Search

Post-training quantization is widely employed to reduce the computational demands of neural networks. Typically, individual substructures, such as layers or blocks of layers, are quantized with the objective of minimizing quantization errors in their pre-activations by fine-tuning the corresponding weights. Deriving this local objective from the global objective of minimizing task loss involves two key simplifications: assuming substructures are mutually independent and ignoring the knowledge of subsequent substructures as well as the task loss. In this work, we assess the effects of these simplifications on weight-only quantization of large language models. We introduce two multi-block fine-tuning strategies and compare them against the baseline of fine-tuning single transformer blocks. The first captures correlations of weights across blocks by jointly optimizing multiple quantized blocks. The second incorporates knowledge of subsequent blocks by minimizing the error in downstream pre-activations rather than focusing solely on the quantized block. Our findings indicate that the effectiveness of these methods depends on the specific network model, with no impact on some models but demonstrating significant benefits for others.

Stroke remains a leading cause of global morbidity and mortality, placing a heavy socioeconomic burden. Over the past decade, advances in endovascular reperfusion therapy and the use of CT and MRI imaging for treatment guidance have significantly improved patient outcomes and are now standard in clinical practice. To develop machine learning algorithms that can extract meaningful and reproducible models of brain function for both clinical and research purposes from stroke images - particularly for lesion identification, brain health quantification, and prognosis - large, diverse, and well-annotated public datasets are essential. While only a few datasets with (sub-)acute stroke data were previously available, several large, high-quality datasets have recently been made publicly accessible. However, these existing datasets include only MRI data. In contrast, our dataset is the first to offer comprehensive longitudinal stroke data, including acute CT imaging with angiography and perfusion, follow-up MRI at 2-9 days, as well as acute and longitudinal clinical data up to a three-month outcome. The dataset includes a training dataset of n = 150 and a test dataset of n = 100 scans. Training data is publicly available, while test data will be used exclusively for model validation. We are making this dataset available as part of the 2024 edition of the Ischemic Stroke Lesion Segmentation (ISLES) challenge (https://www.isles-challenge.org/), which continuously aims to establish benchmark methods for acute and sub-acute ischemic stroke lesion segmentation, aiding in creating open stroke imaging datasets and evaluating cutting-edge image processing algorithms.

Accurate estimation of core (irreversibly damaged tissue) and penumbra (salvageable tissue) volumes is essential for ischemic stroke treatment decisions. Perfusion CT, the clinical standard, estimates these volumes but is affected by variations in deconvolution algorithms, implementations, and thresholds. Core tissue expands over time, with growth rates influenced by thrombus location, collateral circulation, and inherent patient-specific factors. Understanding this tissue growth is crucial for determining the need to transfer patients to comprehensive stroke centers, predicting the benefits of additional reperfusion attempts during mechanical thrombectomy, and forecasting final clinical outcomes. This work presents the ISLES'24 challenge, which addresses final post-treatment stroke infarct prediction from pre-interventional acute stroke imaging and clinical data. ISLES'24 establishes a unique 360-degree setting where all feasibly accessible clinical data are available for participants, including full CT acute stroke imaging, sub-acute follow-up MRI, and clinical tabular data. The contributions of this work are two-fold: first, we introduce a standardized benchmarking of final stroke infarct segmentation algorithms through the ISLES'24 challenge; second, we provide insights into infarct segmentation using multimodal imaging and clinical data strategies by identifying outperforming methods on a finely curated dataset. The outputs of this challenge are anticipated to enhance clinical decision-making and improve patient outcome predictions. All ISLES'24 materials, including data, performance evaluation scripts, and leading algorithmic strategies, are available to the research community following \url{https://isles-24.grand-challenge.org/}.

Vision-language artificial intelligence models (VLMs) possess medical knowledge and can be employed in healthcare in numerous ways, including as image interpreters, virtual scribes, and general decision support systems. However, here, we demonstrate that current VLMs applied to medical tasks exhibit a fundamental security flaw: they can be attacked by prompt injection attacks, which can be used to output harmful information just by interacting with the VLM, without any access to its parameters. We performed a quantitative study to evaluate the vulnerabilities to these attacks in four state of the art VLMs which have been proposed to be of utility in healthcare: Claude 3 Opus, Claude 3.5 Sonnet, Reka Core, and GPT-4o. Using a set of N=297 attacks, we show that all of these models are susceptible. Specifically, we show that embedding sub-visual prompts in medical imaging data can cause the model to provide harmful output, and that these prompts are non-obvious to human observers. Thus, our study demonstrates a key vulnerability in medical VLMs which should be mitigated before widespread clinical adoption., Comment: 57 Pages, 5 Figures

Matching cancer patients to clinical trials is essential for advancing treatment and patient care. However, the inconsistent format of medical free text documents and complex trial eligibility criteria make this process extremely challenging and time-consuming for physicians. We investigated whether the entire trial matching process - from identifying relevant trials among 105,600 oncology-related clinical trials on clinicaltrials.gov to generating criterion-level eligibility matches - could be automated using Large Language Models (LLMs). Using GPT-4o and a set of 51 synthetic Electronic Health Records (EHRs), we demonstrate that our approach identifies relevant candidate trials in 93.3% of cases and achieves a preliminary accuracy of 88.0% when matching patient-level information at the criterion level against a baseline defined by human experts. Utilizing LLM feedback reveals that 39.3% criteria that were initially considered incorrect are either ambiguous or inaccurately annotated, leading to a total model accuracy of 92.7% after refining our human baseline. In summary, we present an end-to-end pipeline for clinical trial matching using LLMs, demonstrating high precision in screening and matching trials to individual patients, even outperforming the performance of qualified medical doctors. Our fully end-to-end pipeline can operate autonomously or with human supervision and is not restricted to oncology, offering a scalable solution for enhancing patient-trial matching in real-world settings., Comment: 149 pages, including Supplements. 3 Main Figures

Recent advances in large language models (LLMs) have increased the demand for comprehensive benchmarks to evaluate their capabilities as human-like agents. Existing benchmarks, while useful, often focus on specific application scenarios, emphasizing task completion but failing to dissect the underlying skills that drive these outcomes. This lack of granularity makes it difficult to deeply discern where failures stem from. Additionally, setting up these environments requires considerable effort, and issues of unreliability and reproducibility sometimes arise, especially in interactive tasks. To address these limitations, we introduce the Massive Multitask Agent Understanding (MMAU) benchmark, featuring comprehensive offline tasks that eliminate the need for complex environment setups. It evaluates models across five domains, including Tool-use, Directed Acyclic Graph (DAG) QA, Data Science and Machine Learning coding, Contest-level programming and Mathematics, and covers five essential capabilities: Understanding, Reasoning, Planning, Problem-solving, and Self-correction. With a total of 20 meticulously designed tasks encompassing over 3K distinct prompts, MMAU provides a comprehensive framework for evaluating the strengths and limitations of LLM agents. By testing 18 representative models on MMAU, we provide deep and insightful analyses. Ultimately, MMAU not only sheds light on the capabilities and limitations of LLM agents but also enhances the interpretability of their performance. Datasets and evaluation scripts of MMAU are released at https://github.com/apple/axlearn/tree/main/docs/research/mmau.

