrowing after BP control was achieved with improvement in the patient's headaches. Further investigation for secondary causes of the patient's elevated BP revealed a right renal mass, and the patient underwent right nephrectomy, and the biopsy results confirmed the diagnosis of pleomorphic sarcoma. CONCLUSION: Our case suggests a possible association between severe acute respiratory syndrome coronavirus 2 with development of RCVS, but further studies are needed to validate this observation, establish a causal relationship and define a pathophysiological mechanism. Considering tumors other than catecholamine-producing masses as a potential risk factor for developing RCVS might lead to earlier detection and treatment of any underlying malignancy in patients whom the main and sole presentation could be RCVS.

Myeloid Sarcoma of the Temporal Bone: A Unique Cause of Hearing Loss, Otalgia, and Facial Nerve Weakness.

OBJECTIVE: To characterize a clinical triad of symptoms associated with myeloid sarcomas of the temporal bone via a review of all previously reported cases. METHODS: Case report and Ovid MEDLINE database literature review. RESULTS: A literature search revealed that a clinical triad of hearing loss, otalgia, and facial nerve weakness are commonly associated with this rare presentation of myeloid sarcoma in the temporal bone. 44% (18/41) of patients presented with all three symptoms, while 76% (31/41) presented with at least two. The presence of t(8;21) was reported in nine patients with myeloid sarcomas of the temporal bone. CONCLUSIONS: Although myeloid sarcomas are exceedingly rare, it is necessary to consider them as part of the differential diagnosis for patients who might present with middle ear and mastoid opacification on computed tomography (CT) scan, hearing loss, otalgia, and facial nerve palsy. Physicians should maintain a high degree of suspicion in patients with a history of acute myelogenous leukemia (AML), especially if previous cytogenetic analysis revealed a t(8;21).

Incidence of Concomitant Semicircular Canal Dehiscence With Otosclerosis.

Objective: The concurrence of otosclerosis and superior semicircular canal dehiscence (SSCD) presents a diagnostic challenge and failure to differentiate between these 2 diagnoses results in mischaracterization and unsuccessful surgery. The objective of this study is to identify the incidence of SSCD in patients who have computed tomography (CT) evidence of otosclerosis. Study Design: Retrospective chart review. Setting: Tertiary referral hospital. Patients: Adults with CT scan of the temporal bone diagnosed with radiological unilateral or bilateral fenestral otosclerosis from January 1995 to April 2018. Methods: Retrospective review of patient imaging from a multi-center tertiary-referral health system from January 1995 to April 2018. Imaging was reviewed to quantify the incidence of SSCD among patients with CT-diagnosed bilateral fenestral otosclerosis. Poor quality imaging was excluded from review. Results: One-thousand two-hundred eight patients (1214 CT scans) were identified with otosclerosis, of which 373 were diagnosed with fenestral otosclerosis (663 ears) with imaging of sufficient quality for review. This population was predominantly female (57.2%) with bilateral fenestral otosclerosis (78%). Of these, 23 ears (3.5%) had definitive evidence of SSCD, with an additional 15 ears (2.3%) with possible radiographic evidence of SSCD. There was no significant difference in laterality between the SSCD and otosclerosis. Conclusions: Among 373 patients with fenestral otosclerosis per CT temporal bone imaging at a tertiary referral hospital, as many as 8.3% of patients had radiographic evidence of SSCD. Given this incidence, it continues to be important to consider SSCD when diagnosing and treating otosclerosis.

Determination of posterolateral oropharyngeal wall thickness and the potential implications for transoral surgical margins in tonsil cancer.

BACKGROUND: Margins in transoral surgery for tonsil cancer can be limited by oropharyngeal wall thickness (OWT), but the normal range is not well established. METHODS: In 240 noncancer subjects, OWT was measured bilaterally in the vicinity of the tonsils with MRI. Statistical analysis was performed to assess for interaction of age, sex, location, and obesity. RESULTS: Mean(SD) OWT measured 3.4(0.6) mm posteriorly, 3.7(2.0) mm between the styloglossus and stylopharyngeus, and 5.3(0.8) mm laterally. OWT was greater in men, correlated with obesity, decreased posteriorly and laterally in the 60-80 versus 40-59 year age groups, and increased when styloglossus/stylopharyngeus were closer. OWT was <5 mm in 36.7%-97.9% of locations, with the largest percentage below this threshold located posteriorly. CONCLUSIONS: OWT is frequently <5 mm, particularly in the posterior and intermuscular areas, suggesting that a smaller surgical margin may need to be accepted in transoral tonsil cancer surgery for anatomic reasons.

Uncertainty quantification in the radiogenomics modeling of EGFR amplification in glioblastoma.

