Search

Rationale and objectives Prostate MRI improves detection of clinically significant prostate cancer; however, its diagnostic performance has wide variation. Artificial intelligence (AI) has the potential to assist radiologists in the detection and classification of prostatic lesions. Herein, we aimed to develop and test a cascaded deep learning detection and classification system trained on biparametric prostate MRI using PI-RADS for assisting radiologists during prostate MRI read out. Materials and Methods T2-weighted, diffusion-weighted (ADC maps, high b value DWI) MRI scans obtained at 3 Tesla from two institutions (n = 1043 in-house and n = 347 Prostate-X, respectively) acquired between 2015 to 2019 were used for model training, validation, testing. All scans were retrospectively reevaluated by one radiologist. Suspicious lesions were contoured and assigned a PI-RADS category. A 3D U-Net-based deep neural network was used to train an algorithm for automated detection and segmentation of prostate MRI lesions. Two 3D residual neural network were used for a 4-class classification task to predict PI-RADS categories 2 to 5 and BPH. Training and validation used 89% (n = 1290 scans) of the data using 5 fold cross-validation, the remaining 11% (n = 150 scans) were used for independent testing. Algorithm performance at lesion level was assessed using sensitivities, positive predictive values (PPV), false discovery rates (FDR), classification accuracy, Dice similarity coefficient (DSC). Additional analysis was conducted to compare AI algorithm's lesion detection performance with targeted biopsy results. Results Median age was 66 years (IQR = 60-71), PSA 6.7 ng/ml (IQR = 4.7-9.9) from in-house cohort. In the independent test set, algorithm correctly detected 111 of 198 lesions leading to 56.1% (49.3%-62.6%) sensitivity. PPV was 62.7% (95% CI 54.7%-70.7%) with FDR of 37.3% (95% CI 29.3%-45.3%). Of 79 true positive lesions, 82.3% were tumor positive at targeted biopsy, whereas of 57 false negative lesions, 50.9% were benign at targeted biopsy. Median DSC for lesion segmentation was 0.359. Overall PI-RADS classification accuracy was 30.8% (95% CI 24.6%-37.8%). Conclusion Our cascaded U-Net, residual network architecture can detect, classify cancer suspicious lesions at prostate MRI with good detection, reasonable classification performance metrics.

Federated learning (FL) is a method used for training artificial intelligence models with data from multiple sources while maintaining data anonymity, thus removing many barriers to data sharing. Here we used data from 20 institutes across the globe to train a FL model, called EXAM (electronic medical record (EMR) chest X-ray AI model), that predicts the future oxygen requirements of symptomatic patients with COVID-19 using inputs of vital signs, laboratory data and chest X-rays. EXAM achieved an average area under the curve (AUC) >0.92 for predicting outcomes at 24 and 72 h from the time of initial presentation to the emergency room, and it provided 16% improvement in average AUC measured across all participating sites and an average increase in generalizability of 38% when compared with models trained at a single site using that site’s data. For prediction of mechanical ventilation treatment or death at 24 h at the largest independent test site, EXAM achieved a sensitivity of 0.950 and specificity of 0.882. In this study, FL facilitated rapid data science collaboration without data exchange and generated a model that generalized across heterogeneous, unharmonized datasets for prediction of clinical outcomes in patients with COVID-19, setting the stage for the broader use of FL in healthcare. Federated learning, a method for training artificial intelligence algorithms that protects data privacy, was used to predict future oxygen requirements of symptomatic patients with COVID-19 using data from 20 different institutes across the globe.

Medical image annotation is a major hurdle for developing precise and robust machine learning models. Annotation is expensive, time-consuming, and often requires expert knowledge, particularly in the medical field. Here, we suggest using minimal user interaction in the form of extreme point clicks to train a segmentation model which, in effect, can be used to speed up medical image annotation. An initial segmentation is generated based on the extreme points utilizing the random walker algorithm. This initial segmentation is then used as a noisy supervision signal to train a fully convolutional network that can segment the organ of interest, based on the provided user clicks. Through experimentation on several medical imaging datasets, we show that the predictions of the network can be refined using several rounds of training with the prediction from the same weakly annotated data. Further improvements are shown utilizing the clicked points within a custom-designed loss and attention mechanism. Our approach has the potential to speed up the process of generating new training datasets for the development of new machine learning and deep learning-based models for, but not exclusively, medical image analysis., Comment: 13 pages, 6 figures, 1 table

Syncope is a common emergency department (ED) chief complaint. Rarely, syncope can be the result of right ventricular outflow obstruction from an intracardiac tumor, such as an intracardiac extension of intravenous leiomyomatosis (IVL). Typically, this type of tumor is confined to the pelvic veins, but in very rare cases, it can extend through the inferior vena cava into the right atrium. Point-of-care ultrasound (POCUS) can be a crucial tool in the ED for identifying intracardiac tumors presenting as syncope and expediting clinical management. We present the case of a 39-year-old female with no prior medical history that presented to the ED having experienced dyspnea on exertion and two syncopal episodes prior to ED admission. POCUS use in the ED elucidated the presence of a right atrial mass and further imaging showed a mass on the patient's uterus. After surgical removal of a portion of the atrial mass, a subsequent biopsy revealed it had leiomyoma-like features; as such, the patient was diagnosed with IVL. This case illustrates the importance of using POCUS in the ED to help determine the etiology of syncope. Although intracardiac extensions of IVL are rare, it is important for emergency physicians to keep this diagnosis in the differential in patients with symptoms or risk factors suggestive of IVL with intracardiac extension.

It is generally recognized that lives are saved by administering high-quality cardiopulmonary resuscitation (CPR) to patients in cardiac arrest. A focused assessment with sonography for trauma (FAST) examination is an effective and non–invasive method for detecting rare complications of CPR, such as hemorrhage from abdominal visceral injury. We report the case of a 56-year-old female suffering from intra-abdominal hemorrhage caused by a liver laceration following CPR. The hemoperitoneum was diagnosed by a FAST examination. Although severe complications of CPR are rare, they can be easily detected with the use of a FAST examination. A FAST examination should be considered as a post-resuscitation approach to assess for life-threatening complications in all patients following cardiopulmonary resuscitation.