Large Language Models (LLMs) with their strong task-handling capabilities have shown remarkable advancements across a spectrum of fields, moving beyond natural language understanding. However, their proficiency within the chemistry domain remains restricted, especially in solving professional molecule-related tasks. This challenge is attributed to their inherent limitations in comprehending molecules using only common textual representations, i.e., SMILES strings. In this study, we seek to enhance the ability of LLMs to comprehend molecules by equipping them with a multi-modal external module, namely MolX. In particular, instead of directly using a SMILES string to represent a molecule, we utilize specific encoders to extract fine-grained features from both SMILES string and 2D molecular graph representations for feeding into an LLM. Moreover, a handcrafted molecular fingerprint is incorporated to leverage its embedded domain knowledge. Then, to establish an alignment between MolX and the LLM's textual input space, the whole model in which the LLM is frozen, is pre-trained with a versatile strategy including a diverse set of tasks. Experimental evaluations show that our proposed method outperforms baselines across 4 downstream molecule-related tasks ranging from molecule-to-text translation to retrosynthesis, with and without fine-tuning the LLM, while only introducing a small number of trainable parameters 0.53% and 0.82%, respectively.

Multimodal artificial intelligence (AI) systems have the potential to enhance clinical decision-making by interpreting various types of medical data. However, the effectiveness of these models across all medical fields is uncertain. Each discipline presents unique challenges that need to be addressed for optimal performance. This complexity is further increased when attempting to integrate different fields into a single model. Here, we introduce an alternative approach to multimodal medical AI that utilizes the generalist capabilities of a large language model (LLM) as a central reasoning engine. This engine autonomously coordinates and deploys a set of specialized medical AI tools. These tools include text, radiology and histopathology image interpretation, genomic data processing, web searches, and document retrieval from medical guidelines. We validate our system across a series of clinical oncology scenarios that closely resemble typical patient care workflows. We show that the system has a high capability in employing appropriate tools (97%), drawing correct conclusions (93.6%), and providing complete (94%), and helpful (89.2%) recommendations for individual patient cases while consistently referencing relevant literature (82.5%) upon instruction. This work provides evidence that LLMs can effectively plan and execute domain-specific models to retrieve or synthesize new information when used as autonomous agents. This enables them to function as specialist, patient-tailored clinical assistants. It also simplifies regulatory compliance by allowing each component tool to be individually validated and approved. We believe, that our work can serve as a proof-of-concept for more advanced LLM-agents in the medical domain., Comment: 91 pages, 2 Figures

Diffusion-weighted MRI (DWI) is essential for stroke diagnosis, treatment decisions, and prognosis. However, image and disease variability hinder the development of generalizable AI algorithms with clinical value. We address this gap by presenting a novel ensemble algorithm derived from the 2022 Ischemic Stroke Lesion Segmentation (ISLES) challenge. ISLES'22 provided 400 patient scans with ischemic stroke from various medical centers, facilitating the development of a wide range of cutting-edge segmentation algorithms by the research community. Through collaboration with leading teams, we combined top-performing algorithms into an ensemble model that overcomes the limitations of individual solutions. Our ensemble model achieved superior ischemic lesion detection and segmentation accuracy on our internal test set compared to individual algorithms. This accuracy generalized well across diverse image and disease variables. Furthermore, the model excelled in extracting clinical biomarkers. Notably, in a Turing-like test, neuroradiologists consistently preferred the algorithm's segmentations over manual expert efforts, highlighting increased comprehensiveness and precision. Validation using a real-world external dataset (N=1686) confirmed the model's generalizability. The algorithm's outputs also demonstrated strong correlations with clinical scores (admission NIHSS and 90-day mRS) on par with or exceeding expert-derived results, underlining its clinical relevance. This study offers two key findings. First, we present an ensemble algorithm (https://github.com/Tabrisrei/ISLES22_Ensemble) that detects and segments ischemic stroke lesions on DWI across diverse scenarios on par with expert (neuro)radiologists. Second, we show the potential for biomedical challenge outputs to extend beyond the challenge's initial objectives, demonstrating their real-world clinical applicability.

Medical image classification requires labeled, task-specific datasets which are used to train deep learning networks de novo, or to fine-tune foundation models. However, this process is computationally and technically demanding. In language processing, in-context learning provides an alternative, where models learn from within prompts, bypassing the need for parameter updates. Yet, in-context learning remains underexplored in medical image analysis. Here, we systematically evaluate the model Generative Pretrained Transformer 4 with Vision capabilities (GPT-4V) on cancer image processing with in-context learning on three cancer histopathology tasks of high importance: Classification of tissue subtypes in colorectal cancer, colon polyp subtyping and breast tumor detection in lymph node sections. Our results show that in-context learning is sufficient to match or even outperform specialized neural networks trained for particular tasks, while only requiring a minimal number of samples. In summary, this study demonstrates that large vision language models trained on non-domain specific data can be applied out-of-the box to solve medical image-processing tasks in histopathology. This democratizes access of generalist AI models to medical experts without technical background especially for areas where annotated data is scarce., Comment: 40 pages, 5 figures

The yield of a chemical reaction quantifies the percentage of the target product formed in relation to the reactants consumed during the chemical reaction. Accurate yield prediction can guide chemists toward selecting high-yield reactions during synthesis planning, offering valuable insights before dedicating time and resources to wet lab experiments. While recent advancements in yield prediction have led to overall performance improvement across the entire yield range, an open challenge remains in enhancing predictions for high-yield reactions, which are of greater concern to chemists. In this paper, we argue that the performance gap in high-yield predictions results from the imbalanced distribution of real-world data skewed towards low-yield reactions, often due to unreacted starting materials and inherent ambiguities in the reaction processes. Despite this data imbalance, existing yield prediction methods continue to treat different yield ranges equally, assuming a balanced training distribution. Through extensive experiments on three real-world yield prediction datasets, we emphasize the urgent need to reframe reaction yield prediction as an imbalanced regression problem. Finally, we demonstrate that incorporating simple cost-sensitive re-weighting methods can significantly enhance the performance of yield prediction models on underrepresented high-yield regions.

Large Language Models (LLMs) have achieved remarkable success across a wide array of tasks. Due to the impressive planning and reasoning abilities of LLMs, they have been used as autonomous agents to do many tasks automatically. Recently, based on the development of using one LLM as a single planning or decision-making agent, LLM-based multi-agent systems have achieved considerable progress in complex problem-solving and world simulation. To provide the community with an overview of this dynamic field, we present this survey to offer an in-depth discussion on the essential aspects of multi-agent systems based on LLMs, as well as the challenges. Our goal is for readers to gain substantial insights on the following questions: What domains and environments do LLM-based multi-agents simulate? How are these agents profiled and how do they communicate? What mechanisms contribute to the growth of agents' capacities? For those interested in delving into this field of study, we also summarize the commonly used datasets or benchmarks for them to have convenient access. To keep researchers updated on the latest studies, we maintain an open-source GitHub repository, dedicated to outlining the research on LLM-based multi-agent systems., Comment: This work is ongoing and we welcome your contribution!

