Lazy Resampling: Fast and information preserving preprocessing for deep learning.

BACKGROUND AND OBJECTIVE: Preprocessing of data is a vital step for almost all deep learning workflows. In computer vision, manipulation of data intensity and spatial properties can improve network stability and can provide an important source of generalisation for deep neural networks. Models are frequently trained with preprocessing pipelines composed of many stages, but these pipelines come with a drawback; each stage that resamples the data costs time, degrades image quality, and adds bias to the output. Long pipelines can also be complex to design, especially in medical imaging, where cropping data early can cause significant artifacts. METHODS: We present Lazy Resampling, a software that rephrases spatial preprocessing operations as a graphics pipeline. Rather than each transform individually modifying the data, the transforms generate transform descriptions that are composited together into a single resample operation wherever possible. This reduces pipeline execution time and, most importantly, limits signal degradation. It enables simpler pipeline design as crops and other operations become non-destructive. Lazy Resampling is designed in such a way that it provides the maximum benefit to users without requiring them to understand the underlying concepts or change the way that they build pipelines. RESULTS: We evaluate Lazy Resampling by comparing traditional pipelines and the corresponding lazy resampling pipeline for the following tasks on Medical Segmentation Decathlon datasets. We demonstrate lower information loss in lazy pipelines vs. traditional pipelines. We demonstrate that Lazy Resampling can avoid catastrophic loss of semantic segmentation label accuracy occurring in traditional pipelines when passing labels through a pipeline and then back through the inverted pipeline. Finally, we demonstrate statistically significant improvements when training UNets for semantic segmentation. CONCLUSION: Lazy Resampling reduces the loss of information that occurs when running processing pipelines that traditionally have multiple resampling steps and enables researchers to build simpler pipelines by making operations such as rotation and cropping effectively non-destructive. It makes it possible to invert labels back through a pipeline without catastrophic loss of accuracy. A reference implementation for Lazy Resampling can be found at https://github.com/KCL-BMEIS/LazyResampling. Lazy Resampling is being implemented as a core feature in MONAI, an open source python-based deep learning library for medical imaging, with a roadmap for a full integration.

A Child with COVID-19 Complicated by Rapidly Progressive Severe Organizing Pneumonia: A Case Report.

A 2-year-old boy tested positive for SARS-CoV-2 and, after 30 days of mild-moderate respiratory symptoms, suddenly deteriorated and required extracorporeal membrane oxygenation. Lung biopsy was performed with findings consistent with organizing pneumonia. He received intensive therapy with high-dose methylprednisolone, intravenous immune globulin, rituximab, and plasmapheresis without improvement. He died after 85 days hospitalization. This case highlights unique presentations of COVID-19 and reaffirms the concept that, while rare in Hawai'i, pediatric COVID-19 is an ongoing problem and that severe, even fatal, disease can occur.

MONAI Label: A framework for AI-assisted interactive labeling of 3D medical images.

The lack of annotated datasets is a major bottleneck for training new task-specific supervised machine learning models, considering that manual annotation is extremely expensive and time-consuming. To address this problem, we present MONAI Label, a free and open-source framework that facilitates the development of applications based on artificial intelligence (AI) models that aim at reducing the time required to annotate radiology datasets. Through MONAI Label, researchers can develop AI annotation applications focusing on their domain of expertise. It allows researchers to readily deploy their apps as services, which can be made available to clinicians via their preferred user interface. Currently, MONAI Label readily supports locally installed (3D Slicer) and web-based (OHIF) frontends and offers two active learning strategies to facilitate and speed up the training of segmentation algorithms. MONAI Label allows researchers to make incremental improvements to their AI-based annotation application by making them available to other researchers and clinicians alike. Additionally, MONAI Label provides sample AI-based interactive and non-interactive labeling applications, that can be used directly off the shelf, as plug-and-play to any given dataset. Significant reduced annotation times using the interactive model can be observed on two public datasets.

Do Gradient Inversion Attacks Make Federated Learning Unsafe?