Intrauterine devices (IUDs) are used worldwide. The 2 types that are used are the levonorgestrel IUD and a copper containing IUD. This is a case study of a 30-year-old female with a levonorgestrel IUD who was diagnosed with a ruptured ectopic pregnancy in the emergency department (ED). Point-of-care urine pregnancy test and point-of-care ultrasound (POCUS) were vital in making this diagnosis and should be utilized in patients assigned female at birth who present with abdominal pain.

A better understanding of temporal relationships between chest CT and labs may provide a reference for disease severity over the disease course. Generalized curves of lung opacity volume and density over time can be used as standardized references from well before symptoms develop to over a month after recovery, when residual lung opacities remain. 739 patients with COVID-19 underwent CT and RT-PCR in an outbreak setting between January 21st and April 12th, 2020. 29 of 739 patients had serial exams (121 CTs and 279 laboratory measurements) over 50 ± 16 days, with an average of 4.2 sequential CTs each. Sequential volumes of total lung, overall opacity and opacity subtypes (ground glass opacity [GGO] and consolidation) were extracted using deep learning and manual segmentation. Generalized temporal curves of CT and laboratory measurements were correlated. Lung opacities appeared 3.4 ± 2.2 days prior to symptom onset. Opacity peaked 1 day after symptom onset. GGO onset was earlier and resolved later than consolidation. Lactate dehydrogenase, and C-reactive protein peaked earlier than procalcitonin and leukopenia. The temporal relationships of quantitative CT features and clinical labs have distinctive patterns and peaks in relation to symptom onset, which may inform early clinical course in patients with mild COVID-19 pneumonia, or may shed light upon chronic lung effects or mechanisms of medical countermeasures in clinical trials.

Objective To demonstrate enabling multi-institutional training without centralizing or sharing the underlying physical data via federated learning (FL). Materials and Methods Deep learning models were trained at each participating institution using local clinical data, and an additional model was trained using FL across all of the institutions. Results We found that the FL model exhibited superior performance and generalizability to the models trained at single institutions, with an overall performance level that was significantly better than that of any of the institutional models alone when evaluated on held-out test sets from each institution and an outside challenge dataset. Discussion The power of FL was successfully demonstrated across 3 academic institutions while avoiding the privacy risk associated with the transfer and pooling of patient data. Conclusion Federated learning is an effective methodology that merits further study to enable accelerated development of models across institutions, enabling greater generalizability in clinical use.

Emergency physicians often rely on heuristics to facilitate clinical decisions due to the large volume of patients they see daily. Consequently, they are vulnerable to error and bias. We report the case of a 69-year-old male that presented to the emergency department (ED) with shortness of breath, productive cough, and dyspnea on exertion. One day prior to ED admission, he was diagnosed with bronchitis; however, point-of-care ultrasound (POCUS) in the ED identified acute pulmonary embolism. This case illustrates the potential dangers of anchoring bias and shows the benefits of using point-of-care ultrasound of the lungs and heart to assist in the diagnosis of acute pulmonary embolism.

The World Health Organization (WHO) categorized coronavirus disease 2019 (COVID-19) as a pandemic due to its high contagion rate and widespread infectivity in February 2020. In the United States (US), one of the public health concerns is the adequacy of resources to treat infected cases. We describe a case of a previously well, 9-year-old obese boy who presented to the emergency department with shortness of breath, fever, abdominal pain and cough with chest pain. He was diagnosed with COVID-19 through significant family contact, confirmed by polymerase chain reaction and found to be at high risk of venous thromboembolism due to abnormal d-dimer. Lung point-of-care ultrasound (POCUS) in the emergency department observed significant lung pathology, including pleural thickening, consolidation, and B lines. A chest X-ray (CXR) found bilateral ground glass opacities and interstitial prominences consistent with viral pneumonia. Our case suggests that lung POCUS can provide adequate and rapid imaging to assess lung pathology of COVID-19 in a pediatric patient. As there is limited literature on use of lung POCUS in pediatric patients infected with SARS-CoV-2, our case emphasizes its function as a potentially efficient modality in bedside assessment.

Late life is a period frequently marked by decline in personal health and heightened need for social support. Consequently, the social networks in which individuals are embedded assume an increasingly central role in the health and wellbeing of older adults. In the present article, I review the state of the literature on social networks and health in later life. By drawing on insights from the sociology of ageing and the life course, I address new developments and current challenges within the field. Chief among these developments and challenges is the recognition that the ageing process does not occur in a vacuum. Rather, individuals are consistently exposed to numerous changes to their social lives which have strong implications for current and future health outcomes. Upon highlighting the latest innovations within the field of networks and health, I conclude with useful directions for future research.

ecent advances in deep learning for medical image segmentation demonstrate expert-level accuracy.ecent advances in deep learning for medical image segmentation demonstrate expert-level accuracy.R However, application of these models in clinically realistic environments can result in poor generalization and decreased accuracy, mainly due to the domain shift across different hospitals, scanner vendors, imaging protocols, and patient populations etc. Common transfer learning and domain adaptation techniques are proposed to address this bottleneck. However, these solutions require data (and annotations) from the target domain to retrain the model, and is therefore restrictive in practice for widespread model deployment. Ideally, we wish to have a trained (locked) model that can work uniformly well across unseen domains without further training. In this paper, we propose a deep stacked transformation approach for domain generalization. Specifically, a series of n stacked transformations are applied to each image during network training. The underlying assumption is that the “expected” domain shift for a specific medical imaging modality could be simulated by applying extensive data augmentation on a single source domain, and consequently, a deep model trained on the augmented “big” data (BigAug) could generalize well on unseen domains. We exploit four surprisingly effective, but previously understudied, image-based characteristics for data augmentation to overcome the domain generalization problem. We train and evaluate the BigAug model (with n = 9 transformations) on three different 3D segmentation tasks (prostate gland, left atrial, left ventricle) covering two medical imaging modalities (MRI and ultrasound) involving eight publicly available challenge datasets. The results show that when training on relatively small dataset (n=10~32 volumes, depending on the size of the available datasets) from a single source domain: (i) BigAug models degrade an average of 11% (Dice score change) from source to unseen domain, substantially better than conventional augmentation (degrading 39%) and CycleGAN-based domain adaptation method (degrading 25%), (ii) BigAug is better than “shallower” stacked transforms (i.e. those with fewer transforms) on unseen domains and demonstrates modest improvement to conventional augmentation on the source domain, (iii) after training with BigAug on one source domain, performance on an unseen domain is similar to training a model from scratch on that domain when using the same number of training samples. When training on large datasets (n=465 volumes) with BigAug, (iv) application to unseen domains reaches the performance of state-of-the-art fully supervised models that are trained and tested on their source domains. These findings establish a strong benchmark for the study of domain generalization in medical imaging, and can be generalized to the design of highly robust deep segmentation models for clinical deployment.