The Circle of Willis (CoW) is an important network of arteries connecting major circulations of the brain. Its vascular architecture is believed to affect the risk, severity, and clinical outcome of serious neuro-vascular diseases. However, characterizing the highly variable CoW anatomy is still a manual and time-consuming expert task. The CoW is usually imaged by two angiographic imaging modalities, magnetic resonance angiography (MRA) and computed tomography angiography (CTA), but there exist limited public datasets with annotations on CoW anatomy, especially for CTA. Therefore we organized the TopCoW Challenge in 2023 with the release of an annotated CoW dataset. The TopCoW dataset was the first public dataset with voxel-level annotations for thirteen possible CoW vessel components, enabled by virtual-reality (VR) technology. It was also the first large dataset with paired MRA and CTA from the same patients. TopCoW challenge formalized the CoW characterization problem as a multiclass anatomical segmentation task with an emphasis on topological metrics. We invited submissions worldwide for the CoW segmentation task, which attracted over 140 registered participants from four continents. The top performing teams managed to segment many CoW components to Dice scores around 90%, but with lower scores for communicating arteries and rare variants. There were also topological mistakes for predictions with high Dice scores. Additional topological analysis revealed further areas for improvement in detecting certain CoW components and matching CoW variant topology accurately. TopCoW represented a first attempt at benchmarking the CoW anatomical segmentation task for MRA and CTA, both morphologically and topologically., Comment: 24 pages, 11 figures, 9 tables. Summary Paper for the MICCAI TopCoW 2023 Challenge

Reaction prediction, a critical task in synthetic chemistry, is to predict the outcome of a reaction based on given reactants. Generative models like Transformer and VAE have typically been employed to predict the reaction product. However, these likelihood-maximization models overlooked the inherent stochastic nature of chemical reactions, such as the multiple ways electrons can be redistributed among atoms during the reaction process. In scenarios where similar reactants could follow different electron redistribution patterns, these models typically predict the most common outcomes, neglecting less frequent but potentially crucial reaction patterns. These overlooked patterns, though rare, can lead to innovative methods for designing synthetic routes and significantly advance synthesis techniques. To break the limits of previous approaches, we propose organizing the mapping space between reactants and electron redistribution patterns in a divide-and-conquer manner. We address the reaction problem by training multiple expert models, each specializing in capturing a type of electron redistribution pattern in reaction. These experts enhance the prediction process by considering both typical and other less common electron redistribution manners. In the inference stage, a dropout strategy is applied to each expert to improve the electron redistribution diversity. The most plausible products are finally predicted through a ranking stage designed to integrate the predictions from multiple experts. Experimental results on the largest reaction prediction benchmark USPTO-MIT show the superior performance of our proposed method compared to baselines.

Molecular Representation Learning (MRL) has proven impactful in numerous biochemical applications such as drug discovery and enzyme design. While Graph Neural Networks (GNNs) are effective at learning molecular representations from a 2D molecular graph or a single 3D structure, existing works often overlook the flexible nature of molecules, which continuously interconvert across conformations via chemical bond rotations and minor vibrational perturbations. To better account for molecular flexibility, some recent works formulate MRL as an ensemble learning problem, focusing on explicitly learning from a set of conformer structures. However, most of these studies have limited datasets, tasks, and models. In this work, we introduce the first MoleculAR Conformer Ensemble Learning (MARCEL) benchmark to thoroughly evaluate the potential of learning on conformer ensembles and suggest promising research directions. MARCEL includes four datasets covering diverse molecule- and reaction-level properties of chemically diverse molecules including organocatalysts and transition-metal catalysts, extending beyond the scope of common GNN benchmarks that are confined to drug-like molecules. In addition, we conduct a comprehensive empirical study, which benchmarks representative 1D, 2D, and 3D molecular representation learning models, along with two strategies that explicitly incorporate conformer ensembles into 3D MRL models. Our findings reveal that direct learning from an accessible conformer space can improve performance on a variety of tasks and models., Comment: ICLR 2024

Large Language Models (LLMs) with strong abilities in natural language processing tasks have emerged and have been applied in various kinds of areas such as science, finance and software engineering. However, the capability of LLMs to advance the field of chemistry remains unclear. In this paper, rather than pursuing state-of-the-art performance, we aim to evaluate capabilities of LLMs in a wide range of tasks across the chemistry domain. We identify three key chemistry-related capabilities including understanding, reasoning and explaining to explore in LLMs and establish a benchmark containing eight chemistry tasks. Our analysis draws on widely recognized datasets facilitating a broad exploration of the capacities of LLMs within the context of practical chemistry. Five LLMs (GPT-4, GPT-3.5, Davinci-003, Llama and Galactica) are evaluated for each chemistry task in zero-shot and few-shot in-context learning settings with carefully selected demonstration examples and specially crafted prompts. Our investigation found that GPT-4 outperformed other models and LLMs exhibit different competitive levels in eight chemistry tasks. In addition to the key findings from the comprehensive benchmark analysis, our work provides insights into the limitation of current LLMs and the impact of in-context learning settings on LLMs' performance across various chemistry tasks. The code and datasets used in this study are available at https://github.com/ChemFoundationModels/ChemLLMBench., Comment: NeurIPS 2023 Datasets and Benchmarks Track camera-ready version

Automated brain tumor segmentation methods have become well-established and reached performance levels offering clear clinical utility. These methods typically rely on four input magnetic resonance imaging (MRI) modalities: T1-weighted images with and without contrast enhancement, T2-weighted images, and FLAIR images. However, some sequences are often missing in clinical practice due to time constraints or image artifacts, such as patient motion. Consequently, the ability to substitute missing modalities and gain segmentation performance is highly desirable and necessary for the broader adoption of these algorithms in the clinical routine. In this work, we present the establishment of the Brain MR Image Synthesis Benchmark (BraSyn) in conjunction with the Medical Image Computing and Computer-Assisted Intervention (MICCAI) 2023. The primary objective of this challenge is to evaluate image synthesis methods that can realistically generate missing MRI modalities when multiple available images are provided. The ultimate aim is to facilitate automated brain tumor segmentation pipelines. The image dataset used in the benchmark is diverse and multi-modal, created through collaboration with various hospitals and research institutions., Comment: Technical report of BraSyn

A myriad of algorithms for the automatic analysis of brain MR images is available to support clinicians in their decision-making. For brain tumor patients, the image acquisition time series typically starts with an already pathological scan. This poses problems, as many algorithms are designed to analyze healthy brains and provide no guarantee for images featuring lesions. Examples include, but are not limited to, algorithms for brain anatomy parcellation, tissue segmentation, and brain extraction. To solve this dilemma, we introduce the BraTS inpainting challenge. Here, the participants explore inpainting techniques to synthesize healthy brain scans from lesioned ones. The following manuscript contains the task formulation, dataset, and submission procedure. Later, it will be updated to summarize the findings of the challenge. The challenge is organized as part of the ASNR-BraTS MICCAI challenge., Comment: 14 pages, 6 figures

