arXiv subjects

Daniel C. Alexander

Publications and source records attributed to Daniel C. Alexander.

107 records · Page 6Linked to original sources

arXiv2019

Multi-Stage Prediction Networks for Data Harmonization

In this paper, we introduce multi-task learning (MTL) to data harmonization (DH); where we aim to harmonize images across different acquisition platforms and sites. This allows us to integrate information from multiple acquisitions and improve the predictive performance and learning efficiency of the harmonization model. Specifically, we introduce the Multi Stage Prediction (MSP) Network, a MTL framework that incorporates neural networks of potentially disparate architectures, trained for different individual acquisition platforms, into a larger architecture that is refined in unison. The MSP utilizes high-level features of single networks for individual tasks, as inputs of additional neural networks to inform the final prediction, therefore exploiting redundancy across tasks to make the most of limited training data. We validate our methods on a dMRI harmonization challenge dataset, where we predict three modern platform types, from one obtained from an old scanner. We show how MTL architectures, such as the MSP, produce around 20\% improvement of patch-based mean-squared error over current state-of-the-art methods and that our MSP outperforms off-the-shelf MTL networks. Our code is available https://github.com/sbb-gh/ .

cs.LG↗

arXiv2019

Machine learning based white matter models with permeability: An experimental study in cuprizone treated in-vivo mouse model of axonal demyelination

The intra-axonal water exchange time τi, a parameter associated with axonal permeability, could be an important biomarker for understanding demyelinating pathologies such as Multiple Sclerosis. Diffusion-Weighted MRI is sensitive to changes in permeability, however, the parameter has remained elusive due to the intractability of the mathematical models that incorporate it. Machine learning based computational models can potentially be used to estimate such parameters, and recently, a theoretical framework using a random forest (RF) suggests this is a promising approach. In this study, we adopt such an RF approach and experimentally investigate its suitability as a biomarker for demyelinating pathologies through direct comparison with histology. For this, we use an in-vivo cuprizone (CPZ) mouse model of demyelination with available ex-vivo electron microscopy (EM) data. We test our model on noise-free simulations and find very strong correlations between the predicted and ground truth parameters. For realistic noise levels as in our in-vivo data, the performance is affected, however, the parameters are still well estimated. We apply our RF model on in-vivo data from 8 CPZ and 8 wild-type (WT) mice and validate the RF estimates using histology. We find a strong correlation between the in-vivo RF estimates of τi and the EM measurements of myelin thickness ({ρ_τ}i = 0.82), and between RF estimates and EM measurements of intra-axonal volume fraction ({ρ_f} = 0.98). When comparing τi in CPZ and WT mice we find a statistically significant decrease in the corpus callosum of the CPZ compared to the WT mice, in line with our expectations that τi is lower in regions where the myelin sheath is damaged. Overall, these results demonstrate the suitability of machine learning compartment models with permeability as a potential biomarker for demyelinating pathologies.

physics.med-ph↗

arXiv2019

Learning From Noisy Labels By Regularized Estimation Of Annotator Confusion

The predictive performance of supervised learning algorithms depends on the quality of labels. In a typical label collection process, multiple annotators provide subjective noisy estimates of the "truth" under the influence of their varying skill-levels and biases. Blindly treating these noisy labels as the ground truth limits the accuracy of learning algorithms in the presence of strong disagreement. This problem is critical for applications in domains such as medical imaging where both the annotation cost and inter-observer variability are high. In this work, we present a method for simultaneously learning the individual annotator model and the underlying true label distribution, using only noisy observations. Each annotator is modeled by a confusion matrix that is jointly estimated along with the classifier predictions. We propose to add a regularization term to the loss function that encourages convergence to the true annotator confusion matrix. We provide a theoretical argument as to how the regularization is essential to our approach both for the case of single annotator and multiple annotators. Despite the simplicity of the idea, experiments on image classification tasks with both simulated and real labels show that our method either outperforms or performs on par with the state-of-the-art methods and is capable of estimating the skills of annotators even with a single label available per image.

