Search

Generative modelling and synthetic data can be a surrogate for real medical imaging datasets, whose scarcity and difficulty to share can be a nuisance when delivering accurate deep learning models for healthcare applications. In recent years, there has been an increased interest in using these models for data augmentation and synthetic data sharing, using architectures such as generative adversarial networks (GANs) or diffusion models (DMs). Nonetheless, the application of synthetic data to tasks such as 3D magnetic resonance imaging (MRI) segmentation remains limited due to the lack of labels associated with the generated images. Moreover, many of the proposed generative MRI models lack the ability to generate arbitrary modalities due to the absence of explicit contrast conditioning. These limitations prevent the user from adjusting the contrast and content of the images and obtaining more generalisable data for training task-specific models. In this work, we propose brainSPADE3D, a 3D generative model for brain MRI and associated segmentations, where the user can condition on specific pathological phenotypes and contrasts. The proposed joint imaging-segmentation generative model is shown to generate high-fidelity synthetic images and associated segmentations, with the ability to combine pathologies. We demonstrate how the model can alleviate issues with segmentation model performance when unexpected pathologies are present in the data., Comment: Accepted for publication at the 2023 Deep Generative Models (DGM4MICCAI) MICCAI workshop (Vancouver, Canada)

Recent advances in generative AI have brought incredible breakthroughs in several areas, including medical imaging. These generative models have tremendous potential not only to help safely share medical data via synthetic datasets but also to perform an array of diverse applications, such as anomaly detection, image-to-image translation, denoising, and MRI reconstruction. However, due to the complexity of these models, their implementation and reproducibility can be difficult. This complexity can hinder progress, act as a use barrier, and dissuade the comparison of new methods with existing works. In this study, we present MONAI Generative Models, a freely available open-source platform that allows researchers and developers to easily train, evaluate, and deploy generative models and related applications. Our platform reproduces state-of-art studies in a standardised way involving different architectures (such as diffusion models, autoregressive transformers, and GANs), and provides pre-trained models for the community. We have implemented these models in a generalisable fashion, illustrating that their results can be extended to 2D or 3D scenarios, including medical images with different modalities (like CT, MRI, and X-Ray data) and from different anatomical areas. Finally, we adopt a modular and extensible approach, ensuring long-term maintainability and the extension of current applications for future features.

Methods for out-of-distribution (OOD) detection that scale to 3D data are crucial components of any real-world clinical deep learning system. Classic denoising diffusion probabilistic models (DDPMs) have been recently proposed as a robust way to perform reconstruction-based OOD detection on 2D datasets, but do not trivially scale to 3D data. In this work, we propose to use Latent Diffusion Models (LDMs), which enable the scaling of DDPMs to high-resolution 3D medical data. We validate the proposed approach on near- and far-OOD datasets and compare it to a recently proposed, 3D-enabled approach using Latent Transformer Models (LTMs). Not only does the proposed LDM-based approach achieve statistically significant better performance, it also shows less sensitivity to the underlying latent representation, more favourable memory scaling, and produces better spatial anomaly maps. Code is available at https://github.com/marksgraham/ddpm-ood, Comment: Accepted at MICCAI 2023

Many web sites are transitioning how they construct their pages. The conventional model is where the content is embedded server-side in the HTML and returned to the client in an HTTP response. Increasingly, sites are moving to a model where the initial HTTP response contains only an HTML skeleton plus JavaScript that makes API calls to a variety of servers for the content (typically in JSON format), and then builds out the DOM client-side, more easily allowing for periodically refreshing the content in a page and allowing dynamic modification of the content. This client-side rendering, now predominant in social media platforms such as Twitter and Instagram, is also being adopted by news outlets, such as CNN.com. When conventional web archiving techniques, such as crawling with Heritrix, are applied to pages that render their content client-side, the JSON responses can become out of sync with the HTML page in which it is to be embedded, resulting in temporal violations on replay. Because the violative JSON is not directly observable in the page (i.e., in the same manner a violative embedded image is), the temporal violations can be difficult to detect. We describe how the top level CNN.com page has used client-side rendering since April 2015 and the impact this has had on web archives. Between April 24, 2015 and July 21, 2016, we found almost 15,000 mementos with a temporal violation of more than 2 days between the base CNN.com HTML and the JSON responses used to deliver the content under the main story. One way to mitigate this problem is to use browser-based crawling instead of conventional crawlers like Heritrix, but browser-based crawling is currently much slower than non-browser-based tools such as Heritrix., Comment: 20 pages, preprint version of paper accepted at the 2023 ACM/IEEE Joint Conference on Digital Libraries (JCDL)