Fully compressible magnetohydrodynamic (MHD) simulations are a fundamental tool for investigating the role of dynamo amplification in the generation of magnetic fields in deep convective layers of stars. The flows that arise in such environments are characterized by low (sonic) Mach numbers (M_son < 0.01 ). In these regimes, conventional MHD codes typically show excessive dissipation and tend to be inefficient as the Courant-Friedrichs-Lewy (CFL) constraint on the time step becomes too strict. In this work we present a new method for efficiently simulating MHD flows at low Mach numbers in a space-dependent gravitational potential while still retaining all effects of compressibility. The proposed scheme is implemented in the finite-volume Seven-League Hydro (SLH) code, and it makes use of a low-Mach version of the five-wave Harten-Lax-van Leer discontinuities (HLLD) solver to reduce numerical dissipation, an implicit-explicit time discretization technique based on Strang splitting to overcome the overly strict CFL constraint, and a well-balancing method that dramatically reduces the magnitude of spatial discretization errors in strongly stratified setups. The solenoidal constraint on the magnetic field is enforced by using a constrained transport method on a staggered grid. We carry out five verification tests, including the simulation of a small-scale dynamo in a star-like environment at M_son ~ 0.001 . We demonstrate that the proposed scheme can be used to accurately simulate compressible MHD flows in regimes of low Mach numbers and strongly stratified setups even with moderately coarse grids.

Molecular representation learning (MRL) is a key step to build the connection between machine learning and chemical science. In particular, it encodes molecules as numerical vectors preserving the molecular structures and features, on top of which the downstream tasks (e.g., property prediction) can be performed. Recently, MRL has achieved considerable progress, especially in methods based on deep molecular graph learning. In this survey, we systematically review these graph-based molecular representation techniques, especially the methods incorporating chemical domain knowledge. Specifically, we first introduce the features of 2D and 3D molecular graphs. Then we summarize and categorize MRL methods into three groups based on their input. Furthermore, we discuss some typical chemical applications supported by MRL. To facilitate studies in this fast-developing area, we also list the benchmarks and commonly used datasets in the paper. Finally, we share our thoughts on future research directions.

Magnetic resonance imaging (MRI) is a central modality for stroke imaging. It is used upon patient admission to make treatment decisions such as selecting patients for intravenous thrombolysis or endovascular therapy. MRI is later used in the duration of hospital stay to predict outcome by visualizing infarct core size and location. Furthermore, it may be used to characterize stroke etiology, e.g. differentiation between (cardio)-embolic and non-embolic stroke. Computer based automated medical image processing is increasingly finding its way into clinical routine. Previous iterations of the Ischemic Stroke Lesion Segmentation (ISLES) challenge have aided in the generation of identifying benchmark methods for acute and sub-acute ischemic stroke lesion segmentation. Here we introduce an expert-annotated, multicenter MRI dataset for segmentation of acute to subacute stroke lesions. This dataset comprises 400 multi-vendor MRI cases with high variability in stroke lesion size, quantity and location. It is split into a training dataset of n=250 and a test dataset of n=150. All training data will be made publicly available. The test dataset will be used for model validation only and will not be released to the public. This dataset serves as the foundation of the ISLES 2022 challenge with the goal of finding algorithmic methods to enable the development and benchmarking of robust and accurate segmentation algorithms for ischemic stroke., Comment: 12 pages, 2 figures

Although machine learning (ML) has shown promise in numerous domains, there are concerns about generalizability to out-of-sample data. This is currently addressed by centrally sharing ample, and importantly diverse, data from multiple sites. However, such centralization is challenging to scale (or even not feasible) due to various limitations. Federated ML (FL) provides an alternative to train accurate and generalizable ML models, by only sharing numerical model updates. Here we present findings from the largest FL study to-date, involving data from 71 healthcare institutions across 6 continents, to generate an automatic tumor boundary detector for the rare disease of glioblastoma, utilizing the largest dataset of such patients ever used in the literature (25,256 MRI scans from 6,314 patients). We demonstrate a 33% improvement over a publicly trained model to delineate the surgically targetable tumor, and 23% improvement over the tumor's entire extent. We anticipate our study to: 1) enable more studies in healthcare informed by large and diverse data, ensuring meaningful results for rare diseases and underrepresented populations, 2) facilitate further quantitative analyses for glioblastoma via performance optimization of our consensus model for eventual public release, and 3) demonstrate the effectiveness of FL at such scale and task complexity as a paradigm shift for multi-site collaborations, alleviating the need for data sharing., Comment: federated learning, deep learning, convolutional neural network, segmentation, brain tumor, glioma, glioblastoma, FeTS, BraTS

Extracting quantitative information of neuronal signals by non-invasive imaging is an outstanding challenge for understanding brain function and pathology. However, state-of-the-art techniques offer low sensitivity to deep electrical sources. Stimulus induced rotary saturation (SIRS) is a recently proposed magnetic resonance imaging (MRI) sequence that detects oscillatory magnetic fields using a spin-lock preparation. Phantom experiments and simulations proved its efficiency and sensitivity, but the susceptibility of the method to field inhomogeneities is still not well understood. In this study, we simulated and analyzed the dynamic of three spin-lock preparations and their response to field inhomogeneities in the presence of a resonant oscillating field. We show that the composite spin-lock preparation is more robust against field variations within the double resonance effect. In addition, we tested the capability of the chosen composite spin-lock preparation to recover information about the spectral components of a composite signal. This study sets the bases to move one step further towards the clinical application of MR-based neuronal current imaging., Comment: 16 pages, 9 figures

Deep learning (DL) models have provided state-of-the-art performance in various medical imaging benchmarking challenges, including the Brain Tumor Segmentation (BraTS) challenges. However, the task of focal pathology multi-compartment segmentation (e.g., tumor and lesion sub-regions) is particularly challenging, and potential errors hinder translating DL models into clinical workflows. Quantifying the reliability of DL model predictions in the form of uncertainties could enable clinical review of the most uncertain regions, thereby building trust and paving the way toward clinical translation. Several uncertainty estimation methods have recently been introduced for DL medical image segmentation tasks. Developing scores to evaluate and compare the performance of uncertainty measures will assist the end-user in making more informed decisions. In this study, we explore and evaluate a score developed during the BraTS 2019 and BraTS 2020 task on uncertainty quantification (QU-BraTS) and designed to assess and rank uncertainty estimates for brain tumor multi-compartment segmentation. This score (1) rewards uncertainty estimates that produce high confidence in correct assertions and those that assign low confidence levels at incorrect assertions, and (2) penalizes uncertainty measures that lead to a higher percentage of under-confident correct assertions. We further benchmark the segmentation uncertainties generated by 14 independent participating teams of QU-BraTS 2020, all of which also participated in the main BraTS segmentation task. Overall, our findings confirm the importance and complementary value that uncertainty estimates provide to segmentation algorithms, highlighting the need for uncertainty quantification in medical image analyses. Finally, in favor of transparency and reproducibility, our evaluation code is made publicly available at: https://github.com/RagMeh11/QU-BraTS., Comment: Accepted for publication at the Journal of Machine Learning for Biomedical Imaging (MELBA): https://www.melba-journal.org/papers/2022:026.html