cs.LG↗

arXiv2019

Adaptive Neural Trees

Deep neural networks and decision trees operate on largely separate paradigms; typically, the former performs representation learning with pre-specified architectures, while the latter is characterised by learning hierarchies over pre-specified features with data-driven architectures. We unite the two via adaptive neural trees (ANTs) that incorporates representation learning into edges, routing functions and leaf nodes of a decision tree, along with a backpropagation-based training algorithm that adaptively grows the architecture from primitive modules (e.g., convolutional layers). We demonstrate that, whilst achieving competitive performance on classification and regression datasets, ANTs benefit from (i) lightweight inference via conditional computation, (ii) hierarchical separation of features useful to the task e.g. learning meaningful class associations, such as separating natural vs. man-made objects, and (iii) a mechanism to adapt the architecture to the size and complexity of the training dataset.

cs.NE↗

arXiv2019

In utero diffusion MRI: challenges, advances, and applications

In utero diffusion MRI provides unique opportunities to non-invasively study the microstructure of tissue during fetal development. A wide range of developmental processes, such as the growth of white matter tracts in the brain, the maturation of placental villous trees, or the fibres in the fetal heart remain to be studied and understood in detail. Advances in fetal interventions and surgery furthermore increase the need for ever more precise antenatal diagnosis from fetal MRI. However, the specific properties of the in utero environment, such as fetal and maternal motion, increased field-of-view, tissue interfaces and safety considerations, are significant challenges for most MRI techniques, and particularly for diffusion. Recent years have seen major improvements, driven by the development of bespoke techniques adapted to these specific challenges in both acquisition and processing. Fetal diffusion MRI, an emerging research tool, is now adding valuable novel information for both research and clinical questions. This paper will highlight specific challenges, outline strategies to target them, and discuss two main applications: fetal brain connectomics and placental maturation.

physics.med-ph↗

arXiv2019

Combined Diffusion-Relaxometry MRI to Identify Dysfunction in the Human Placenta

Purpose: A combined diffusion-relaxometry MR acquisition and analysis pipeline for in-vivo human placenta, which allows for exploration of coupling between T2* and apparent diffusion coefficient (ADC) measurements in a sub 10 minute scan time. Methods: We present a novel acquisition combining a diffusion prepared spin-echo with subsequent gradient echoes. The placentas of 17 pregnant women were scanned in-vivo, including both healthy controls and participants with various pregnancy complications. We estimate the joint T2*-ADC spectra using an inverse Laplace transform. Results: T2*-ADC spectra demonstrate clear quantitative separation between normal and dysfunctional placentas. Conclusions: Combined T2*-diffusivity MRI is promising for assessing fetal and maternal health during pregnancy. The T2*-ADC spectrum potentially provides additional information on tissue microstructure, compared to measuring these two contrasts separately. The presented method is immediately applicable to the study of other organs.

physics.med-ph↗

arXiv2019

Accurate estimation of microscopic diffusion anisotropy and its time dependence in the mouse brain

Microscopic diffusion anisotropy (μA) has been recently gaining increasing attention for its ability to decouple the average compartment anisotropy from orientation dispersion. Advanced diffusion MRI sequences, such as double diffusion encoding (DDE) and double oscillating diffusion encoding (DODE) have been used for mapping μA. However, the time-dependence of μA has not been investigated insofar, and furthermore, the accuracy of μA estimation vis-à-vis different b-values was not assessed. Here, we investigate both these concepts using theory, simulation, and experiments in the mouse brain. In the first part, simulations and experimental results show that the conventional estimation of microscopic anisotropy from the difference of D(O)DE sequences with parallel and orthogonal gradient directions yields values that highly depend on the choice of b-value. To mitigate this undesirable bias, we propose a multi-shell approach that harnesses a polynomial fit of the signal difference up to third order terms in b-value. In simulations, this approach yields more accurate μA metrics, which are similar to the ground truth values. The second part of this work uses the proposed multi-shell method to estimate the time/frequency dependence of μA. The data shows either an increase or no change in μA with frequency depending on the region of interest, both in white and gray matter. When comparing the experimental results with simulations, it emerges that simple geometric models such as infinite cylinders with either negligible or finite radii cannot replicate the measured trend, and more complex models, which, for example, incorporate structure along the fibre direction are required. Thus, measuring the time dependence of microscopic anisotropy can provide valuable information for characterizing tissue microstructure.