Out-of-distribution detection is crucial to the safe deployment of machine learning systems. Currently, unsupervised out-of-distribution detection is dominated by generative-based approaches that make use of estimates of the likelihood or other measurements from a generative model. Reconstruction-based methods offer an alternative approach, in which a measure of reconstruction error is used to determine if a sample is out-of-distribution. However, reconstruction-based approaches are less favoured, as they require careful tuning of the model's information bottleneck - such as the size of the latent dimension - to produce good results. In this work, we exploit the view of denoising diffusion probabilistic models (DDPM) as denoising autoencoders where the bottleneck is controlled externally, by means of the amount of noise applied. We propose to use DDPMs to reconstruct an input that has been noised to a range of noise levels, and use the resulting multi-dimensional reconstruction error to classify out-of-distribution inputs. We validate our approach both on standard computer-vision datasets and on higher dimension medical datasets. Our approach outperforms not only reconstruction-based methods, but also state-of-the-art generative-based approaches. Code is available at https://github.com/marksgraham/ddpm-ood.

In order to achieve good performance and generalisability, medical image segmentation models should be trained on sizeable datasets with sufficient variability. Due to ethics and governance restrictions, and the costs associated with labelling data, scientific development is often stifled, with models trained and tested on limited data. Data augmentation is often used to artificially increase the variability in the data distribution and improve model generalisability. Recent works have explored deep generative models for image synthesis, as such an approach would enable the generation of an effectively infinite amount of varied data, addressing the generalisability and data access problems. However, many proposed solutions limit the user's control over what is generated. In this work, we propose brainSPADE, a model which combines a synthetic diffusion-based label generator with a semantic image generator. Our model can produce fully synthetic brain labels on-demand, with or without pathology of interest, and then generate a corresponding MRI image of an arbitrary guided style. Experiments show that brainSPADE synthetic data can be used to train segmentation models with performance comparable to that of models trained on real data., Comment: 12 pages, 2 (+2 App.) figures, 3 tables. Accepted at Simulation and Synthesis in Medical Imaging workshop (MICCAI 2022)

Human anatomy, morphology, and associated diseases can be studied using medical imaging data. However, access to medical imaging data is restricted by governance and privacy concerns, data ownership, and the cost of acquisition, thus limiting our ability to understand the human body. A possible solution to this issue is the creation of a model able to learn and then generate synthetic images of the human body conditioned on specific characteristics of relevance (e.g., age, sex, and disease status). Deep generative models, in the form of neural networks, have been recently used to create synthetic 2D images of natural scenes. Still, the ability to produce high-resolution 3D volumetric imaging data with correct anatomical morphology has been hampered by data scarcity and algorithmic and computational limitations. This work proposes a generative model that can be scaled to produce anatomically correct, high-resolution, and realistic images of the human brain, with the necessary quality to allow further downstream analyses. The ability to generate a potentially unlimited amount of data not only enables large-scale studies of human anatomy and pathology without jeopardizing patient privacy, but also significantly advances research in the field of anomaly detection, modality synthesis, learning under limited data, and fair and ethical AI. Code and trained models are available at: https://github.com/AmigoLab/SynthAnatomy., Comment: 13 pages, 3 figures, 2 tables, accepted at SASHIMI MICCAI 2022