Radiogenomics uses machine-learning (ML) to directly connect the morphologic and physiological appearance of tumors on clinical imaging with underlying genomic features. Despite extensive growth in the area of radiogenomics across many cancers, and its potential role in advancing clinical decision making, no published studies have directly addressed uncertainty in these model predictions. We developed a radiogenomics ML model to quantify uncertainty using transductive Gaussian Processes (GP) and a unique dataset of 95 image-localized biopsies with spatially matched MRI from 25 untreated Glioblastoma (GBM) patients. The model generated predictions for regional EGFR amplification status (a common and important target in GBM) to resolve the intratumoral genetic heterogeneity across each individual tumor-a key factor for future personalized therapeutic paradigms. The model used probability distributions for each sample prediction to quantify uncertainty, and used transductive learning to reduce the overall uncertainty. We compared predictive accuracy and uncertainty of the transductive learning GP model against a standard GP model using leave-one-patient-out cross validation. Additionally, we used a separate dataset containing 24 image-localized biopsies from 7 high-grade glioma patients to validate the model. Predictive uncertainty informed the likelihood of achieving an accurate sample prediction. When stratifying predictions based on uncertainty, we observed substantially higher performance in the group cohort (75% accuracy, n = 95) and amongst sample predictions with the lowest uncertainty (83% accuracy, n = 72) compared to predictions with higher uncertainty (48% accuracy, n = 23), due largely to data interpolation (rather than extrapolation). On the separate validation set, our model achieved 78% accuracy amongst the sample predictions with lowest uncertainty. We present a novel approach to quantify radiogenomics uncertainty to enhance model performance and clinical interpretability. This should help integrate more reliable radiogenomics models for improved medical decision-making.

Imaging of PTEN-related abnormalities in the central nervous system.

The phosphatase and tensin homolog (PTEN) located at 10q23.31 is a tumor suppressor gene expressed ubiquitously, and loss of function mutations lead to aberrant growth, angiogenesis, and an increased risk for a variety of tumors. PTEN mutations have been associated with multiple abnormalities in the central nervous system, and a number of clinical phenotypes are now attributed to germline PTEN mutations, collectively referred to as PTEN hamartoma tumor syndrome (PHTS). Most notably, these include Cowden syndrome (CS), Bannayan-Riley-Ruvalcaba syndrome (BRRS), and autism spectrum disorders with macrocephaly. It is important to recognize the neuroimaging features associated with PTEN mutations to not only avoid misdiagnosis in cases of known PHTS but also to guide genetic testing in patients who do not yet have an established diagnosis. In this review, the central nervous system imaging features of PTEN-related disorders are discussed.

Multiple reader comparison of 2D TOF, 3D TOF, and CEMRA in screening of the carotid bifurcations: Time to reconsider routine contrast use?

BACKGROUND AND PURPOSE: MR contrast-enhanced techniques are undergoing increased scrutiny since the FDA applied a warning for gadolinium-based MR contrast agents due to gadolinium deposition within multiple organ systems. While CE-MRA provides excellent image quality, is it required in a screening carotid study? This study compares 2D TOF and 3D TOF MRA vs. CE-MRA in defining carotid stenosis in a large clinical patient population, and with multiple readers with varying experience. MATERIALS AND METHODS: 200 consecutive patients had their carotid bifurcations evaluated with 2D TOF, 3D TOF and CE-MRA sequences by 6 board-certified neuroradiologists. Stenosis and quality of examinations were defined for each study. Inter-rater reliability was assessed using two-way random effects intraclass correlation coefficients. Intra-reader reliability was computed via weighted Cohen's κ. Weighted Cohen's κ were also computed to assess agreement in stenosis ratings between enhanced images and unenhanced images. RESULTS: Agreement between unenhanced and enhanced ratings was substantial with a pooled weighted κ of 0.733 (0.628-0.811). For 5 of the 6 readers, the combination of unenhanced 2D TOF and 3D TOF showed better agreement with contrast-enhanced than either 2D TOF or 3D TOF alone. Intra-reader reliability was substantial. CONCLUSIONS: The combination of 2D TOF and 3D TOF MRA showed substantial agreement with CE-MRA regarding degree of carotid stenosis in this large outpatient population across multiple readers of varying experience. Given the scrutiny that GBCA are undergoing due to concerns regarding CNS and soft tissue deposition, it seems prudent to reserve CE-MRA for cases which are not satisfactorily answered by the nonenhanced study or other noninvasive examinations.

Deep neural network to locate and segment brain tumors outperformed the expert technicians who created the training data.

Purpose: Deep learning (DL) algorithms have shown promising results for brain tumor segmentation in MRI. However, validation is required prior to routine clinical use. We report the first randomized and blinded comparison of DL and trained technician segmentations. Approach: We compiled a multi-institutional database of 741 pretreatment MRI exams. Each contained a postcontrast T1-weighted exam, a T2-weighted fluid-attenuated inversion recovery exam, and at least one technician-derived tumor segmentation. The database included 729 unique patients (470 males and 259 females). Of these exams, 641 were used for training the DL system, and 100 were reserved for testing. We developed a platform to enable qualitative, blinded, controlled assessment of lesion segmentations made by technicians and the DL method. On this platform, 20 neuroradiologists performed 400 side-by-side comparisons of segmentations on 100 test cases. They scored each segmentation between 0 (poor) and 10 (perfect). Agreement between segmentations from technicians and the DL method was also evaluated quantitatively using the Dice coefficient, which produces values between 0 (no overlap) and 1 (perfect overlap). Results: The neuroradiologists gave technician and DL segmentations mean scores of 6.97 and 7.31, respectively ( p < 0.00007 ). The DL method achieved a mean Dice coefficient of 0.87 on the test cases. Conclusions: This was the first objective comparison of automated and human segmentation using a blinded controlled assessment study. Our DL system learned to outperform its "human teachers" and produced output that was better, on average, than its training data.