physics.med-ph↗

arXiv2019

DIVE: A spatiotemporal progression model of brain pathology in neurodegenerative disorders

Here we present DIVE: Data-driven Inference of Vertexwise Evolution. DIVE is an image-based disease progression model with single-vertex resolution, designed to reconstruct long-term patterns of brain pathology from short-term longitudinal data sets. DIVE clusters vertex-wise biomarker measurements on the cortical surface that have similar temporal dynamics across a patient population, and concurrently estimates an average trajectory of vertex measurements in each cluster. DIVE uniquely outputs a parcellation of the cortex into areas with common progression patterns, leading to a new signature for individual diseases. DIVE further estimates the disease stage and progression speed for every visit of every subject, potentially enhancing stratification for clinical trials or management. On simulated data, DIVE can recover ground truth clusters and their underlying trajectory, provided the average trajectories are sufficiently different between clusters. We demonstrate DIVE on data from two cohorts: the Alzheimer's Disease Neuroimaging Initiative (ADNI) and the Dementia Research Centre (DRC), UK, containing patients with Posterior Cortical Atrophy (PCA) as well as typical Alzheimer's disease (tAD). DIVE finds similar spatial patterns of atrophy for tAD subjects in the two independent datasets (ADNI and DRC), and further reveals distinct patterns of pathology in different diseases (tAD vs PCA) and for distinct types of biomarker data: cortical thickness from Magnetic Resonance Imaging (MRI) vs amyloid load from Positron Emission Tomography (PET). Finally, DIVE can be used to estimate a fine-grained spatial distribution of pathology in the brain using any kind of voxelwise or vertexwise measures including Jacobian compression maps, fractional anisotropy (FA) maps from diffusion imaging or other PET measures. DIVE source code is available online: https://github.com/mrazvan22/dive

cs.CV↗

arXiv2018

TADPOLE Challenge: Prediction of Longitudinal Evolution in Alzheimer's Disease

The Alzheimer's Disease Prediction Of Longitudinal Evolution (TADPOLE) Challenge compares the performance of algorithms at predicting future evolution of individuals at risk of Alzheimer's disease. TADPOLE Challenge participants train their models and algorithms on historical data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) study or any other datasets to which they have access. Participants are then required to make monthly forecasts over a period of 5 years from January 2018, of three key outcomes for ADNI-3 rollover participants: clinical diagnosis, Alzheimer's Disease Assessment Scale Cognitive Subdomain (ADAS-Cog13), and total volume of the ventricles. These individual forecasts are later compared with the corresponding future measurements in ADNI-3 (obtained after the TADPOLE submission deadline). The first submission phase of TADPOLE was open for prize-eligible submissions between 15 June and 15 November 2017. The submission system remains open via the website: https://tadpole.grand-challenge.org, although since 15 November 2017 submissions are not eligible for the first round of prizes. This paper describes the design of the TADPOLE Challenge.

q-bio.PE↗

arXiv2018

Deeper Image Quality Transfer: Training Low-Memory Neural Networks for 3D Images

In this paper we address the memory demands that come with the processing of 3-dimensional, high-resolution, multi-channeled medical images in deep learning. We exploit memory-efficient backpropagation techniques, to reduce the memory complexity of network training from being linear in the network's depth, to being roughly constant $ - $ permitting us to elongate deep architectures with negligible memory increase. We evaluate our methodology in the paradigm of Image Quality Transfer, whilst noting its potential application to various tasks that use deep learning. We study the impact of depth on accuracy and show that deeper models have more predictive power, which may exploit larger training sets. We obtain substantially better results than the previous state-of-the-art model with a slight memory increase, reducing the root-mean-squared-error by $ 13\% $. Our code is publicly available.