Transthyretin amyloidosis is an increasingly recognized cause of restrictive cardiomyopathy related to amyloid fibril deposition in cardiac tissues. As treatment therapies have emerged for transthyretin amyloidosis (ATTR), so has interest in using biomarkers to identify disease prior to advanced presentation. Lower levels of transthyretin and retinol binding protein-4 have been demonstrated in patients with pathogenic mutations of transthyretin either with or without clinical disease. Levels associate with the severity of mutations as well as response to treatment with transthyretin stabilizers or small interfering RNA molecules which silence transthyretin production. Transthyretin stability is the rate limiting step of amyloid fibril formation and directly measuring transthyretin kinetic stability has the potential to identify patients as risk as well as therapeutic response to treatment regardless of pathogenic or wild-type genetics. In addition, non-antibody protein-based peptide probes have been developed that directedly measure misfolded transthyretin oligomers due to transthyretin breakdown. Although promising, both TTR kinetic and protein peptide probes remain in early stages of clinical investigation. Transthyretin, retinol binding protein-4, transthyretin kinetic stability, and protein-based peptide probes have potential as biomarkers to facilitate an earlier ATTR diagnosis for patients with pathogenic transthyretin mutations.

Radiologists should be aware of the many hereditary cancer syndromes that confer increased risks of breast and gynecologic malignancies in individuals with deleterious gene mutations, as well as th...

Artificial intelligence (AI) methods for the automatic detection and quantification of COVID-19 lesions in chest computed tomography (CT) might play an important role in the monitoring and management of the disease. We organized an international challenge and competition for the development and comparison of AI algorithms for this task, which we supported with public data and state-of-the-art benchmark methods. Board Certified Radiologists annotated 295 public images from two sources (A and B) for algorithms training (n=199, source A), validation (n=50, source A) and testing (n=23, source A; n=23, source B). There were 1,096 registered teams of which 225 and 98 completed the validation and testing phases, respectively. The challenge showed that AI models could be rapidly designed by diverse teams with the potential to measure disease or facilitate timely and patient-specific interventions. This paper provides an overview and the major outcomes of the COVID-19 Lung CT Lesion Segmentation Challenge - 2020.

Annotation-efficient deep learning refers to methods and practices that yield high-performance deep learning models without the use of massive carefully labeled training datasets. This paradigm has recently attracted attention from the medical imaging research community because (1) it is difficult to collect large, representative medical imaging datasets given the diversity of imaging protocols, imaging devices, and patient populations, (2) it is expensive to acquire accurate annotations from medical experts even for moderately sized medical imaging datasets, and (3) it is infeasible to adapt data-hungry deep learning models to detect and diagnose rare diseases whose low prevalence hinders data collection.

Access to cognitive stimulation through social interactions is a key mechanism used to explain the association between personal networks, cognitive health, and brain structure in older adults. However, little research has assessed how best to operationalize access to novel or diverse social stimuli using social network measures, many of which were designed to study information diffusion within large whole networks (e.g., structural holes and bridging social capital). Using data from 277 adults in the Social Networks and Alzheimer Disease (SNAD) study, we aimed to evaluate such measures for use in research on cognitive aging using personal social networks. We found a positive association between individual measures of structural holes and cognitive health, but not with brain structure. Further, we extracted a latent measure of bridging social capital using multiple individual measures (i.e., structural holes, network diversity, weak ties, and network size) and found it was significantly associated with cognitive health and brain structure, supporting the utility of this concept and related measures in the study of cognitive aging. Finally, individual measures may underestimate the effects of multidimensional bridging social capital on cognitive health and brain structure compared to a latent measure that combines them.

INTRODUCTION AND OBJECTIVE:The Prostate Imaging-Reporting and Data System (PI-RADS) version 2.1 has become the standard in risk categorization of prostate lesions. However, there is significant int...

Purpose For the planning and navigation of neurosurgery, we have developed a fully convolutional network (FCN)-based method for brain structure segmentation on magnetic resonance (MR) images. The capability of an FCN depends on the quality of the training data (i.e., raw data and annotation data) and network architectures. The improvement of annotation quality is a significant concern because it requires much labor for labeling organ regions. To address this problem, we focus on skip connection architectures and reveal which skip connections are effective for training FCNs using sparsely annotated brain images. Methods We tested 2D FCN architectures with four different types of skip connections. The first was a U-Net architecture with horizontal skip connections that transfer feature maps at the same scale from the encoder to the decoder. The second was a U-Net++ architecture with dense convolution layers and dense horizontal skip connections. The third was a full-resolution residual network (FRRN) architecture with vertical skip connections that pass feature maps between each downsampled scale path and the full-resolution scale path. The last one was a hybrid architecture with a combination of horizontal and vertical skip connections. We validated the effect of skip connections on medical image segmentation from sparse annotation based on these four FCN architectures, which were trained under the same conditions. Results For multi-class segmentation of the cerebrum, cerebellum, brainstem, and blood vessels from sparsely annotated MR images, we performed a comparative evaluation of segmentation performance among the above four FCN approaches: U-Net, U-Net++, FRRN, and hybrid architectures. The experimental results show that the horizontal skip connections in the U-Net architectures were effective for the segmentation of larger-sized objects, while the vertical skip connections in the FRRN architecture improved the segmentation of smaller-sized objects. The hybrid architecture with both horizontal and vertical skip connections achieved the best results of the four FCN architectures. We then performed an ablation study to explore which skip connections in the FRRN architecture contributed to the improved segmentation of blood vessels. In the ablation study, we compared the segmentation performance between architectures with a horizontal path (HP), a horizontal path and vertical up paths (HP+VUPs), a horizontal path and vertical down paths (HP+VDPs), and a horizontal path and vertical up and down paths (FRRN). We found that the vertical up paths were effective in improving the segmentation of smaller-sized objects. Conclusions This paper investigated which skip connection architectures were effective for multi-class brain segmentation from sparse annotation. Consequently, using vertical skip connections with horizontal skip connections allowed FCNs to improve segmentation performance. This article is protected by copyright. All rights reserved.

