The University of California San Francisco Brain Metastases Stereotactic Radiosurgery (UCSF-BMSR) MRI Dataset.

Supplemental material is available for this article.

Magnetic Resonance Imaging Prognostic Findings for Visual and Mortality Outcomes in Acute Invasive Fungal Rhinosinusitis.

PURPOSE: To identify initial, preintervention magnetic resonance imaging (MRI) findings that are predictive of visual and mortality outcomes in acute invasive fungal rhinosinusitis (AIFRS). DESIGN: Retrospective cohort study. PARTICIPANTS: Patients with histopathologically or microbiologically confirmed AIFRS cared for at a single, tertiary academic institution between January 2000 and February 2020. METHODS: A retrospective review of MRI scans and clinical records of patients with confirmed diagnosis of AIFRS was performed. For each radiologic characteristic, a modified Poisson regression with robust standard errors was used to estimate the risk ratio for blindness. A multivariate Cox proportional hazards model was used to study AIFRS-specific risk factors associated with mortality. MAIN OUTCOME MEASURE: Identification of initial, preintervention MRI findings associated with visual and mortality outcomes. RESULTS: The study comprised 78 patients (93 orbits, 63 with unilateral disease and 15 with bilateral disease) with AIFRS. The leading causes of immunosuppression were hematologic malignancy (38%) and diabetes mellitus (36%). Mucormycota constituted 56% of infections, and Ascomycota constituted 37%. The overall death rate resulting from infection was 38%. Risk factors for poor visual acuity outcomes on initial MRI included involvement of the orbital apex (relative risk [RR], 2.0; 95% confidence interval [CI], 1.1-3.8; P = 0.026) and cerebral arteries (RR, 1.8; 95% CI, 1.3-2.5; P < 0.001). Increased mortality was associated with involvement of the facial soft tissues (hazard ratio [HR], 4.9; 95% CI, 1.3-18.2; P = 0.017), nasolacrimal drainage apparatus (HR, 5.0; 95% CI, 1.5-16.1; P = 0.008), and intracranial space (HR, 3.5; 95% CI, 1.4-8.6; P = 0.006). Orbital soft tissue involvement was associated with decreased mortality (HR, 0.3; 95% CI, 0.1-0.6; P = 0.001). CONCLUSIONS: Extrasinonasal involvement in AIFRS typically signals advanced infection with the facial soft tissues most commonly affected. The initial, preintervention MRI is prognostic for a poor visual acuity outcome when orbital apex or cerebral arterial involvement, or both, are present. Facial soft tissues, nasolacrimal drainage apparatus, intracranial involvement, or a combination thereof is associated with increased mortality risk, whereas orbital soft tissue involvement is correlated with a reduced risk of mortality.

Development and web deployment of an automated neuroradiology MRI protocoling tool with natural language processing.

BACKGROUND: A systematic approach to MRI protocol assignment is essential for the efficient delivery of safe patient care. Advances in natural language processing (NLP) allow for the development of accurate automated protocol assignment. We aim to develop, evaluate, and deploy an NLP model that automates protocol assignment, given the clinician indication text. METHODS: We collected 7139 spine MRI protocols (routine or contrast) and 990 head MRI protocols (routine brain, contrast brain, or other) from a single institution. Protocols were split into training (n = 4997 for spine MRI; n = 839 for head MRI), validation (n = 1071 for spine MRI, fivefold cross-validation used for head MRI), and test (n = 1071 for spine MRI; n = 151 for head MRI) sets. fastText and XGBoost were used to develop 2 NLP models to classify spine and head MRI protocols, respectively. A Flask-based web app was developed to be deployed via Heroku. RESULTS: The spine MRI model had an accuracy of 83.38% and a receiver operator characteristic area under the curve (ROC-AUC) of 0.8873. The head MRI model had an accuracy of 85.43% with a routine brain protocol ROC-AUC of 0.9463 and contrast brain protocol ROC-AUC of 0.9284. Cancer, infectious, and inflammatory related keywords were associated with contrast administration. Structural anatomic abnormalities and stroke/altered mental status were indicative of routine spine and brain MRI, respectively. Error analysis revealed increasing the sample size may improve performance for head MRI protocols. A web version of the model is provided for demonstration and deployment. CONCLUSION: We developed and web-deployed two NLP models that accurately predict spine and head MRI protocol assignment, which could improve radiology workflow efficiency.

Three-dimensional U-Net Convolutional Neural Network for Detection and Segmentation of Intracranial Metastases.