The BraTS 2021 challenge celebrates its 10th anniversary and is jointly organized by the Radiological Society of North America (RSNA), the American Society of Neuroradiology (ASNR), and the Medical Image Computing and Computer Assisted Interventions (MICCAI) society. Since its inception, BraTS has been focusing on being a common benchmarking venue for brain glioma segmentation algorithms, with well-curated multi-institutional multi-parametric magnetic resonance imaging (mpMRI) data. Gliomas are the most common primary malignancies of the central nervous system, with varying degrees of aggressiveness and prognosis. The RSNA-ASNR-MICCAI BraTS 2021 challenge targets the evaluation of computational algorithms assessing the same tumor compartmentalization, as well as the underlying tumor's molecular characterization, in pre-operative baseline mpMRI data from 2,040 patients. Specifically, the two tasks that BraTS 2021 focuses on are: a) the segmentation of the histologically distinct brain tumor sub-regions, and b) the classification of the tumor's O[6]-methylguanine-DNA methyltransferase (MGMT) promoter methylation status. The performance evaluation of all participating algorithms in BraTS 2021 will be conducted through the Sage Bionetworks Synapse platform (Task 1) and Kaggle (Task 2), concluding in distributing to the top ranked participants monetary awards of $60,000 collectively., Comment: 19 pages, 2 figures, 1 table

We prove that in the Cayley graph of any braid group modulo its center $B_n/Z(B_n)$, equipped with Garside's generating set, the axes of all pseudo-Anosov braids are strongly contracting. More generally, we consider a Garside group $G$ of finite type with cyclic center. We prove that in the Cayley graph of $G/Z(G)$, equipped with the Garside generators, the axis of any Morse element is strongly contracting. As a consequence, we prove that Morse elements act loxodromically on the additional length graph of $G$., Comment: 25 pages, 9 figures

This manuscript describes the first challenge on Federated Learning, namely the Federated Tumor Segmentation (FeTS) challenge 2021. International challenges have become the standard for validation of biomedical image analysis methods. However, the actual performance of participating (even the winning) algorithms on "real-world" clinical data often remains unclear, as the data included in challenges are usually acquired in very controlled settings at few institutions. The seemingly obvious solution of just collecting increasingly more data from more institutions in such challenges does not scale well due to privacy and ownership hurdles. Towards alleviating these concerns, we are proposing the FeTS challenge 2021 to cater towards both the development and the evaluation of models for the segmentation of intrinsically heterogeneous (in appearance, shape, and histology) brain tumors, namely gliomas. Specifically, the FeTS 2021 challenge uses clinically acquired, multi-institutional magnetic resonance imaging (MRI) scans from the BraTS 2020 challenge, as well as from various remote independent institutions included in the collaborative network of a real-world federation (https://www.fets.ai/). The goals of the FeTS challenge are directly represented by the two included tasks: 1) the identification of the optimal weight aggregation approach towards the training of a consensus model that has gained knowledge via federated learning from multiple geographically distinct institutions, while their data are always retained within each institution, and 2) the federated evaluation of the generalizability of brain tumor segmentation models "in the wild", i.e. on data from institutional distributions that were not part of the training datasets.

Stars form by accreting material from their surrounding disks. There is a consensus that matter flowing through the disk is channelled onto the stellar surface by the stellar magnetic field. This is thought to be strong enough to truncate the disk close to the so-called corotation radius where the disk rotates at the same rate as the star. Spectro-interferometric studies in young stellar objects show that Hydrogen is mostly emitted in a region of a few milliarcseconds across, usually located within the dust sublimation radius. Its origin is still a matter of debate and it can be interpreted as coming from the stellar magnetosphere, a rotating wind or a disk. In the case of intermediate-mass Herbig AeBe stars, the fact that the Br gamma emission is spatially resolved rules out that most of the emission comes from the magnetosphere. This is due to the weak magnetic fields (some tenths of G) detected in these sources, resulting in very compact magnetospheres. In the case of T Tauri sources, their larger magnetospheres should make them easier to resolve. However, the small angular size of the magnetosphere (a few tenths of milliarcseconds), along with the presence of winds emitting in Hydrogen make the observations interpretation challenging. Here, we present direct evidence of magnetospheric accretion by spatially resolving the inner disk of the 60 pc T Tauri star TW Hydrae through optical long baseline interferometry. We find that the hydrogen near-infrared emission comes from a region approximately 3.5 stellar radii (R*) across. This region is within the continuum dusty disk emitting region (Rcont = 7 R*) and smaller than the corotation radius which is twice as big. This indicates that the hydrogen emission originates at the accretion columns, as expected in magnetospheric accretion models, rather than in a wind emitted at much larger distance (>1au)., Comment: Publish in Nature

The recent success of graph neural networks has significantly boosted molecular property prediction, advancing activities such as drug discovery. The existing deep neural network methods usually require large training dataset for each property, impairing their performances in cases (especially for new molecular properties) with a limited amount of experimental data, which are common in real situations. To this end, we propose Meta-MGNN, a novel model for few-shot molecular property prediction. Meta-MGNN applies molecular graph neural network to learn molecular representation and builds a meta-learning framework for model optimization. To exploit unlabeled molecular information and address task heterogeneity of different molecular properties, Meta-MGNN further incorporates molecular structure, attribute based self-supervised modules and self-attentive task weights into the former framework, strengthening the whole learning model. Extensive experiments on two public multi-property datasets demonstrate that Meta-MGNN outperforms a variety of state-of-the-art methods., Comment: To appear in WWW 2021 (long paper); Code is available at https://github.com/zhichunguo/Meta-MGNN

Predicting the final ischaemic stroke lesion provides crucial information regarding the volume of salvageable hypoperfused tissue, which helps physicians in the difficult decision-making process of treatment planning and intervention. Treatment selection is influenced by clinical diagnosis, which requires delineating the stroke lesion, as well as characterising cerebral blood flow dynamics using neuroimaging acquisitions. Nonetheless, predicting the final stroke lesion is an intricate task, due to the variability in lesion size, shape, location and the underlying cerebral haemodynamic processes that occur after the ischaemic stroke takes place. Moreover, since elapsed time between stroke and treatment is related to the loss of brain tissue, assessing and predicting the final stroke lesion needs to be performed in a short period of time, which makes the task even more complex. Therefore, there is a need for automatic methods that predict the final stroke lesion and support physicians in the treatment decision process. We propose a fully automatic deep learning method based on unsupervised and supervised learning to predict the final stroke lesion after 90 days. Our aim is to predict the final stroke lesion location and extent, taking into account the underlying cerebral blood flow dynamics that can influence the prediction. To achieve this, we propose a two-branch Restricted Boltzmann Machine, which provides specialized data-driven features from different sets of standard parametric Magnetic Resonance Imaging maps. These data-driven feature maps are then combined with the parametric Magnetic Resonance Imaging maps, and fed to a Convolutional and Recurrent Neural Network architecture. We evaluated our proposal on the publicly available ISLES 2017 testing dataset, reaching a Dice score of 0.38, Hausdorff Distance of 29.21 mm, and Average Symmetric Surface Distance of 5.52 mm., Comment: Accepted at Medical Image Analysis (MedIA)