PURPOSE: To identify distinguishing CT radiomic features of pancreatic ductal adenocarcinoma (PDAC) and to investigate whether radiomic analysis with machine learning can distinguish between patients who have PDAC and those who do not. MATERIALS AND METHODS: This retrospective study included contrast material–enhanced CT images in 436 patients with PDAC and 479 healthy controls from 2012 to 2018 from Taiwan that were randomly divided for training and testing. Another 100 patients with PDAC (enriched for small PDACs) and 100 controls from Taiwan were identified for testing (from 2004 to 2011). An additional 182 patients with PDAC and 82 healthy controls from the United States were randomly divided for training and testing. Images were processed into patches. An XGBoost (https://xgboost.ai/) model was trained to classify patches as cancerous or noncancerous. Patients were classified as either having or not having PDAC on the basis of the proportion of patches classified as cancerous. For both patch-based and patient-based classification, the models were characterized as either a local model (trained on Taiwanese data only) or a generalized model (trained on both Taiwanese and U.S. data). Sensitivity, specificity, and accuracy were calculated for patch- and patient-based analysis for the models. RESULTS: The median tumor size was 2.8 cm (interquartile range, 2.0–4.0 cm) in the 536 Taiwanese patients with PDAC (mean age, 65 years ± 12 [standard deviation]; 289 men). Compared with normal pancreas, PDACs had lower values for radiomic features reflecting intensity and higher values for radiomic features reflecting heterogeneity. The performance metrics for the developed generalized model when tested on the Taiwanese and U.S. test data sets, respectively, were as follows: sensitivity, 94.7% (177 of 187) and 80.6% (29 of 36); specificity, 95.4% (187 of 196) and 100% (16 of 16); accuracy, 95.0% (364 of 383) and 86.5% (45 of 52); and area under the curve, 0.98 and 0.91. CONCLUSION: Radiomic analysis with machine learning enabled accurate detection of PDAC at CT and could identify patients with PDAC. Keywords: CT, Computer Aided Diagnosis (CAD), Pancreas, Computer Applications-Detection/Diagnosis Supplemental material is available for this article. © RSNA, 2021 An earlier incorrect version of this article appeared online. This article was corrected on July 30, 2021.

Research suggests social connectedness may help older adults with dementia maintain cognitive functionality and quality of life. However, little is known about its specific social and biological mechanisms. This paper proposes two pathways through social bridging (i.e., cognitive enrichment through expansive social networks) and bonding (i.e., neuroendocrine benefits of integration in cohesive social networks). We provide preliminary evidence for these pathways using neuroimaging, cognitive, and egocentric social network data from the Social Networks and Alzheimer's Disease (SNAD) study (N = 280). We found that network size, density, and presence of weak ties (i.e., social bridging) moderated the association between brain atrophy and cognitive function, while marriage/cohabitation (i.e., social bonding) moderated the association between perceived stress and cognitive function. We argue that social connectedness may have downstream implications for multiple pathophysiological processes in cognitive aging, even negating existing structural damage to the brain, making it a strong candidate for clinical or policy intervention.

Central retinal artery occlusion (CRAO) is a rare, but serious, diagnosis that can lead to blindness, most often due to thromboembolic disease. In the emergency department (ED), CRAO can present as acute, painless loss of vision. Physicians need quick ways to rule in this diagnosis due to the time-sensitive nature of the event. We describe two patients in this cases series who present to the same ED with unilateral painless vision loss and histories that include notable risk factors such as thromboembolic and atherosclerotic disease. Upon arrival, point-of-care ultrasound (POCUS) done at the bedside allowed for quick determination of CRAO. ​​​​​​​The importance of this case series is to emphasize the efficacy of POCUS in evaluating patients with painless vision loss in the ED setting.

Loss-of-function mutations in progranulin (GRN) are a major autosomal dominant cause of frontotemporal dementia. Most pathogenicGRNmutations result in progranulin haploinsufficiency, which is thought to cause frontotemporal dementia inGRNmutation carriers. Progranulin haploinsufficiency may drive frontotemporal dementia pathogenesis by disrupting lysosomal function, as patients withGRNmutations on both alleles develop the lysosomal storage disorder neuronal ceroid lipofuscinosis, and frontotemporal dementia patients withGRNmutations (FTD-GRN) also accumulate lipofuscin. The specific lysosomal deficits caused by progranulin insufficiency remain unclear, but emerging data indicate that progranulin insufficiency may impair lysosomal sphingolipid-metabolizing enzymes. We investigated the effects of progranulin insufficiency on sphingolipid-metabolizing enzymes in the inferior frontal gyrus of FTD-GRNpatients using fluorogenic activity assays, biochemical profiling of enzyme levels and posttranslational modifications, and quantitative neuropathology. Of the enzymes studied, only β-glucocerebrosidase exhibited impairment in FTD-GRNpatients. Brains from FTD-GRNpatients had lower activity than controls, which was associated with lower levels of mature β-glucocerebrosidase protein and accumulation of insoluble, incompletely glycosylated β-glucocerebrosidase. Immunostaining revealed loss of neuronal β-glucocerebrosidase in FTD-GRNpatients. To investigate the effects of progranulin insufficiency on β-glucocerebrosidase outside of the context of neurodegeneration, we investigated β-glucocerebrosidase activity in progranulin-insufficient mice. Brains fromGrn−/−mice had lower β-glucocerebrosidase activity than wild-type littermates, which was corrected by AAV-progranulin gene therapy. These data show that progranulin insufficiency impairs β-glucocerebrosidase activity in the brain. This effect is strongest in neurons and may be caused by impaired β-glucocerebrosidase processing.

Dieser Ubersichtsartikel beschreibt alle relevanten Aspekte entsprechend der neuen S2k-Leitlinie der Deutschen Gesellschaft fur Urologie (DGU) zur Diagnostik und Behandlung der IC/BPS (interstitielle Zystitis/Blasenschmerzsyndrom). Eine Aufstellung notwendiger und optionaler Untersuchungen und der Notwendigkeit einer Ausschlussdiagnostik werden zusammenfassend dargestellt und gewertet. Auch entsprechend aufgelistete Therapieoptionen (von konservativen, oral-medikamentosen, komplementar-medizinischen bis hin zu intravesikalen und interventionell-operativen Verfahren) geben dem Leser einen guten Uberblick uber die Inhalte der Leitlinie und moglicher therapeutischer Ansatze. Letztlich werden auch Ubersichten der Empfehlungen inklusive Konsens der Leitliniengruppe in verschiedenen Infoboxen beschrieben.