cs.CV↗

arXiv2018

A generative model of realistic brain cells with application to numerical simulation of diffusion-weighted MR signal

In this work, we introduce a novel computational framework that we developed to use numerical simulations to investigate the complexity of brain tissue at a microscopic level with a detail never realised before. Directly inspired by the advances in computational neuroscience for modelling brain cells, we propose a generative model that enables us to simulate molecular diffusion within realistic digitalised brain cells, such as neurons and glia, in a completely controlled and flexible fashion. We validate our new approach by showing an excellent match between the morphology and simulated DW-MR signal of the generated digital model of brain cells and those of digital reconstruction of real brain cells from available open-access databases. We demonstrate the versatility and potentiality of the framework by showing a select set of examples of relevance for the DW-MR community. Further development is ongoing, which will support even more realistic conditions like dense packing of numerous 3D complex cell structures and varying cell surface permeability.

physics.med-ph↗

arXiv2018

Uncertainty in multitask learning: joint representations for probabilistic MR-only radiotherapy planning

Multi-task neural network architectures provide a mechanism that jointly integrates information from distinct sources. It is ideal in the context of MR-only radiotherapy planning as it can jointly regress a synthetic CT (synCT) scan and segment organs-at-risk (OAR) from MRI. We propose a probabilistic multi-task network that estimates: 1) intrinsic uncertainty through a heteroscedastic noise model for spatially-adaptive task loss weighting and 2) parameter uncertainty through approximate Bayesian inference. This allows sampling of multiple segmentations and synCTs that share their network representation. We test our model on prostate cancer scans and show that it produces more accurate and consistent synCTs with a better estimation in the variance of the errors, state of the art results in OAR segmentation and a methodology for quality assurance in radiotherapy treatment planning.

cs.CV↗

arXiv2017

ESCAPE - Echo SCraper and ClAssifier of PErsons: A novel tool to facilitate using voice-controlled devices for research

Smart devices have become common place in many homes, and these devices can be utilized to provide support for people with mental or physical deficits. Voice-controlled assistants are a class of smart device that collect a large amount of data in the home. In this work we present Echo SCraper and ClAssifier of Persons (ESCAPE), an open source software for the extraction of Amazon Echo interaction data, and speaker recognition on that data. We show that ESCAPE is able to extract data from a voice-controlled assistant and classify with accuracy who is talking, based on a small number of labeled audio data. Using ESCAPE to extract interactions recorded over 3 months in the first author's home yields a rich dataset of transcribed audio recordings. Our results demonstrate that using this software the Amazon Echo can be used to study participants in a naturalistic setting with minimal intrusion. We also discuss the potential for usage of voice-controlled devices together with ESCAPE to understand how diseases affect individuals, and how these data can be used to monitor disease processes in general.

cs.HC↗

arXiv2017

Bayesian Image Quality Transfer with CNNs: Exploring Uncertainty in dMRI Super-Resolution

In this work, we investigate the value of uncertainty modeling in 3D super-resolution with convolutional neural networks (CNNs). Deep learning has shown success in a plethora of medical image transformation problems, such as super-resolution (SR) and image synthesis. However, the highly ill-posed nature of such problems results in inevitable ambiguity in the learning of networks. We propose to account for intrinsic uncertainty through a per-patch heteroscedastic noise model and for parameter uncertainty through approximate Bayesian inference in the form of variational dropout. We show that the combined benefits of both lead to the state-of-the-art performance SR of diffusion MR brain images in terms of errors compared to ground truth. We further show that the reduced error scores produce tangible benefits in downstream tractography. In addition, the probabilistic nature of the methods naturally confers a mechanism to quantify uncertainty over the super-resolved output. We demonstrate through experiments on both healthy and pathological brains the potential utility of such an uncertainty measure in the risk assessment of the super-resolved images for subsequent clinical use.

