Explore the vast range of titles available.

Breast cancer remains a global challenge, causing over 600,000 deaths in 2018 (ref. 1 ). To achieve earlier cancer detection, health organizations worldwide recommend screening mammography, which is estimated to decrease breast cancer mortality by 20–40% (refs. 2 , 3 ). Despite the clear value of screening mammography, significant false positive and false negative rates along with non-uniformities in expert reader availability leave opportunities for improving quality and access 4 , 5 . To address these limitations, there has been much recent interest in applying deep learning to mammography 6 – 18 , and these efforts have highlighted two key difficulties: obtaining large amounts of annotated training data and ensuring generalization across populations, acquisition equipment and modalities. Here we present an annotation-efficient deep learning approach that (1) achieves state-of-the-art performance in mammogram classification, (2) successfully extends to digital breast tomosynthesis (DBT; ‘3D mammography’), (3) detects cancers in clinically negative prior mammograms of patients with cancer, (4) generalizes well to a population with low screening rates and (5) outperforms five out of five full-time breast-imaging specialists with an average increase in sensitivity of 14%. By creating new ‘maximum suspicion projection’ (MSP) images from DBT data, our progressively trained, multiple-instance learning approach effectively trains on DBT exams using only breast-level labels while maintaining localization-based interpretability. Altogether, our results demonstrate promise towards software that can improve the accuracy of and access to screening mammography worldwide. A generalizable and interpretable artificial-intelligence system achieves clinical accuracy for screening and early breast-cancer detection on 2D and 3D mammograms.

The wide variety of treatment options that exist for glioblastoma, including surgery, ionizing radiation, anti-neoplastic chemotherapies, anti-angiogenic therapies, and active or passive immunotherapies, all may alter aspects of vascular permeability within the tumor and/or normal parenchyma. These alterations manifest as changes in the degree of contrast enhancement or T2-weighted signal hyperintensity on standard anatomic MRI scans, posing a potential challenge for accurate radiographic response assessment for identifying anti-tumor effects. The current review highlights the challenges that remain in differentiating true disease progression from changes due to radiation therapy, including pseudoprogression and radionecrosis, as well as immune or inflammatory changes that may occur as either an undesired result of cytotoxic therapy or as a desired consequence of immunotherapies.

•Ipilimumab is an immunomodulating drug used for treatment of metastatic melanoma.•Ipilimumab use may be associated with autoimmune lymphocytic hypophysitis.•Symptoms include headache, fatigue, vision changes, endocrinopathy, hyponatremia.•MRI findings of pituitary gland and stalk enlargement with homogeneous enhancement.•Concern for ipilimumab hypophysitis is important for early diagnosis and treatment. Ipilimumab is an immunomodulating drug for use in treatment of unresectable or metastatic melanoma with autoimmune lymphocytic hypophysitis as a reported complication. We describe three recent cases of ipilimumab associated autoimmune hypophysitis (IAH) at our institution, and provide a selected literature review showing its variable clinical presentation, imaging appearance and treatment in order to expedite early and appropriate IAH management. Patients had variable clinical presentation of hypophysitis, including headache, fatigue, visual changes, endocrinopathy, and/or hyponatremia. Contrast enhanced MRI showed symmetric pituitary gland and stalk enlargement in all of our cases and received a presumptive diagnosis of IAH. Following cessation of therapy and treatment there was normalization of pituitary morphology at follow-up MRI and return to clinical baseline. Varying clinical presentation can complicate the diagnosis of lymphocytic hypophysitis. One must be cognizant of its overall clinical and radiologic picture in patients receiving ipilimumab, now commonly used for the treatment of metastatic melanoma.