PURPOSE: To develop and validate a neural network for automated detection and segmentation of intracranial metastases on brain MRI studies obtained for stereotactic radiosurgery treatment planning. MATERIALS AND METHODS: In this retrospective study, 413 patients (average age, 61 years ± 12 [standard deviation]; 238 women) with a total of 5202 intracranial metastases (median volume, 0.05 cm3; interquartile range, 0.02-0.18 cm3) undergoing stereotactic radiosurgery at one institution were included (January 2017 to February 2020). A total of 563 MRI examinations were performed among the patients, and studies were split into training (n = 413), validation (n = 50), and test (n = 100) datasets. A three-dimensional (3D) U-Net convolutional network was trained and validated on 413 T1 postcontrast or subtraction scans, and several loss functions were evaluated. After model validation, 100 discrete test patients, who underwent imaging after the training and validation patients, were used for final model evaluation. Performance for detection and segmentation of metastases was evaluated using Dice scores, false discovery rates, and false-negative rates, and a comparison with neuroradiologist interrater reliability was performed. RESULTS: The median Dice score for segmenting enhancing metastases in the test set was 0.75 (interquartile range, 0.63-0.84). There were strong correlations between manually segmented and predicted metastasis volumes (r = 0.98, P < .001) and between the number of manually segmented and predicted metastases (R = 0.95, P < .001). Higher Dice scores were strongly correlated with larger metastasis volumes on a logarithmically transformed scale (r = 0.71). Sensitivity across the whole test sample was 70.0% overall and 96.4% for metastases larger than 6 mm. There was an average of 0.46 false-positive results per scan, with the positive predictive value being 91.5%. In comparison, the median Dice score between two neuroradiologists was 0.85 (interquartile range, 0.80-0.89), with sensitivity across the test sample being 87.9% overall and 98.4% for metastases larger than 6 mm. CONCLUSION: A 3D U-Net-based convolutional neural network was able to segment brain metastases with high accuracy and perform detection at the level of human interrater reliability for metastases larger than 6 mm.Keywords: Adults, Brain/Brain Stem, CNS, Feature detection, MR-Imaging, Neural Networks, Neuro-Oncology, Quantification, Segmentation© RSNA, 2021.

Uncommon malignant renal tumors and atypical presentation of common ones: a guide for radiologists.

OBJECTIVE: While the typical imaging features of the more common RCC subtypes have previously been described, they can at times have unusual, but distinguishing features. Rarer renal tumors span a broad range of imaging features, but they may also have characteristic presentations. We review the key imaging features of atypical presentations of malignant renal tumors and uncommon malignant renal tumors. CONCLUSION: Renal tumors have many different presentation patterns, but knowledge of the distinguishing MR and CT features can help identify both atypical presentation of common malignancies and uncommon renal tumors.

Paraneoplastic neurologic syndrome as a presentation of underlying breast malignancy.

While breast cancer most commonly presents as a screen-detected mammographic finding or a breast symptom, in very rare instances it may first present as a paraneoplastic neurologic syndrome (PNS; Surg Case Rep, 2015;1:59; Ann Neurol 2004;56:715). Fewer than 1% of breast cancer patients have PNS, and an even smaller percentage initially present with neurologic symptoms (J Neurol Neurosurg Psychiatry, 2004;75:ii43). We report a case series of three patients who presented with neurological disorders suspicious for PNS, and were subsequently found to have underlying breast cancer. We follow this with a discussion of key clinical features of management considerations in paraneoplastic syndromes secondary to breast malignancy.

Torsion of a giant mesocolic lipoma in a child with Bannayan-Riley-Ruvalcaba syndrome.

A 6-year-old boy with Bannayan-Riley-Ruvalcaba syndrome (BRRS) presented to the emergency department with periumbilical abdominal pain for 12 h. A contrast-enhanced abdominal and pelvis CT examination revealed significant interval change in the size and appearance of a previously seen hypoattenuating right mesocolic fatty mass suggestive for lipoma, first observed 5 months prior. This lesion demonstrated new enhancing internal septations, a thickened capsule, interval development of adjacent mesenteric fat stranding and engorgement of the mesenteric vessels. Given the short follow-up interval and acute clinical presentation, imaging findings were suggestive for torsion. We present this case for the unusual imaging findings as well as to highlight the differential diagnosis for abdominal fat containing lesions by imaging in patients with BRRS and other hamartomatous syndromes.

Adjuvant immunotherapy for non-small cell lung cancer.

Non-small cell lung cancer (NSCLC) is the biggest cancer killer in the United States and worldwide. In 2011, there are estimated to be 221,130 new cases of lung cancer in the United States. Over a million people will die of lung cancer worldwide this year alone. When possible, surgery to remove the tumor is the best treatment strategy for patients with NSCLC. However, even with adjuvant (postoperative) chemotherapy and radiation, more than 40% of patients will develop recurrences locally or systemically and ultimately succumb to their disease. Thus, there is an urgent need for developing superior approaches to treat patients who undergo surgery for NSCLC to eliminate residual disease that is likely responsible for these recurrences. Our group and others have been interested in using immunotherapy to augment the efficacy of current treatment strategies. Immunotherapy is very effective against minimal disease burden and small deposits of tumor cells that are accessible by the circulating immune cells. Therefore, this strategy may be ideally suited as an adjunct to surgery to seek and destroy microscopic tumor deposits that remain after surgery. This review describes the mechanistic underpinnings of immunotherapy and how it is currently being used to target residual disease and prevent postoperative recurrences after pulmonary resection in NSCLC.