cs.CV↗

arXiv2017

Disease Progression Modeling and Prediction through Random Effect Gaussian Processes and Time Transformation

The development of statistical approaches for the joint modelling of the temporal changes of imaging, biochemical, and clinical biomarkers is of paramount importance for improving the understanding of neurodegenerative disorders, and for providing a reference for the prediction and quantification of the pathology in unseen individuals. Nonetheless, the use of disease progression models for probabilistic predictions still requires investigation, for example for accounting for missing observations in clinical data, and for accurate uncertainty quantification. We tackle this problem by proposing a novel Gaussian process-based method for the joint modeling of imaging and clinical biomarker progressions from time series of individual observations. The model is formulated to account for individual random effects and time reparameterization, allowing non-parametric estimates of the biomarker evolution, as well as high flexibility in specifying correlation structure, and time transformation models. Thanks to the Bayesian formulation, the model naturally accounts for missing data, and allows for uncertainty quantification in the estimate of evolutions, as well as for probabilistic prediction of disease staging in unseen patients. The experimental results show that the proposed model provides a biologically plausible description of the evolution of Alzheimer's pathology across the whole disease time-span as well as remarkable predictive performance when tested on a large clinical cohort with missing observations.

stat.AP↗

arXiv2016

Diffusion MRI microstructure models with in vivo human brain Connectom data: results from a multi-group comparison

A large number of mathematical models have been proposed to describe the measured signal in diffusion-weighted (DW) magnetic resonance imaging (MRI) and infer properties about the white matter microstructure. However, a head-to-head comparison of DW-MRI models is critically missing in the field. To address this deficiency, we organized the "White Matter Modeling Challenge" during the International Symposium on Biomedical Imaging (ISBI) 2015 conference. This competition aimed at identifying the DW-MRI models that best predict unseen DW data. in vivo DW-MRI data was acquired on the Connectom scanner at the A.A.Martinos Center (Massachusetts General Hospital) using gradients strength of up to 300 mT/m and a broad set of diffusion times. We focused on assessing the DW signal prediction in two regions: the genu in the corpus callosum, where the fibres are relatively straight and parallel, and the fornix, where the configuration of fibres is more complex. The challenge participants had access to three-quarters of the whole dataset, and their models were ranked on their ability to predict the remaining unseen quarter of data. In this paper we provide both an overview and a more in-depth description of each evaluated model, report the challenge results, and infer trends about the model characteristics that were associated with high model ranking. This work provides a much needed benchmark for DW-MRI models. The acquired data and model details for signal prediction evaluation are provided online to encourage a larger scale assessment of diffusion models in the future.

physics.med-ph↗

arXiv2013

Diffusion imaging with stimulated echoes: signal models and experiment design

Purpose: Stimulated echo acquisition mode (STEAM) diffusion MRI can be advantageous over pulsed-gradient spin-echo (PGSE) for diffusion times that are long compared to $\ttwo$. It is important therefore for biomedical diffusion imaging applications at 7T and above where $\ttwo$ is short. However, imaging gradients in the STEAM sequence contribute much greater diffusion weighting than in PGSE, but are often ignored during post-processing. We demonstrate here that this can severely bias parameter estimates. Method: We present models for the STEAM signal for free and restricted diffusion that account for crusher and slice-select (butterfly) gradients to avoid such bias. The butterfly gradients also disrupt experiment design, typically by skewing gradient-vectors towards the slice direction. We propose a simple compensation to the diffusion gradient vector specified to the scanner that counterbalances the butterfly gradients to preserve the intended experiment design. Results: High-field data fixed from a monkey brain experiments demonstrate the need for both the compensation during acquisition and correct modelling during post-processing for both diffusion tensor imaging and ActiveAx axon-diameter index mapping. Simulations support the results and indicate a similar need in in-vivo human applications. Conclusion: Correct modelling and compensation are important for practical applications of STEAM diffusion MRI.

physics.med-ph↗