•Contemporary DSC methodology should be used to establish rCBV thresholds.•Proper data reduction approaches best evaluate rCBV-based disease status.•A rCBV threshold of 1.5 best distinguishes toxoplasmosis from lymphoma. CNS toxoplasmosis and lymphoma are often indistinguishable by conventional contrast-enhanced MRI. There is limited literature on the diagnostic efficacy of dynamic susceptibility contrast (DSC) MRI for differentiating these entities. This study assesses the clinical utility of relative cerebral blood volume (rCBV) for making a diagnosis and determines rCBV thresholds for differentiation using contemporary DSC-MRI. Thirteen patients with 25 lesions (13 toxoplasmosis and 12 lymphoma) and pre-treatment DSC-MRI were identified retrospectively. Volumetric regions of interest of segmented enhancement were used to extract mean rCBV normalized to normal-appearing white matter for each lesion. We compared average mean rCBV between all toxoplasmosis and lymphoma lesions using a general mixed model. Three models were also compared for evaluating rCBV-based disease status in each patient: 1) mean rCBV of each lesion using a generalized estimating equation, 2) volume-weighted mean rCBV, and 3) maximum mean rCBV of all lesions using logistic regression. The average mean rCBV for all toxoplasmosis lesions was 0.98 (95% CI 0.55–1.41) compared to 2.07 (95% CI 1.71–2.43) for all lymphoma lesions, a significant difference (1.09, 95% CI 0.53–1.65, p=0.0013). For the three models used to evaluate rCBV-based disease status in each patient, a significant relationship was observed, with an optimal rCBV threshold of approximately 1.5 for distinguishing lymphoma from toxoplasmosis in each model. RCBV derived from contemporary DSC-MRI is helpful for distinguishing between cerebral toxoplasmosis and cerebral lymphoma on an individual patient basis and may facilitate more timely initiation of appropriate directed therapy.

The systematic investigation of susceptibility-induced contrast in MRI is important to better interpret the influence of microvascular and microcellular morphology on DSC-MRI derived perfusion data. Recently, a novel computational approach called the Finite Perturber Method (FPM), which enables the study of susceptibility-induced contrast in MRI arising from arbitrary microvascular morphologies in 3D has been developed. However, the FPM has lower efficiency in simulating water diffusion especially for complex tissues. In this work, an improved computational approach that combines the FPM with a matrix-based finite difference method (FDM), which we call the Finite Perturber the Finite Difference Method (FPFDM), has been developed in order to efficiently investigate the influence of vascular and extravascular morphological features on susceptibility-induced transverse relaxation. The current work provides a framework for better interpreting how DSC-MRI data depend on various phenomena, including contrast agent leakage in cancerous tissues and water diffusion rates. In addition, we illustrate using simulated and micro-CT extracted tissue structures the improved FPFDM along with its potential applications and limitations.

Predictive tools for stratifying neonatal hydrocephalus into low‐ and high‐risk groups for cerebrospinal fluid (CSF) diversion are currently lacking. We developed and validated an artificial intelligence (AI) model that integrates multimodal imaging and clinical data to predict CSF diversion needs. The development cohort included 116 neonates with suspicion of raised intracranial pressure (ICP) from a Chinese tertiary referral hospital (80 with intracranial pressure > 80 mm H2O, 36 with intracranial pressure ≤ 80 mm H2O). The external validation cohort consisted of 21 neonates with hydrocephalus from an American medical center, categorized by etiology: prenatal myelomeningocele (MMC) closure (n = 5), postnatal MMC closure (n = 6), and post‐hemorrhagic hydrocephalus (PHH) (n = 10). Inclusion criteria required available MRI and complete clinical follow‐up to confirm CSF diversion outcomes. The primary outcome was the need for CSF diversion. Model performance was assessed using under the receiver operating characteristics curve (AUC), sensitivity, and specificity. The hybrid AI model achieved an AUC of 0.824 in the development cohort in predicting raised ICP, outperforming both the clinical‐only model (AUC 0.528, p < 0.001) and the image‐only model (AUC 0.685, p = 0.007). In the external validation cohort, the fused MRI‐based model achieved an AUC of 0.808. The model correctly predicted CSF diversion in 4/5 prenatal MMC, 4/6 postnatal MMC, and 9/10 PHH cases. The AI model demonstrated robust performance in predicting the need for CSF diversion, particularly in PHH cases, and has the potential to assist decision‐making, especially in settings with limited pediatric neurosurgical expertise. Future work should focus on further refining model performance for complex etiologies such as MMC‐associated hydrocephalus. Our study developed and validated an artificial intelligence (AI) model that integrates multimodal MRIs and clinical data to predict CSF diversion needs. The AI model was trained in a developmental cohort including 116 neonates with suspicion of raised intracranial pressure (ICP) and then validated in an external validation cohort consisting of 21 neonates with hydrocephalus.

