Arrow Research search

Author name cluster

Stephen Smith

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

34 papers
1 author row

Possible papers

34

YNIMG Journal 2022 Journal Article

Accurate predictions of individual differences in task-evoked brain activity from resting-state fMRI using a sparse ensemble learner

  • Ying-Qiu Zheng
  • Seyedeh-Rezvan Farahibozorg
  • Weikang Gong
  • Hossein Rafipoor
  • Saad Jbabdi
  • Stephen Smith

Modelling and predicting individual differences in task-fMRI activity can have a wide range of applications from basic to clinical neuroscience. It has been shown that models based on resting-state activity can have high predictive accuracy. Here we propose several improvements to such models. Using a sparse ensemble learner, we show that (i) features extracted using Stochastic Probabilistic Functional Modes (sPROFUMO) outperform the previously proposed dual-regression approach, (ii) that the shape and overall intensity of individualised task activations can be modelled separately and explicitly, (iii) training the model on predicting residual differences in brain activity further boosts individualised predictions. These results hold for both surface-based analyses of the Human Connectome Project data as well as volumetric analyses of UK-biobank data. Overall, our model achieves state of the art prediction accuracy on par with the test-retest reliability of task-fMRI scans, suggesting that it has potential to supplement traditional task localisers.

Details DOI

NeurIPS Conference 2022 Conference Paper

Biologically-plausible backpropagation through arbitrary timespans via local neuromodulators

  • Yuhan Helena Liu
  • Stephen Smith
  • Stefan Mihalas
  • Eric Shea-Brown
  • Uygar Sümbül

The spectacular successes of recurrent neural network models where key parameters are adjusted via backpropagation-based gradient descent have inspired much thought as to how biological neuronal networks might solve the corresponding synaptic credit assignment problem [1, 2, 3]. There is so far little agreement, however, as to how biological networks could implement the necessary backpropagation through time, given widely recognized constraints of biological synaptic network signaling architectures. Here, we propose that extra-synaptic diffusion of local neuromodulators such as neuropeptides may afford an effective mode of backpropagation lying within the bounds of biological plausibility. Going beyond existing temporal truncation-based gradient approximations [4, 5, 6], our approximate gradient-based update rule, ModProp, propagates credit information through arbitrary time steps. ModProp suggests that modulatory signals can act on receiving cells by convolving their eligibility traces via causal, time-invariant and synapse-type-specific filter taps. Our mathematical analysis of ModProp learning, together with simulation results on benchmark temporal tasks, demonstrate the advantage of ModProp over existing biologically-plausible temporal credit assignment rules. These results suggest a potential neuronal mechanism for signaling credit information related to recurrent interactions over a longer time horizon. Finally, we derive an in-silico implementation of ModProp that could serve as a low-complexity and causal alternative to backpropagation through time.

PDF Details

YNIMG Journal 2021 Journal Article

The nonhuman primate neuroimaging and neuroanatomy project

  • Takuya Hayashi
  • Yujie Hou
  • Matthew F Glasser
  • Joonas A Autio
  • Kenneth Knoblauch
  • Miho Inoue-Murayama
  • Tim Coalson
  • Essa Yacoub

Multi-modal neuroimaging projects such as the Human Connectome Project (HCP) and UK Biobank are advancing our understanding of human brain architecture, function, connectivity, and their variability across individuals using high-quality non-invasive data from many subjects. Such efforts depend upon the accuracy of non-invasive brain imaging measures. However, 'ground truth' validation of connectivity using invasive tracers is not feasible in humans. Studies using nonhuman primates (NHPs) enable comparisons between invasive and non-invasive measures, including exploration of how "functional connectivity" from fMRI and "tractographic connectivity" from diffusion MRI compare with long-distance connections measured using tract tracing. Our NonHuman Primate Neuroimaging & Neuroanatomy Project (NHP_NNP) is an international effort (6 laboratories in 5 countries) to: (i) acquire and analyze high-quality multi-modal brain imaging data of macaque and marmoset monkeys using protocols and methods adapted from the HCP; (ii) acquire quantitative invasive tract-tracing data for cortical and subcortical projections to cortical areas; and (iii) map the distributions of different brain cell types with immunocytochemical stains to better define brain areal boundaries. We are acquiring high-resolution structural, functional, and diffusion MRI data together with behavioral measures from over 100 individual macaques and marmosets in order to generate non-invasive measures of brain architecture such as myelin and cortical thickness maps, as well as functional and diffusion tractography-based connectomes. We are using classical and next-generation anatomical tracers to generate quantitative connectivity maps based on brain-wide counting of labeled cortical and subcortical neurons, providing ground truth measures of connectivity. Advanced statistical modeling techniques address the consistency of both kinds of data across individuals, allowing comparison of tracer-based and non-invasive MRI-based connectivity measures. We aim to develop improved cortical and subcortical areal atlases by combining histological and imaging methods. Finally, we are collecting genetic and sociality-associated behavioral data in all animals in an effort to understand how genetic variation shapes the connectome and behavior.