The BraTS dataset contains a mixture of high-grade and low-grade gliomas, which have a rather different appearance: previous studies have shown that performance can be improved by separated training on low-grade gliomas (LGGs) and high-grade gliomas (HGGs), but in practice this information is not available at test time to decide which model to use. By contrast with HGGs, LGGs often present no sharp boundary between the tumor core and the surrounding edema, but rather a gradual reduction of tumor-cell density. Utilizing our 3D-to-2D fully convolutional architecture, DeepSCAN, which ranked highly in the 2019 BraTS challenge and was trained using an uncertainty-aware loss, we separate cases into those with a confidently segmented core, and those with a vaguely segmented or missing core. Since by assumption every tumor has a core, we reduce the threshold for classification of core tissue in those cases where the core, as segmented by the classifier, is vaguely defined or missing. We then predict survival of high-grade glioma patients using a fusion of linear regression and random forest classification, based on age, number of distinct tumor components, and number of distinct tumor cores. We present results on the validation dataset of the Multimodal Brain Tumor Segmentation Challenge 2020 (segmentation and uncertainty challenge), and on the testing set, where the method achieved 4th place in Segmentation, 1st place in uncertainty estimation, and 1st place in Survival prediction., Comment: Presented (virtually) in the MICCAI Brainles workshop 2020. Accepted for publication in Brainles proceedings

51 Oph is a Herbig Ae/Be star that exhibits strong near-infrared CO ro-vibrational emission at 2.3 micron, most likely originating in the innermost regions of a circumstellar disc. We aim to obtain the physical and geometrical properties of the system by spatially resolving the circumstellar environment of the inner gaseous disc. We used the second-generation VLTI/GRAVITY to spatially resolve the continuum and the CO overtone emission. We obtained data over 12 baselines with the auxiliary telescopes and derive visibilities, and the differential and closure phases as a function of wavelength. We used a simple LTE ring model of the CO emission to reproduce the spectrum and CO line displacements. Our interferometric data show that the star is marginally resolved at our spatial resolution, with a radius of 10.58+-2.65 Rsun.The K-band continuum emission from the disc is inclined by 63+-1 deg, with a position angle of 116+-1 deg, and 4+-0.8 mas (0.5+-0.1 au) across. The visibilities increase within the CO line emission, indicating that the CO is emitted within the dust-sublimation radius.By modelling the CO bandhead spectrum, we derive that the CO is emitted from a hot (T=1900-2800 K) and dense (NCO=(0.9-9)x10^21 cm^-2) gas. The analysis of the CO line displacement with respect to the continuum allows us to infer that the CO is emitted from a region 0.10+-0.02 au across, well within the dust-sublimation radius. The inclination and position angle of the CO line emitting region is consistent with that of the dusty disc. Our spatially resolved interferometric observations confirm the CO ro-vibrational emission within the dust-free region of the inner disc. Conventional disc models exclude the presence of CO in the dust-depleted regions of Herbig AeBe stars. Ad hoc models of the innermost disc regions, that can compute the properties of the dust-free inner disc, are therefore required., Comment: Accepted in A&A

The Silicon Photomultiplier (SiPM) is a mature photodetector concept that is applied in a variety of applications ranging from medical imaging to automotive LiDAR systems. Over the last few years, improvements of the sensor performance are gradually approaching to a saturation. In this work we present our new concept to overcome the intrinsic limitations of planar configurations of electrodes. Our non-planar technology is based on focusing and enhancing the electric fields by tip-like electrodes. The shape of the electric field and the lack of typical micro-cell edges, allows us to exclude cell separation boundaries and eliminate dead space around active cell areas. Our design provides a high-density micro-cell layout with a high geometric efficiency. It resolves the well-known trade-off between the detection efficiency and the dynamic range. The first "Tip Avalanche Photodiode" (TAPD) prototypes show a remarkable geometric efficiency above 80% for a micro-cell pitch of 15$\mathrm{\mu}$m. This directly translates into a photon detection efficiency (PDE) record peak value of 73% at 600nm with respect to the state-of-the-art SiPMs. Moreover, the PDE remains above a value of 45% up to a wavelength of 800nm with another record value of 22% at 905nm. The reduced micro-cell capacity allows for a fast recovery time below 4ns, which improves the operation at high photon rates. Overall, the TAPD is anticipated to be a very promising SiPM generation for various wide-spectral and high-dynamic-range applications in health science, biophysics, particle physics and LiDARs.

The search for a first-order phase transition in strongly interacting matter is one of the major objectives in the exploration of the phase diagram of Quantum Chromodynamics (QCD). In the present work we investigate dilepton radiation from the hot and dense fireballs created in Au-Au collisions at projectile energies of 1-2 $A$GeV for potential signatures of a first-order transition. Toward this end, we employ a hydrodynamic simulation with two different equations of state, with and without a phase transition. The latter is constrained by susceptibilities at vanishing chemical potential from lattice-QCD as well as neutron star properties, while the former is implemented via modification of the mean-fields in the quark phase. We find that the latent heat involved in the first-order transition leads to a substantial increase in the low-mass thermal emission signal, by about a factor of two above the cross-over scenario., Comment: 14 pages, 8 figures

This article in memory of Patrick Dehornoy (1952-2019) is an invitation to Garside theory for mainstream geometric group theorists interested in mapping class groups, curve complexes, and the geometry of Artin-Tits groups.

We use PIONIER data from the ESO archive and GRAVITY data that were obtained in June 2017 with the four 8m telescopes. We use a parametric disk model and the 3D radiative transfer code MCFOST to reproduce the Spectral Energy Distribution and match the interferometric observations. To match the SED , our model requires a stellar luminosity of 2.5 Lsun, higher than any previously determined values. Such a high value is needed to accommodate the circumstellar extinction caused by the highly inclined disk, which has been neglected in previous studies. While using an effective temperature of 4800 K determined through high-resolution spectroscopy, we derive a stellar radius of 2.29 Rsun. These revised fundamental parameters, when combined with the mass estimates available , lead to an age of 0.5-2.0 Ma for RY Lup, in better agreement with the age of the Lupus association than previous determinations. Our disk model nicely reproduces the interferometric GRAVITY data and is in good agreement with the PIONIER ones. We derive an inner rim location at 0.12~au from the central star. This model corresponds to an inclination of the inner disk of 50deg, which is in mild tension with previous determinations of a more inclined outer disk from SPHERE (70 deg in NIR) and ALMA(67 $\pm$5 deg) images, but consistent with the inclination determination from the ALMA CO spectra (55$\pm$5deg). Increasing the inclination of the inner disk to 70 deg leads to a higher line-of-sight extinction and therefore requires a higher stellar luminosity of 4.65 Lsun to match the observed flux levels. This luminosity would translate to a stellar radius of 3.13~Rsun, leading to an age of 2-3~Ma, and a stellar mass of about 2 Msun, in disagreement with the observed dynamical mass estimate of 1.3-1.5 Msun. Critically, this high-inclination inner disk model also fails to reproduce the visibilities observed with GRAVITY., Comment: Accepted for publication in A&A; 11 pages, 5 figures, 3 tables