Palliative care is a field of medicine that delivers patient-centered care for individuals and their families suffering from serious illness at all stages of the disease trajectory. It addresses the major priorities of relieving suffering, establishing goals of care, and managing physical symptoms while integrating the psychosocial, cultural, spiritual, and existential complexities of coping with chronic illness. This article discusses the role of palliative care in the health care system. It reviews the importance of prognostication, disease trajectory, and communication. The role of the primary care physician as part of a multidisciplinary team member delivering primary palliative care is emphasized.

OBJECTIVE To determine the safety and efficacy of advanced practice provider (APP)-performed newborn circumcisions (NBCs), we reviewed outcomes of NBCs performed by pediatric urologists and APPs. We hypothesize comparable clinical outcomes between the groups. METHODS All urology performed NBCs during a 5-year period were reviewed, including time surrounding implementation of the APP-led clinic. Return to emergency department (ED) rates, return to operating room (OR) rates, and intraprocedure bleeding requiring intervention were reviewed. Fisher exact and Mann-Whitney testing were utilized. RESULTS There were no statistically significant differences in rates of intraprocedure bleeding, return to ED in 30days, return to OR for revision or other related penile surgery, or the overall number of patients with complications between the groups. Thirteen patients had complications in the APP cohort, compared to 8 in the urologist cohort. There was a difference in age and weight, with urologists performing NBCs on older and heavier patients. There was no difference in clinical outcomes between children over and under 10 pounds (4.5 kg). There was a significant difference in the need for revision circumcision when comparing children older vs younger than 30days (1.9% vs 0%, P = 0.034). CONCLUSION An APP-led NBC clinic is both safe and feasible. The widely used age and weight cutoffs for NBC need to be further evaluated, as there was no significant difference in clinical outcomes. This practice design provides pediatric urologists more time to focus on the most complex patients, both in the clinic and OR.

Background Patients with a prenatal diagnosis of lower urinary tract obstruction (LUTO) may undergo prenatal interventions, such as vesicoamniotic shunt (VAS) placement, as a temporary solution for relieving urinary tract obstruction. A recent FDA communication has raised awareness of the potential neurocognitive adverse effects of anesthesia in children. We hypothesized as to whether a prenatal LUTO staging system was predictive of the number of anesthesia events for prenatally diagnosed LUTO patients. Methods We retrospectively reviewed the prenatal and postnatal clinical records for patients with prenatally diagnosed LUTO from 2012 to 2015. Patients were stratified by prenatal VAS status and by LUTO disease severity according to Ruano et al. (Ultrasound Obstet Gynecol. 2016). Results 31 patients were identified with a prenatal LUTO diagnosis, and postnatal records were available for 21 patients (seven patients in each stage). When combining prenatal and postnatal anesthesia, there was a significant difference in the number of anesthesia encounters by stage (1.6, 3.7, and 6.7 for Stage I, II, and III respectively, p = .034). Upon univariate analysis, higher gestational age (GA) at birth was associated with a decreased number of anesthesia events in the first year (p = .031). Conclusions The majority of infants with prenatally diagnosed LUTO will undergo postnatal procedures with general anesthesia exposure in the first year of life. Patients with higher prenatal LUTO severity experienced a higher number of both prenatal and postnatal anesthesia encounters. In addition, higher GA at birth was associated with fewer anesthesia encounters in the first year. Level of evidence This is a prognostic study with Level IV evidence.

Objective To evaluate the management and clinical outcomes of nonfunctioning upper pole moieties treated with either upper pole heminephrectomy or upper pole preservation with lower ureteral reconstruction at a single tertiary institution. Methods After Institutional Review Board (IRB) approval, patients with duplicated systems undergoing upper pole heminephrectomy, ureteroureterostomy, or common sheath ureteral reimplantation from 2012-2017 were identified. Only patients with a nonfunctioning upper pole moiety on ultrasound or renal scan were included. Patients undergoing upper pole heminephrectomy were compared to those undergoing upper pole preservation with respect to demographics, anatomic variations preoperatively, and postoperative outcomes. Results Twenty-seven (57%) patients underwent upper pole preservation with lower ureteral reconstruction; 20 (43%) patients underwent upper pole heminephrectomy. Patients undergoing lower ureteral reconstruction were older (1.63 vs 2.76 years, P = .018) and more commonly presented with lower pole vesicoureteral reflux (67% vs 25%, P = .008). No significant difference in postoperative complications was seen between the two groups. After ureteroureterostomy, one patient developed new onset symptomatic reflux to the upper pole requiring intravesical reimplantation. In the heminephrectomy group, 4 of 11 patients with ureteroceles had ureterocelectomy with concomitant lower pole reimplantation. After heminephrectomy, two additional patients required further interventions: ureterocele excision and transurethral polyp excision. Conclusion For patients with nonfunctional upper poles, lower tract reconstruction is a safe alternative to upper pole heminephrectomy. No significant difference in outcomes was seen. Considering that nearly 1 of 3 of patients with upper pole heminephrectomy required additional lower urinary tract procedures, pursuing upper pole preservation with lower urinary tract reconstruction may be favorable.

T-cell lymphoblastic lymphoma (T-cell LBL) is an uncommon diagnosis for acute dyspnea in pediatric emergencies. This case details a 13-year-old boy presenting to the ED with dyspnea, who was diagnosed with T-cell LBL. It was a unique presentation in which there was no obvious mediastinal mass on the examination or primary imaging. As a safe and cost-effective modality for a patient that was too unstable to transfer to the radiology department for computed tomography, point-of-care ultrasound (POCUS) was useful in the patient’s rapid assessment for suspected pericardial and pleural effusion. This case highlights the advantage of early utilization of POCUS for pediatric patients with dyspnea.