Deep generative models have emerged as promising tools for detecting arbitrary anomalies in data, dispensing with the necessity for manual labelling. Recently, autoregressive transformers have achieved state-of-the-art performance for anomaly detection in medical imaging. Nonetheless, these models still have some intrinsic weaknesses, such as requiring images to be modelled as 1D sequences, the accumulation of errors during the sampling process, and the significant inference times associated with transformers. Denoising diffusion probabilistic models are a class of non-autoregressive generative models recently shown to produce excellent samples in computer vision (surpassing Generative Adversarial Networks), and to achieve log-likelihoods that are competitive with transformers while having fast inference times. Diffusion models can be applied to the latent representations learnt by autoencoders, making them easily scalable and great candidates for application to high dimensional data, such as medical images. Here, we propose a method based on diffusion models to detect and segment anomalies in brain imaging. By training the models on healthy data and then exploring its diffusion and reverse steps across its Markov chain, we can identify anomalous areas in the latent space and hence identify anomalies in the pixel space. Our diffusion models achieve competitive performance compared with autoregressive approaches across a series of experiments with 2D CT and MRI data involving synthetic and real pathological lesions with much reduced inference times, making their usage clinically viable.

In a clinical setting it is essential that deployed image processing systems are robust to the full range of inputs they might encounter and, in particular, do not make confidently wrong predictions. The most popular approach to safe processing is to train networks that can provide a measure of their uncertainty, but these tend to fail for inputs that are far outside the training data distribution. Recently, generative modelling approaches have been proposed as an alternative; these can quantify the likelihood of a data sample explicitly, filtering out any out-of-distribution (OOD) samples before further processing is performed. In this work, we focus on image segmentation and evaluate several approaches to network uncertainty in the far-OOD and near-OOD cases for the task of segmenting haemorrhages in head CTs. We find all of these approaches are unsuitable for safe segmentation as they provide confidently wrong predictions when operating OOD. We propose performing full 3D OOD detection using a VQ-GAN to provide a compressed latent representation of the image and a transformer to estimate the data likelihood. Our approach successfully identifies images in both the far- and near-OOD cases. We find a strong relationship between image likelihood and the quality of a model's segmentation, making this approach viable for filtering images unsuitable for segmentation. To our knowledge, this is the first time transformers have been applied to perform OOD detection on 3D image data. Code is available at github.com/marksgraham/transformer-ood., Comment: Accepted at MIDL 2022 (Oral)

The Covid-19 pandemic has led to the rise of remote work with consequences for the global division of work. Remote work could connect labour markets, but it could also increase spatial polarisation. However, our understanding of the geographies of remote work is limited. Specifically, does remote work bring jobs to rural areas or is it concentrating in large cities, and how do skill requirements affect competition for jobs and wages? We use data from a fully remote labour market - an online labour platform - to show that remote work is polarised along three dimensions. First, countries are globally divided: North American, European, and South Asian remote workers attract most jobs, while many Global South countries participate only marginally. Secondly, remote jobs are pulled to urban regions; rural areas fall behind. Thirdly, remote work is polarised along the skill axis: workers with in-demand skills attract profitable jobs, while others face intense competition and obtain low wages. The findings suggest that remote work is shaped by agglomerative forces, which are deepening the gap between urban and rural areas. To make remote work an effective tool for rural development, it needs to be embedded in local skill-building and labour market programmes., Comment: This is an Accepted Manuscript of an article published in the journal PLOS ONE

Consider n nodes communicating over an unreliable broadcast channel. Each node has a single packet that needs to be communicated to all other nodes. Time is slotted, and a time slot is long enough for each node to broadcast one packet. Each broadcast reaches a random subset of nodes. The objective is to minimise the time until all nodes have received all packets. We study two schemes, (i) random relaying, and (ii) random linear network coding, and analyse their performance in an asymptotic regime in which n tends to infinity. Simulation results for a wide range of n are also presented., Comment: Submitted to IEEE Transactions on Information Theory

The Covid Symptom Study, a smartphone-based surveillance study on COVID-19 symptoms in the population, is an exemplar of big data citizen science. Over 4.7 million participants and 189 million unique assessments have been logged since its introduction in March 2020. The success of the Covid Symptom Study creates technical challenges around effective data curation for two reasons. Firstly, the scale of the dataset means that it can no longer be easily processed using standard software on commodity hardware. Secondly, the size of the research group means that replicability and consistency of key analytics used across multiple publications becomes an issue. We present ExeTera, an open source data curation software designed to address scalability challenges and to enable reproducible research across an international research group for datasets such as the Covid Symptom Study dataset.