The formation and the evolution of protoplanetary disks are important stages in the lifetime of stars. The processes of disk evolution and planet formation are intrinsically linked. We spatially resolve with GRAVITY/VLTI in the K-band the sub au-scale region of 27 stars to gain statistical understanding of their properties. We look for correlations with stellar parameters, such as luminosity, mass, temperature and age. Our sample also cover a range of various properties in terms of reprocessed flux, flared or flat morphology, and gaps. We developed semi-physical geometrical models to fit our interferometric data. Our best models correspond to smooth and wide rings, implying that wedge-shaped rims at the dust sublimation edge are favored, as found in the H-band. The closure phases are generally non-null with a median value of ~10 deg, indicating spatial asymmetries of the intensity distributions. Multi-size grain populations could explain the closure phase ranges below 20-25 deg but other scenarios should be invoked to explain the largest ones. Our measurements extend the Radius-Luminosity relation to ~1e4 Lsun and confirm the significant spread around the mean relation observed in the H-band. Gapped sources exhibit a large N-to-K band size ratio and large values of this ratio are only observed for the members of our sample that would be older than 1 Ma, less massive, and with lower luminosity. In the 2 Ms mass range, we observe a correlation in the increase of the relative age with the transition from group II to group I, and an increase of the N-to-K size ratio. However, the size of the current sample does not yet permit us to invoke a clear universal evolution mechanism across the HAeBe mass range. The measured locations of the K-band emission suggest that these disks might be structured by forming young planets, rather than by depletion due to EUV, FUV, and X-ray photo-evaporation., Comment: Accepted for publication in A&A; 23 pages, 16 figures, 7 tables

Deep learning methods for brain tumor segmentation are typically trained in an ad hoc fashion on all available data. Brain tumors are tremendously heterogeneous in image appearance and labeled training data is limited. We argue that incorporation of additional prior information, specifically tumor grade, associated with tumor imaging phenotypes during model training can significantly improve segmentation performance. Two strategies for incorporation of tumor grade during model training are proposed and their impact on segmentation performance is demonstrated on the BRATS 2018 dataset., Comment: Accepted as extended abstract for MIDL 2019 [arXiv:1907.08612]

The detection of new or enlarged white-matter lesions in multiple sclerosis is a vital task in the monitoring of patients undergoing disease-modifying treatment for multiple sclerosis. However, the definition of 'new or enlarged' is not fixed, and it is known that lesion-counting is highly subjective, with high degree of inter- and intra-rater variability. Automated methods for lesion quantification hold the potential to make the detection of new and enlarged lesions consistent and repeatable. However, the majority of lesion segmentation algorithms are not evaluated for their ability to separate progressive from stable patients, despite this being a pressing clinical use-case. In this paper we show that change in volumetric measurements of lesion load alone is not a good method for performing this separation, even for highly performing segmentation methods. Instead, we propose a method for identifying lesion changes of high certainty, and establish on a dataset of longitudinal multiple sclerosis cases that this method is able to separate progressive from stable timepoints with a very high level of discrimination (AUC = 0.99), while changes in lesion volume are much less able to perform this separation (AUC = 0.71). Validation of the method on a second external dataset confirms that the method is able to generalize beyond the setting in which it was trained, achieving an accuracy of 83% in separating stable and progressive timepoints. Both lesion volume and count have previously been shown to be strong predictors of disease course across a population. However, we demonstrate that for individual patients, changes in these measures are not an adequate means of establishing no evidence of disease activity. Meanwhile, directly detecting tissue which changes, with high confidence, from non-lesion to lesion is a feasible methodology for identifying radiologically active patients.

In applications of supervised learning applied to medical image segmentation, the need for large amounts of labeled data typically goes unquestioned. In particular, in the case of brain anatomy segmentation, hundreds or thousands of weakly-labeled volumes are often used as training data. In this paper, we first observe that for many brain structures, a small number of training examples, (n=9), weakly labeled using Freesurfer 6.0, plus simple data augmentation, suffice as training data to achieve high performance, achieving an overall mean Dice coefficient of $0.84 \pm 0.12$ compared to Freesurfer over 28 brain structures in T1-weighted images of $\approx 4000$ 9-10 year-olds from the Adolescent Brain Cognitive Development study. We then examine two varieties of heteroscedastic network as a method for improving classification results. An existing proposal by Kendall and Gal, which uses Monte-Carlo inference to learn to predict the variance of each prediction, yields an overall mean Dice of $0.85 \pm 0.14$ and showed statistically significant improvements over 25 brain structures. Meanwhile a novel heteroscedastic network which directly learns the probability that an example has been mislabeled yielded an overall mean Dice of $0.87 \pm 0.11$ and showed statistically significant improvements over all but one of the brain structures considered. The loss function associated to this network can be interpreted as performing a form of learned label smoothing, where labels are only smoothed where they are judged to be uncertain.

The goal of this mostly expository paper is to present several candidates for hyperbolic structures on irreducible Artin-Tits groups of spherical type and to elucidate some relations between them. Most constructions are algebraic analogues of previously known hyperbolic structures on Artin braid groups coming from natural actions of these groups on curve graphs and (modified) arc graphs of punctured disks.

Quantification of cerebral white matter hyperintensities (WMH) of presumed vascular origin is of key importance in many neurological research studies. Currently, measurements are often still obtained from manual segmentations on brain MR images, which is a laborious procedure. Automatic WMH segmentation methods exist, but a standardized comparison of the performance of such methods is lacking. We organized a scientific challenge, in which developers could evaluate their method on a standardized multi-center/-scanner image dataset, giving an objective comparison: the WMH Segmentation Challenge (https://wmh.isi.uu.nl/). Sixty T1+FLAIR images from three MR scanners were released with manual WMH segmentations for training. A test set of 110 images from five MR scanners was used for evaluation. Segmentation methods had to be containerized and submitted to the challenge organizers. Five evaluation metrics were used to rank the methods: (1) Dice similarity coefficient, (2) modified Hausdorff distance (95th percentile), (3) absolute log-transformed volume difference, (4) sensitivity for detecting individual lesions, and (5) F1-score for individual lesions. Additionally, methods were ranked on their inter-scanner robustness. Twenty participants submitted their method for evaluation. This paper provides a detailed analysis of the results. In brief, there is a cluster of four methods that rank significantly better than the other methods, with one clear winner. The inter-scanner robustness ranking shows that not all methods generalize to unseen scanners. The challenge remains open for future submissions and provides a public platform for method evaluation., Comment: Accepted for publication in IEEE Transactions on Medical Imaging

To date, infrared interferometry at best achieved contrast ratios of a few times $10^{-4}$ on bright targets. GRAVITY, with its dual-field mode, is now capable of high contrast observations, enabling the direct observation of exoplanets. We demonstrate the technique on HR8799, a young planetary system composed of four known giant exoplanets. We used the GRAVITY fringe tracker to lock the fringes on the central star, and integrated off-axis on the HR8799e planet situated at 390 mas from the star. Data reduction included post-processing to remove the flux leaking from the central star and to extract the coherent flux of the planet. The inferred K band spectrum of the planet has a spectral resolution of 500. We also derive the astrometric position of the planet relative to the star with a precision on the order of 100$\,\mu$as. The GRAVITY astrometric measurement disfavors perfectly coplanar stable orbital solutions. A small adjustment of a few degrees to the orbital inclination of HR 8799 e can resolve the tension, implying that the orbits are close to, but not strictly coplanar. The spectrum, with a signal-to-noise ratio of $\approx 5$ per spectral channel, is compatible with a late-type L brown dwarf. Using Exo-REM synthetic spectra, we derive a temperature of $1150\pm50$\,K and a surface gravity of $10^{4.3\pm0.3}\,$cm/s$^{2}$. This corresponds to a radius of $1.17^{+0.13}_{-0.11}\,R_{\rm Jup}$ and a mass of $10^{+7}_{-4}\,M_{\rm Jup}$, which is an independent confirmation of mass estimates from evolutionary models. Our results demonstrate the power of interferometry for the direct detection and spectroscopic study of exoplanets at close angular separations from their stars., Comment: published in A&A