Objectives The cognitive reserve hypothesis has been proposed as a key mechanism explaining the link between social networks and cognitive function but has rarely been empirically tested using neuroimaging data. This study examines whether social network attributes moderate the association between amygdalar volume and cognitive function. Methods Data were from the Social Networks in Alzheimer Disease study (N = 154) and Indiana Alzheimer’s Disease Research Center. Social networks were measured using the PhenX Social Network Battery. Regional data from magnetic resonance imaging (amygdalar volume [AV]) were analyzed using FreeSurfer software. Cognitive function was measured using the Montreal Cognitive Assessment (MoCA) and consensus diagnosis. Linear regression analyses were conducted to test the moderating role of social networks on the association between AV and cognitive function. Results Participants with greater ability to span multiple social roles and subgroups within their networks scored higher on the MoCA after adjusting for sociodemographic variables, depression, frequency of contact, and AV. Social networks moderated the association between AV and cognitive function. Discussion Among participants who engaged in diverse and loosely connected social networks, the expected adverse cognitive effects of brain volume in regions implicated in socioemotional processing were attenuated. These findings suggest that cognitive stimulation achieved through social interaction with a diverse array of social relationships across multiple contexts may help promote cognitive reserve.

Recently, neural architecture search (NAS) has been applied to automatically search high-performance networks for medical image segmentation. The NAS search space usually contains a network topology level (controlling connections among cells with different spatial scales) and a cell level (operations within each cell). Existing methods either require long searching time for large-scale 3D image datasets, or are limited to pre-defined topologies (such as U-shaped or single-path). In this work, we focus on three important aspects of NAS in 3D medical image segmentation: flexible multi-path network topology, high search efficiency, and budgeted GPU memory usage. A novel differentiable search framework is proposed to support fast gradient-based search within a highly flexible network topology search space. The discretization of the searched optimal continuous model in differentiable scheme may produce a sub-optimal final discrete model (discretization gap). Therefore, we propose a topology loss to alleviate this problem. In addition, the GPU memory usage for the searched 3D model is limited with budget constraints during search. Our Differentiable Network Topology Search scheme (DiNTS) is evaluated on the Medical Segmentation Decathlon (MSD) challenge, which contains ten challenging segmentation tasks. Our method achieves the state-of-the-art performance and the top ranking on the MSD challenge leaderboard., CVPR2021 oral

Objectives To evaluate a novel scoring system that combines several prenatal parameters for selecting ideal candidates for fetal intervention, and for predicting postnatal survival in patients with severe fetal lower urinary tract obstruction (LUTO). Methods We retrospectively reviewed all cases of severe LUTO evaluated for fetal intervention in a single large fetal center between January 2013 and December 2017. A scoring system for determining fetal candidacy for intervention was retrospectively developed based on postnatal outcomes. The proposed scoring system included fetal urinary biochemistry, renal ultrasound parameters, initial bladder volume, and degree of bladder refill. Relevant demographic characteristics, ultrasound reports and laboratory results were reviewed. Receiver operating characteristic (ROC) curves were used to select the cut-off values for initial bladder volume and degree of bladder refill and to evaluate the performance of the scoring system in predicting postnatal death. Results Of the 79 LUTO patients evaluated, 31 were eligible for the study. The overall 6-month postnatal survival was 64.5 % (20/31). A scoring system (0−8) was suggested with 2 points for unfavorable biochemistry, 4 points for ultrasound evidence of dysplastic kidneys, 1 point for inadequate initial bladder volume and 1 point for inadequate bladder refill. Scores>3 (N = 7) were associated with 0 % 6-month survival. The ROC curve for predicting postnatal mortality showed area under curve (AUC) of 0.82 (95 % CI 0.65−0.99). Subgroup analysis within subjects who underwent fetal intervention (N = 22) also confirmed the significance of the distribution of the scoring system between groups who survived and those who did not after adjustment for GA at delivery (p = 0.01). Conclusion We propose a novel scoring system for antenatal evaluation of patients with severe LUTO which may be useful in selecting those candidates most appropriate for intervention and in counseling parents about predicted postnatal outcome.

Deep learning models for medical image segmentation are primarily data-driven. Models trained with more data lead to improved performance and generalizability. However, training is a computationally expensive process because multiple hyper-parameters need to be tested to find the optimal setting for best performance. In this work, we focus on accelerating the estimation of hyper-parameters by proposing two novel methodologies: proxy data and proxy networks. Both can be useful for estimating hyper-parameters more efficiently. We test the proposed techniques on CT and MR imaging modalities using well-known public datasets. In both cases using one dataset for building proxy data and another data source for external evaluation. For CT, the approach is tested on spleen segmentation with two datasets. The first dataset is from the medical segmentation decathlon (MSD), where the proxy data is constructed, the secondary dataset is utilized as an external validation dataset. Similarly, for MR, the approach is evaluated on prostate segmentation where the first dataset is from MSD and the second dataset is PROSTATEx. First, we show higher correlation to using full data for training when testing on the external validation set using smaller proxy data than a random selection of the proxy data. Second, we show that a high correlation exists for proxy networks when compared with the full network on validation Dice score. Third, we show that the proposed approach of utilizing a proxy network can speed up an AutoML framework for hyper-parameter search by 3.3\(\times \), and by 4.4\(\times \) if proxy data and proxy network are utilized together.

Multiple instance learning (MIL) is a key algorithm for classification of whole slide images (WSI). Histology WSIs can have billions of pixels, which create enormous computational and annotation challenges. Typically, such images are divided into a set of patches (a bag of instances), where only bag-level class labels are provided. Deep learning based MIL methods calculate instance features using convolutional neural network (CNN). Our proposed approach is also deep learning based, with the following two contributions: Firstly, we propose to explicitly account for dependencies between instances during training by embedding self-attention Transformer blocks to capture dependencies between instances. For example, a tumor grade may depend on the presence of several particular patterns at different locations in WSI, which requires to account for dependencies between patches. Secondly, we propose an instance-wise loss function based on instance pseudo-labels. We compare the proposed algorithm to multiple baseline methods, evaluate it on the PANDA challenge dataset, the largest publicly available WSI dataset with over 11K images, and demonstrate state-of-the-art results.