Federated learning (FL) allows the collaborative training of AI models without needing to share raw data. This capability makes it especially interesting for healthcare applications where patient and data privacy is of utmost concern. However, recent works on the inversion of deep neural networks from model gradients raised concerns about the security of FL in preventing the leakage of training data. In this work, we show that these attacks presented in the literature are impractical in FL use-cases where the clients' training involves updating the Batch Normalization (BN) statistics and provide a new baseline attack that works for such scenarios. Furthermore, we present new ways to measure and visualize potential data leakage in FL. Our work is a step towards establishing reproducible methods of measuring data leakage in FL and could help determine the optimal tradeoffs between privacy-preserving techniques, such as differential privacy, and model accuracy based on quantifiable metrics.

miR-15b-5p promotes expression of proinflammatory cytokines in human placenta by inhibiting Apelin signaling pathway.

INTRODUCTION: The timing of parturition at end of human gestation may be controlled by fetal signals. The signaling molecules contributing to activation of human labor may be mediated by fetal exosomes. In this study, we focused on investigation of the role of microRNAs (miRNAs) derived from fetal exosomes in the regulation of human placental gene expression. METHODS: Using immunofluorescent labeling, array-based miRNA profiling assay, target prediction analysis, and conducting a variety of biochemical approaches including miRNA mimics, dual-luciferase, siRNA-mediated gene silencing, and immunohistochemical staining assay in primary trophoblast culture and formalin-fixed paraffin-embedded placental tissues, we examined whether fetal exosomal miRNAs can stimulate expression of proinflammatory mediators in human placenta. RESULTS: We showed placental uptake of exosomes derived from the umbilical artery, and found that 9 fetal exosomal miRNAs: let-7i-5p, miR-185-5p, miR-15b-5p, miR-376c-3p, miR-548d-5p, miR-92b-3p, miR-16-5p, and miR-1301-3p were significantly increased in placentas of women delivering following term labor compared to those delivering by Cesarean section in the late preterm period. Target prediction analysis identified miR-15b-5p of particular interest, because one of its predicted targets is Apelin, a potent inhibitor of proinflammatory mediators. We further found that miR-15b-5p repressed Apelin protein levels and activated pro-labor hormones and cytokines including IL-1, IL-6, IL-8, and TNF-α. DISCUSSION: These data suggest a potential fetal-to-placental signal that could play a role in the length of human gestation and onset of human labor.

Pediatric Code Blue: How Prepared Are We? A Self-Efficacy Assessment Project.

Pediatric advanced life support (PALS) recertification every two years is inadequate to maintain proficiency. The authors hypothesized that a standardized, recurring curriculum may enhance retention of cardiopulmonary resuscitation (CPR) skills. Monthly in situ mock code training and an annual online self-directed learning module were implemented for pediatric intensive care unit nurses, pediatric residents, and respiratory therapists at a women and children's hospital. The in situ mock codes were linked to PALS training self-efficacy (pre- and post-mock code) and feedback related surveys. CPR knowledge was assessed using an online module with pre- and post-tests. A total of 82 in situ mock code surveys and 137 online modules were completed over a 20-month period. Medical knowledge (P < .05 for 7/10 questions) and self-confidence improved (P < .001. Several staff reported a negative impact on their patient care assignments in order to participate in the mock code. However, a significant number of participants (65%) concurred with the benefits of monthly mock codes. The curriculum improved CPR efficacy by improving knowledge-based retention as well as self-confidence in their skills.

Online Learning and Residents' Acquisition of Mechanical Ventilation Knowledge: Sequencing Matters.