Convolutional neural networks trained on publicly available medical imaging datasets (source domain) rarely generalise to different scanners or acquisition protocols (target domain). This motivates the active field of domain adaptation. While some approaches to the problem require labeled data from the target domain, others adopt an unsupervised approach to domain adaptation (UDA). Evaluating UDA methods consists of measuring the model's ability to generalise to unseen data in the target domain. In this work, we argue that this is not as useful as adapting to the test set directly. We therefore propose an evaluation framework where we perform test-time UDA on each subject separately. We show that models adapted to a specific target subject from the target domain outperform a domain adaptation method which has seen more data of the target domain but not this specific target subject. This result supports the thesis that unsupervised domain adaptation should be used at test-time, even if only using a single target-domain subject, Comment: Accepted at MICCAI 2020

Many atlases used for brain parcellation are hierarchically organised, progressively dividing the brain into smaller sub-regions. However, state-of-the-art parcellation methods tend to ignore this structure and treat labels as if they are `flat'. We introduce a hierarchically-aware brain parcellation method that works by predicting the decisions at each branch in the label tree. We further show how this method can be used to model uncertainty separately for every branch in this label tree. Our method exceeds the performance of flat uncertainty methods, whilst also providing decomposed uncertainty estimates that enable us to obtain self-consistent parcellations and uncertainty maps at any level of the label hierarchy. We demonstrate a simple way these decision-specific uncertainty maps may be used to provided uncertainty-thresholded tissue maps at any level of the label tree., Comment: To be published in the MICCAI 2020 workshop: Uncertainty for Safe Utilization of Machine Learning in Medical Imaging

We study the link between the kinematic-morphology of galaxies, as inferred from integral-field stellar kinematics, and their relation between mass and star formation rate (SFR). Our sample consists of $\sim 3200$ galaxies with integral-field spectroscopic data from the MaNGA survey with available determinations of their effective stellar angular momentum within the half-light radius $\lambda_{R_e}$. We find that for star-forming galaxies, namely along the star formation main sequence (SFMS), the $\lambda_{R_e}$ values remain large and almost unchanged over about two orders of magnitude in stellar mass, with the exception of the lowest masses $\mathcal{M}_{\star}\lesssim2\times10^{9} \mathcal{M}_{\odot}$, where $\lambda_{R_e}$ slightly decreases. The SFMS is dominated by spiral galaxies with small bulges. Below the SFMS, but above the characteristic stellar mass $\mathcal{M}_{\rm crit}\approx2\times10^{11} \mathcal{M}_{\odot}$, there is a sharp decrease in $\lambda_{R_e}$ with decreasing star formation rate: massive galaxies well below the SFMS are mainly slow-rotator early-type galaxies, namely genuinely spheroidal galaxies without disks. Below the SFMS and below $\mathcal{M}_{\rm crit}$ the decrease of $\lambda_{R_e}$ with decreasing SFR becomes modest or nearly absent: low-mass galaxies well below the SFMS, are fast-rotator early-type galaxies, and contain fast-rotating stellar disks like their star-forming counterparts. We also find a small but clear environmental dependence for the massive galaxies: in the mass range $10^{10.9}-10^{11.5} \mathcal{M}_{\odot}$, galaxies in rich groups or denser regions or classified as central galaxies have lower values of $\lambda_{R_e}$. While no environmental dependence is found for galaxies of lower mass. We discuss how our results can be understood as due to the different star formation and mass assembly histories of galaxies with varying mass., Comment: 22 pages, 12 figures, LaTeX. Accepted for publication on MNRAS

The increasing efficiency and compactness of deep learning architectures, together with hardware improvements, have enabled the complex and high-dimensional modelling of medical volumetric data at higher resolutions. Recently, Vector-Quantised Variational Autoencoders (VQ-VAE) have been proposed as an efficient generative unsupervised learning approach that can encode images to a small percentage of their initial size, while preserving their decoded fidelity. Here, we show a VQ-VAE inspired network can efficiently encode a full-resolution 3D brain volume, compressing the data to $0.825\%$ of the original size while maintaining image fidelity, and significantly outperforming the previous state-of-the-art. We then demonstrate that VQ-VAE decoded images preserve the morphological characteristics of the original data through voxel-based morphology and segmentation experiments. Lastly, we show that such models can be pre-trained and then fine-tuned on different datasets without the introduction of bias.