Active learning is a unique abstraction of machine learning techniques where the model/algorithm could guide users for annotation of a set of data points that would be beneficial to the model, unlike passive machine learning. The primary advantage being that active learning frameworks select data points that can accelerate the learning process of a model and can reduce the amount of data needed to achieve full accuracy as compared to a model trained on a randomly acquired data set. Multiple frameworks for active learning combined with deep learning have been proposed, and the majority of them are dedicated to classification tasks. Herein, we explore active learning for the task of segmentation of medical imaging data sets. We investigate our proposed framework using two datasets: 1.) MRI scans of the hippocampus, 2.) CT scans of pancreas and tumors. This work presents a query-by-committee approach for active learning where a joint optimizer is used for the committee. At the same time, we propose three new strategies for active learning: 1.) increasing frequency of uncertain data to bias the training data set; 2.) Using mutual information among the input images as a regularizer for acquisition to ensure diversity in the training dataset; 3.) adaptation of Dice log-likelihood for Stein variational gradient descent (SVGD). The results indicate an improvement in terms of data reduction by achieving full accuracy while only using 22.69 % and 48.85 % of the available data for each dataset, respectively., Comment: 19 pages, 13 figures, Transactions of Medical Imaging

The boron-10 based Multi-Grid detector is being developed as an alternative to helium-3 based neutron detectors. At the European Spallation Source, the detector will be used for time-of-flight neutron spectroscopy at cold to thermal neutron energies. The objective of this work is to investigate fine time- and energy-resolved effects of the Multi-Grid detector, down to a few $\mu$eV, while comparing it to the performance of a typical helium-3 tube. Furthermore, it is to characterize differences between the detector technologies in terms of internal scattering, as well as the time reconstruction of ~ $\mu$s short neutron pulses. The data were taken at the Helmholtz Zentrum Berlin, where the Multi-Grid detector and a helium-3 tube were installed at the ESS test beamline, V20. Using a Fermi-chopper, the neutron beam of the reactor was chopped into a few tens of $\mu$s wide pulses before reaching the detector, located a few tens of cm downstream. The data of the measurements show an agreement between the derived and calculated neutron detection efficiency curve. The data also provide fine details on the effect of internal scattering, and how it can be reduced. For the first time, the chopper resolution was comparable to the timing resolution of the Multi-Grid detector. This allowed a detailed study of time- and energy resolved effects, as well as a comparison with a typical helium-3 tube., Comment: 37 pages, 21 figures

Introduction. The fight against COVID-19 poses questions as to the clinical presentation, course, diagnosis, and treatment of the condition. This case study presents a patient infected with COVID-19 and suggests with additional research, that bedside ultrasound may be used to diagnose severity of disease and potentially, prognosticate functional lung recovery without using unnecessary resources and exposing additional healthcare professionals to infection. Case Report. A 46-year-old male presented to the emergency department (ED) with cough, fever, and shortness of breath. Chest X-ray showed patchy airspace opacities bilaterally. Rapid testing resulted positive for SARS-CoV-2. Bedside ultrasound showed abnormal lung parenchyma, with diffuse comet tail artifacts, consistent with interstitial pulmonary edema. Following a prolonged intubation, patient’s abnormal lung ultrasound findings are resolved.

Federated learning (FL) is a method used for training artificial intelligence models with data from multiple sources while maintaining data anonymity, thus removing many barriers to data sharing. Here we used data from 20 institutes across the globe to train a FL model, called EXAM (electronic medical record (EMR) chest X-ray AI model), that predicts the future oxygen requirements of symptomatic patients with COVID-19 using inputs of vital signs, laboratory data and chest X-rays. EXAM achieved an average area under the curve (AUC)0.92 for predicting outcomes at 24 and 72 h from the time of initial presentation to the emergency room, and it provided 16% improvement in average AUC measured across all participating sites and an average increase in generalizability of 38% when compared with models trained at a single site using that site's data. For prediction of mechanical ventilation treatment or death at 24 h at the largest independent test site, EXAM achieved a sensitivity of 0.950 and specificity of 0.882. In this study, FL facilitated rapid data science collaboration without data exchange and generated a model that generalized across heterogeneous, unharmonized datasets for prediction of clinical outcomes in patients with COVID-19, setting the stage for the broader use of FL in healthcare.

Building robust deep learning-based models requires diverse training data, ideally from several sources. However, these datasets cannot be combined easily because of patient privacy concerns or regulatory hurdles, especially if medical data is involved. Federated learning (FL) is a way to train machine learning models without the need for centralized datasets. Each FL client trains on their local data while only sharing model parameters with a global server that aggregates the parameters from all clients. At the same time, each client’s data can exhibit differences and inconsistencies due to the local variation in the patient population, imaging equipment, and acquisition protocols. Hence, the federated learned models should be able to adapt to the local particularities of a client’s data. In this work, we combine FL with an AutoML technique based on local neural architecture search by training a “supernet”. Furthermore, we propose an adaptation scheme to allow for personalized model architectures at each FL client’s site. The proposed method is evaluated on four different datasets from 3D prostate MRI and shown to improve the local models’ performance after adaptation through selecting an optimal path through the AutoML supernet.

OBJECTIVE. Deep learning applications in radiology often suffer from overfitting, limiting generalization to external centers. The objective of this study was to develop a high-quality prostate segmentation model capable of maintaining a high degree of performance across multiple independent datasets using transfer learning and data augmentation. MATERIALS AND METHODS. A retrospective cohort of 648 patients who underwent prostate MRI between February 2015 and November 2018 at a single center was used for training and validation. A deep learning approach combining 2D and 3D architecture was used for training, which incorporated transfer learning. A data augmentation strategy was used that was specific to the deformations, intensity, and alterations in image quality seen on radiology images. Five independent datasets, four of which were from outside centers, were used for testing, which was conducted with and without fine-tuning of the original model. The Dice similarity coefficient was used to evaluate model performance. RESULTS. When prostate segmentation models utilizing transfer learning were applied to the internal validation cohort, the mean Dice similarity coefficient was 93.1 for whole prostate and 89.0 for transition zone segmentations. When the models were applied to multiple test set cohorts, the improvement in performance achieved using data augmentation alone was 2.2% for the whole prostate models and 3.0% for the transition zone segmentation models. However, the best test-set results were obtained with models fine-tuned on test center data with mean Dice similarity coefficients of 91.5 for whole prostate segmentation and 89.7 for transition zone segmentation. CONCLUSION. Transfer learning allowed for the development of a high-performing prostate segmentation model, and data augmentation and fine-tuning approaches improved performance of a prostate segmentation model when applied to datasets from external centers.