OBJECTIVE: Rapid advancements in medicine and changing standards in medical education require new, efficient educational strategies. We investigated whether an online intervention could increase residents' knowledge and improve knowledge retention in mechanical ventilation when compared with a clinical rotation and whether the timing of intervention had an impact on overall knowledge gains. DESIGN: A prospective, interventional crossover study conducted from October 2015 to December 2017. SETTING: Multicenter study conducted in 33 PICUs across eight countries. SUBJECTS: Pediatric categorical residents rotating through the PICU for the first time. We allocated 483 residents into two arms based on rotation date to use an online intervention either before or after the clinical rotation. INTERVENTIONS: Residents completed an online virtual mechanical ventilation simulator either before or after a 1-month clinical rotation with a 2-month period between interventions. MEASUREMENTS AND MAIN RESULTS: Performance on case-based, multiple-choice question tests before and after each intervention was used to quantify knowledge gains and knowledge retention. Initial knowledge gains in residents who completed the online intervention (average knowledge gain, 6.9%; SD, 18.2) were noninferior compared with those who completed 1 month of a clinical rotation (average knowledge gain, 6.1%; SD, 18.9; difference, 0.8%; 95% CI, -5.05 to 6.47; p = 0.81). Knowledge retention was greater following completion of the online intervention when compared with the clinical rotation when controlling for time (difference, 7.6%; 95% CI, 0.7-14.5; p = 0.03). When the online intervention was sequenced before (average knowledge gain, 14.6%; SD, 15.4) rather than after (average knowledge gain, 7.0%; SD, 19.1) the clinical rotation, residents had superior overall knowledge acquisition (difference, 7.6%; 95% CI, 2.01-12.97;p = 0.008). CONCLUSIONS: Incorporating an interactive online educational intervention prior to a clinical rotation may offer a strategy to prime learners for the upcoming rotation, augmenting clinical learning in graduate medical education.

L445P mutation on heavy chain stabilizes IgG4 under acidic conditions.

IgG4, a common type of therapeutic antibody, is less stable during manufacturing processes compared with IgG1. Aggregation and fragmentation are the two main challenges. Here, we report instability of the heavy chain (HC) C-terminal region under acidic conditions, which leads to cleavage and aggregation. Leu445, at the C-terminal region of the HC in IgG4, plays a critical role in its acid-induced fragmentation and subsequent aggregation. We found that mutating HC C-terminal Leu445 to Pro (the corresponding residue in IgG1) in IgG4_CDR-X significantly reduces fragmentation and aggregation, while mutating Pro445 to Leu in IgG1_CDR-X promotes fragmentation and aggregation. HC C-terminal Gly446 cleavage was observed in low pH citrate buffer and resulted in further fragmentation and aggregation, whereas, glycine buffer can completely inhibit the cleavage and aggregation. It is proposed that cleavages occur through acid-induced hydrolysis under acidic conditions and glycine stabilizes IgG4 via two main mechanisms: 1) product feedback inhibition of the hydrolysis reaction, and 2) stabilization of protein conformation by direct interaction with the peptide backbone and charged side chains. Experiments using IgG4 molecules IgG4_CDR-Y and IgG4_CDR-Z with the same CH domains as IgG4_CDR-X, but different complementarity-determining regions (CDRs), indicate that the stability of the HC C-terminal region is also closely related to the sequence of the CDRs. The stability of IgG4_CDR-X is significantly improved when binding to its target. Both observations suggest that there are potential interactions between Fab and CH2-CH3 domains, which could be the key factor affecting the stability of IgG antibodies.

Privacy-preserving Federated Brain Tumour Segmentation.

Due to medical data privacy regulations, it is often infeasible to collect and share patient data in a centralised data lake. This poses challenges for training machine learning algorithms, such as deep convolutional networks, which often require large numbers of diverse training examples. Federated learning sidesteps this difficulty by bringing code to the patient data owners and only sharing intermediate model training updates among them. Although a high-accuracy model could be achieved by appropriately aggregating these model updates, the model shared could indirectly leak the local training examples. In this paper, we investigate the feasibility of applying differential-privacy techniques to protect the patient data in a federated learning setup. We implement and evaluate practical federated learning systems for brain tumour segmentation on the BraTS dataset. The experimental results show that there is a trade-off between model performance and privacy protection costs.