Background Delirium occurs frequently in patients with stroke, but the role of preexisting neural substrates in delirium pathogenesis remains unclear. We sought to explore associations between acute and chronic neural substrates of delirium in patients with intracerebral hemorrhage (ICH). Methods Using data from a single-center ICH registry, we identified consecutive patients with acute nontraumatic ICH and available magnetic resonance imaging scans. Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition criteria were used to classify each patient as delirious or nondelirious during their hospitalization. Magnetic resonance imaging scans were processed and analyzed using semiautomated software, with volumetric measurement of acute ICH volume as well as white matter hyperintensity volume (WMHV) and gray and white matter volumes from the contralateral hemisphere. We tested associations between WMHV and incident delirium using multivariable regression models, and then determined the predictive accuracy of these neuroimaging models via area under the curve (AUC) analysis. Results Of 139 patients in our cohort (mean [standard deviation] age 67.3 [17.3] years, 53% male), 58 (42%) patients experienced delirium. In our primary analyses, WMHV was significantly associated with delirium after adjusting for ICH features (odds ratio 1.56 per 10 cm 3 , 95% confidence interval 1.13–2.13), and this association was strengthened after further adjustment for segmented brain volume in patients with high-resolution scans (odds ratio 1.89 per 10 cm 3 , 95% confidence interval 1.24–2.86). Neuroimaging-based models predicted delirium with high accuracy (AUC 0.81), especially in patients with Glasgow Coma Scale score > 13 (AUC 0.85) and smaller ICH (AUC 0.91). Conclusions Chronic white matter disease is independently associated with delirium in patients with acute ICH, and neuroimaging biomarkers may have utility in predicting delirium occurrence.

ObjectivesCT angiography (CTA) is essential in acute stroke to detect emergent large vessel occlusions (ELVO) and must be interpreted by radiologists with and without subspecialized training. Additionally, grayscale inversion has been suggested to improve diagnostic accuracy in other radiology applications. This study examines diagnostic performance in ELVO detection between neuroradiologists, non-neuroradiologists, and radiology residents using standard and grayscale inversion viewing methods.MethodsA random, counterbalanced experimental design was used, where 18 radiologists with varying experiences interpreted the same patient images with and without grayscale inversion. Confirmed positive and negative ELVO cases were randomly ordered using a balanced design. Sensitivity, specificity, positive and negative predictive values as well as confidence, subjective assessment of image quality, time to ELVO detection, and overall interpretation time were examined between grayscale inversion (on/off) by experience level using generalized mixed modeling assuming a binary, negative binomial, and binomial distributions, respectively.ResultsAll groups of radiologists had high sensitivity and specificity for ELVO detection (all > .94). Neuroradiologists were faster than non-neuroradiologists and residents in interpretation time, with a mean of 47 s to detect ELVO, as compared with 59 and 74 s, respectively. Residents were subjectively less confident than attending physicians. With respect to grayscale inversion, no differences were observed between groups with grayscale inversion vs. standard viewing for diagnostic performance (p = 0.30), detection time (p = .45), overall interpretation time (p = .97), and confidence (p = .20).ConclusionsDiagnostic performance in ELVO detection with CTA was high across all levels of radiologist training level. Grayscale inversion offered no significant detection advantage.Key Points• Stroke is an acute vascular syndrome that requires acute vascular imaging.• Proximal large vessel occlusions can be identified quickly and accurately by radiologists across all training levels.• Grayscale inversion demonstrated minimal detectable benefit in the detection of proximal large vessel occlusions.