Introduction: Amyloid cardiomyopathy (ACM) is recognized increasingly as a cause of advanced heart failure. There are limited data describing clinical outcomes for ACM patients who receive durable mechanical circulatory support (MCS). Methods: This analysis included adult patients in the Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) with a primary diagnosis of dilated cardiomyopathy (DCM, N=19,921), non-amyloid restrictive cardiomyopathy (RCM, N=248), or ACM (N=48) who received durable MCS of any type or strategy from 2005-2017. We compared baseline characteristics and adverse events for the 3 groups. The primary endpoint was death with heart transplantation as a competing risk with follow-up censored at device explantation. Results: Patients with ACM (N=48) were older (61 y [IQR 55-69] versus 58 y [IQR 49-66] for DCM and 55 y [IQR 46-62] for non-amyloid RCM, P Conclusion: Compared to DCM or non-amyloid RCM patients, those with durable MCS for ACM have the highest risk of early adverse events and lowest survival. These data highlight concerns with durable MCS for ACM.

Although it is widely accepted that personal networks influence health and illness, network recall remains a major concern. This concern is heightened when studying a population that is vulnerable to cognitive decline. Given these issues, we use data from the Social Network in Alzheimer Disease project to explore similarities and discrepancies between the network perceptions of focal participants and study partners. By leveraging data on a sample of older adults with normal cognition, mild cognitive impairment, and early stage dementia, we explore how cognitive impairment influences older adults’ perceptions of their personal networks. We find that the average individual is more likely to omit weaker, peripheral ties from their self-reported networks than stronger, central ties. Despite observing only moderate levels of focal-partner corroboration across our sample, we find minimal evidence of perceptual differences across diagnostic groups. We offer two broad conclusions. First, self-reported network data, though imperfect, offer a reasonable account of the core people in one’s life. Second, our findings assuage concerns that cognitively impaired older adults have skewed perceptions of their personal networks.

Chest CT is emerging as a valuable diagnostic tool for clinical management of COVID-19 associated lung disease. Artificial intelligence (AI) has the potential to aid in rapid evaluation of CT scans for differentiation of COVID-19 findings from other clinical entities. Here we show that a series of deep learning algorithms, trained in a diverse multinational cohort of 1280 patients to localize parietal pleura/lung parenchyma followed by classification of COVID-19 pneumonia, can achieve up to 90.8% accuracy, with 84% sensitivity and 93% specificity, as evaluated in an independent test set (not included in training and validation) of 1337 patients. Normal controls included chest CTs from oncology, emergency, and pneumonia-related indications. The false positive rate in 140 patients with laboratory confirmed other (non COVID-19) pneumonias was 10%. AI-based algorithms can readily identify CT scans with COVID-19 associated pneumonia, as well as distinguish non-COVID related pneumonias with high specificity in diverse patient populations., Chest CT is emerging as a valuable diagnostic tool for clinical management of COVID-19 associated lung disease. Here, the authors present a multinational study on the application of deep learning algorithms for COVID-19 diagnosis against multiple lung conditions as controls.

Although having achieved great success in medical image segmentation, deep learning-based approaches usually require large amounts of well-annotated data, which can be extremely expensive in the field of medical image analysis. Unlabeled data, on the other hand, is much easier to acquire. Semi-supervised learning and unsupervised domain adaptation both take the advantage of unlabeled data, and they are closely related to each other. In this paper, we propose uncertainty-aware multi-view co-training (UMCT), a unified framework that addresses these two tasks for volumetric medical image segmentation. Our framework is capable of efficiently utilizing unlabeled data for better performance. We firstly rotate and permute the 3D volumes into multiple views and train a 3D deep network on each view. We then apply co-training by enforcing multi-view consistency on unlabeled data, where an uncertainty estimation of each view is utilized to achieve accurate labeling. Experiments on the NIH pancreas segmentation dataset and a multi-organ segmentation dataset show state-of-the-art performance of the proposed framework on semi-supervised medical image segmentation. Under unsupervised domain adaptation settings, we validate the effectiveness of this work by adapting our multi-organ segmentation model to two pathological organs from the Medical Segmentation Decathlon Datasets. Additionally, we show that our UMCT-DA model can even effectively handle the challenging situation where labeled source data is inaccessible, demonstrating strong potentials for real-world applications., 19 pages, 6 figures, to appear in Medical Image Analysis. This article is an extension of the conference paper arXiv:1811.12506

3D convolution neural networks (CNN) have been proved very successful in parsing organs or tumours in 3D medical images, but it remains sophisticated and time-consuming to choose or design proper 3D networks given different task contexts. Recently, Neural Architecture Search (NAS) is proposed to solve this problem by searching for the best network architecture automatically. However, the inconsistency between search stage and deployment stage often exists in NAS algorithms due to memory constraints and large search space, which could become more serious when applying NAS to some memory and time consuming tasks, such as 3D medical image segmentation. In this paper, we propose coarse-to-fine neural architecture search (C2FNAS) to automatically search a 3D segmentation network from scratch without inconsistency on network size or input size. Specifically, we divide the search procedure into two stages: 1) the coarse stage, where we search the macro-level topology of the network, i.e. how each convolution module is connected to other modules; 2) the fine stage, where we search at micro-level for operations in each cell based on previous searched macro-level topology. The coarse-to-fine manner divides the search procedure into two consecutive stages and meanwhile resolves the inconsistency. We evaluate our method on 10 public datasets from Medical Segmentation Decalthon (MSD) challenge, and achieve state-of-the-art performance with the network searched using one dataset, which demonstrates the effectiveness and generalization of our searched models., CVPR 2020

A large literature emphasizes the importance of social relationships during the caregiving process. Yet these issues are seldom presented in a social network framework that examines the structure of caregivers’ personal networks. In this study, I examine how older caregivers experience changes in personal network structure. Using two waves from the National Social Life, Health, and Aging Project, I investigate whether caregivers are more or less likely to exhibit bridging or bonding capital potential compared to noncaregivers. I find that older adults transitioning into caregiving are more likely to develop the ability to bridge social ties within their personal networks than noncaregivers despite potential constraints in their personal freedom. Caregivers in the latter stages, meanwhile, do not differ from noncaregivers in terms of network change. These findings have implications for older adults’ potential to pool resources across social domains as well as negotiate stress and well-being during the caregiving process.