Generative Adversarial Networks (GANs) have become exceedingly popular in a wide range of data-driven research fields, due in part to their success in image generation. Their ability to generate new samples, often from only a small amount of input data, makes them an exciting research tool in areas with limited data resources. One less-explored application of GANs is the synthesis of speech and audio samples. Herein, we propose a set of extensions to the WaveGAN paradigm, a recently proposed approach for sound generation using GANs. The aim of these extensions - preprocessing, Audio-to-Audio generation, skip connections and progressive structures - is to improve the human likeness of synthetic speech samples. Scores from listening tests with 30 volunteers demonstrated a moderate improvement (Cohen's d coefficient of 0.65) in human likeness using the proposed extensions compared to the original WaveGAN approach., Comment: 7 pages, 2 figures

Segmentation of white matter lesions and deep grey matter structures is an important task in the quantification of magnetic resonance imaging in multiple sclerosis. In this paper we explore segmentation solutions based on convolutional neural networks (CNNs) for providing fast, reliable segmentations of lesions and grey-matter structures in multi-modal MR imaging, and the performance of these methods when applied to out-of-centre data. We trained two state-of-the-art fully convolutional CNN architectures on the 2016 MSSEG training dataset, which was annotated by seven independent human raters: a reference implementation of a 3D Unet, and a more recently proposed 3D-to-2D architecture (DeepSCAN). We then retrained those methods on a larger dataset from a single centre, with and without labels for other brain structures. We quantified changes in performance owing to dataset shift, and changes in performance by adding the additional brain-structure labels. We also compared performance with freely available reference methods. Both fully-convolutional CNN methods substantially outperform other approaches in the literature when trained and evaluated in cross-validation on the MSSEG dataset, showing agreement with human raters in the range of human inter-rater variability. Both architectures showed drops in performance when trained on single-centre data and tested on the MSSEG dataset. When trained with the addition of weak anatomical labels derived from Freesurfer, the performance of the 3D Unet degraded, while the performance of the DeepSCAN net improved. Overall, the DeepSCAN network predicting both lesion and anatomical labels was the best-performing network examined., Comment: Substantially revised version after comments from reviewers, including comparison to 3D Unet

Deep learning for regression tasks on medical imaging data has shown promising results. However, compared to other approaches, their power is strongly linked to the dataset size. In this study, we evaluate 3D-convolutional neural networks (CNNs) and classical regression methods with hand-crafted features for survival time regression of patients with high grade brain tumors. The tested CNNs for regression showed promising but unstable results. The best performing deep learning approach reached an accuracy of 51.5% on held-out samples of the training set. All tested deep learning experiments were outperformed by a Support Vector Classifier (SVC) using 30 radiomic features. The investigated features included intensity, shape, location and deep features. The submitted method to the BraTS 2018 survival prediction challenge is an ensemble of SVCs, which reached a cross-validated accuracy of 72.2% on the BraTS 2018 training set, 57.1% on the validation set, and 42.9% on the testing set. The results suggest that more training data is necessary for a stable performance of a CNN model for direct regression from magnetic resonance images, and that non-imaging clinical patient information is crucial along with imaging information., Comment: Contribution to The International Multimodal Brain Tumor Segmentation (BraTS) Challenge 2018, survival prediction task

Gliomas are the most common primary brain malignancies, with different degrees of aggressiveness, variable prognosis and various heterogeneous histologic sub-regions, i.e., peritumoral edematous/invaded tissue, necrotic core, active and non-enhancing core. This intrinsic heterogeneity is also portrayed in their radio-phenotype, as their sub-regions are depicted by varying intensity profiles disseminated across multi-parametric magnetic resonance imaging (mpMRI) scans, reflecting varying biological properties. Their heterogeneous shape, extent, and location are some of the factors that make these tumors difficult to resect, and in some cases inoperable. The amount of resected tumor is a factor also considered in longitudinal scans, when evaluating the apparent tumor for potential diagnosis of progression. Furthermore, there is mounting evidence that accurate segmentation of the various tumor sub-regions can offer the basis for quantitative image analysis towards prediction of patient overall survival. This study assesses the state-of-the-art machine learning (ML) methods used for brain tumor image analysis in mpMRI scans, during the last seven instances of the International Brain Tumor Segmentation (BraTS) challenge, i.e., 2012-2018. Specifically, we focus on i) evaluating segmentations of the various glioma sub-regions in pre-operative mpMRI scans, ii) assessing potential tumor progression by virtue of longitudinal growth of tumor sub-regions, beyond use of the RECIST/RANO criteria, and iii) predicting the overall survival from pre-operative mpMRI scans of patients that underwent gross total resection. Finally, we investigate the challenge of identifying the best ML algorithms for each of these tasks, considering that apart from being diverse on each instance of the challenge, the multi-institutional mpMRI BraTS dataset has also been a continuously evolving/growing dataset., Comment: The International Multimodal Brain Tumor Segmentation (BraTS) Challenge

This work presents an interferometric study of the massive-binary fraction in the Orion Trapezium Cluster with the recently comissioned GRAVITY instrument. We observe a total of 16 stars of mainly OB spectral type. We find three previously unknown companions for $\theta ^1$ Ori B, $\theta ^2$ Ori B, and $\theta ^2$ Ori C. We determine a separation for the previously suspected companion of NU Ori. We confirm four companions for $\theta ^1$ Ori A, $\theta ^1$ Ori C, $\theta ^1$ Ori D, and $\theta ^2$ Ori A, all with substantially improved astrometry and photometric mass estimates. We refine the orbit of the eccentric high-mass binary $\theta ^1$ Ori C and we are able to derive a new orbit for $\theta ^1$ Ori D. We find a system mass of 21.7 $M_{\odot}$ and a period of $53$ days. Together with other previously detected companions seen in spectroscopy or direct imaging, eleven of the 16 high-mass stars are multiple systems. We obtain a total number of 22 companions with separations up to 600 AU. The companion fraction of the early B and O stars in our sample is about 2, significantly higher than in earlier studies of mostly OB associations. The separation distribution hints towards a bimodality. Such a bimodality has been previously found in A stars, but rarely in OB binaries, which up to this point have been assumed to be mostly compact with a tail of wider companions. We also do not find a substantial population of equal-mass binaries. The observed distribution of mass ratios declines steeply with mass, and like the direct star counts, indicates that our companions follow a standard power law initial mass function. Again, this is in contrast to earlier findings of flat mass ratio distributions in OB associations. We exclude collision as a dominant formation mechanism but find no clear preference for core accretion or competitive accretion., Comment: Accepted for publication in A&A