Background: The invasion of glioblastoma cells beyond the visible tumor margin depicted by conventional neuroimaging is believed to mediate recurrence and predict poor survival. Radiomic biomarkers that are associated with the direction and extent of tumor infiltration are, however, non-existent. Methods: Patients from a single center with newly diagnosed glioblastoma (n = 7) underwent preoperative Q-space magnetic resonance imaging (QSI; 3T, 64 gradient directions, b = 1000 s/mm2) between 2018 and 2019. Tumors were manually segmented, and patterns of inter-voxel coherence spatially intersecting each segmentation were generated to represent tumor-associated tractography. One patient additionally underwent regional biopsy of diffusion tract- versus non-tract-associated tissue during tumor resection for RNA sequencing. Imaging data from this cohort were compared with a historical cohort of n = 66 glioblastoma patients who underwent similar QSI scans. Associations of tractography-derived metrics with survival were assessed using t-tests, linear regression, and Kaplan–Meier statistics. Patient-derived glioblastoma xenograft (PDX) mice generated with the sub-hippocampal injection of human-derived glioblastoma stem cells (GSCs) were scanned under high-field conditions (QSI, 7T, 512 gradient directions), and tumor-associated tractography was compared with the 3D microscopic reconstruction of immunostained GSCs. Results: In the principal enrollment cohort of patients with glioblastoma, all cases displayed tractography patterns with tumor-intersecting tract bundles extending into brain parenchyma, a phenotype which was reproduced in PDX mice as well as in a larger comparison cohort of glioblastoma patients (n = 66), when applying similar methods. Reconstructed spatial patterns of GSCs in PDX mice closely mirrored tumor-associated tractography. On a Kaplan–Meier survival analysis of n = 66 patients, the calculated intra-tumoral mean diffusivity predicted the overall survival (p = 0.037), as did tractography-associated features including mean tract length (p = 0.039) and mean projecting tract length (p = 0.022). The RNA sequencing of human tissue samples (n = 13 tumor samples from a single patient) revealed the overexpression of transcripts which regulate cell motility in tract-associated samples. Conclusions: QSI discriminates tumor-specific patterns of inter-voxel coherence believed to represent white matter pathways which may be susceptible to glioblastoma invasion. These findings may lay the groundwork for future work on therapeutic targeting, patient stratification, and prognosis in glioblastoma.

Functional diffusion mapping (fDM) is a cancer imaging technique that quantifies voxelwise changes in apparent diffusion coefficient (ADC). Previous studies have shown value of fDMs in bevacizumab therapy for recurrent glioblastoma multiforme (GBM). The aim of the present study was to implement explicit criteria for diffusion MRI quality control and independently evaluate fDM performance in a multicenter clinical trial (RTOG 0625/ACRIN 6677). A total of 123 patients were enrolled in the current multicenter trial and signed institutional review board-approved informed consent at their respective institutions. MRI was acquired prior to and 8 weeks following therapy. A 5-point QC scoring system was used to evaluate DWI quality. fDM performance was evaluated according to the correlation of these metrics with PFS and OS at the first follow-up time-point. Results showed ADC variability of 7.3% in NAWM and 10.5% in CSF. A total of 68% of patients had usable DWI data and 47% of patients had high quality DWI data when also excluding patients that progressed before the first follow-up. fDM performance was improved by using only the highest quality DWI. High pre-treatment contrast enhancing tumor volume was associated with shorter PFS and OS. A high volume fraction of increasing ADC after therapy was associated with shorter PFS, while a high volume fraction of decreasing ADC was associated with shorter OS. In summary, DWI in multicenter trials are currently of limited value due to image quality. Improvements in consistency of image quality in multicenter trials are necessary for further advancement of DWI biomarkers.