Details DOI

NeurIPS Conference 2020 Conference Paper

ICAM: Interpretable Classification via Disentangled Representations and Feature Attribution Mapping

  • Cher Bass
  • Mariana da Silva
  • Carole Sudre
  • Petru-Daniel Tudosiu
  • Stephen Smith
  • Emma Robinson

Feature attribution (FA), or the assignment of class-relevance to different locations in an image, is important for many classification problems but is particularly crucial within the neuroscience domain, where accurate mechanistic models of behaviours, or disease, require knowledge of all features discriminative of a trait. At the same time, predicting class relevance from brain images is challenging as phenotypes are typically heterogeneous, and changes occur against a background of significant natural variation. Here, we present a novel framework for creating class specific FA maps through image-to-image translation. We propose the use of a VAE-GAN to explicitly disentangle class relevance from background features for improved interpretability properties, which results in meaningful FA maps. We validate our method on 2D and 3D brain image datasets of dementia (ADNI dataset), ageing (UK Biobank), and (simulated) lesion detection. We show that FA maps generated by our method outperform baseline FA methods when validated against ground truth. More significantly, our approach is the first to use latent space sampling to support exploration of phenotype variation.

PDF Details

YNIMG Journal 2020 Journal Article

Towards HCP-Style macaque connectomes: 24-Channel 3T multi-array coil, MRI sequences and preprocessing

  • Joonas A. Autio
  • Matthew F. Glasser
  • Takayuki Ose
  • Chad J. Donahue
  • Matteo Bastiani
  • Masahiro Ohno
  • Yoshihiko Kawabata
  • Yuta Urushibata

Macaque monkeys are an important animal model where invasive investigations can lead to a better understanding of the cortical organization of primates including humans. However, the tools and methods for noninvasive image acquisition (e. g. MRI RF coils and pulse sequence protocols) and image data preprocessing have lagged behind those developed for humans. To resolve the structural and functional characteristics of the smaller macaque brain, high spatial, temporal, and angular resolutions combined with high signal-to-noise ratio are required to ensure good image quality. To address these challenges, we developed a macaque 24-channel receive coil for 3-T MRI with parallel imaging capabilities. This coil enables adaptation of the Human Connectome Project (HCP) image acquisition protocols to the in-vivo macaque brain. In addition, we adapted HCP preprocessing methods to the macaque brain, including spatial minimal preprocessing of structural, functional MRI (fMRI), and diffusion MRI (dMRI). The coil provides the necessary high signal-to-noise ratio and high efficiency in data acquisition, allowing four- and five-fold accelerations for dMRI and fMRI. Automated FreeSurfer segmentation of cortex, reconstruction of cortical surface, removal of artefacts and nuisance signals in fMRI, and distortion correction of dMRI all performed well, and the overall quality of basic neurobiological measures was comparable with those for the HCP. Analyses of functional connectivity in fMRI revealed high sensitivity as compared with those from publicly shared datasets. Tractography-based connectivity estimates correlated with tracer connectivity similarly to that achieved using ex-vivo dMRI. The resulting HCP-style in vivo macaque MRI data show considerable promise for analyzing cortical architecture and functional and structural connectivity using advanced methods that have previously only been available in studies of the human brain.

Details DOI

YNIMG Journal 2019 Journal Article

Optimising neonatal fMRI data analysis: Design and validation of an extended dHCP preprocessing pipeline to characterise noxious-evoked brain activity in infants

  • Luke Baxter
  • Sean Fitzgibbon
  • Fiona Moultrie
  • Sezgi Goksan
  • Mark Jenkinson
  • Stephen Smith
  • Jesper Andersson
  • Eugene Duff

The infant brain is unlike the adult brain, with considerable differences in morphological, neurodynamic, and haemodynamic features. As the majority of current MRI analysis tools were designed for use in adults, a primary objective of the Developing Human Connectome Project (dHCP) is to develop optimised methodological pipelines for the analysis of neonatal structural, resting state, and diffusion MRI data. Here, in an independent neonatal dataset we have extended and optimised the dHCP fMRI preprocessing pipeline for the analysis of stimulus-response fMRI data. We describe and validate this extended dHCP fMRI preprocessing pipeline to analyse changes in brain activity evoked following an acute noxious stimulus applied to the infant's foot. We compare the results obtained from this extended dHCP pipeline to results obtained from a typical FSL FEAT-based analysis pipeline, evaluating the pipelines' outputs using a wide range of tests. We demonstrate that a substantial increase in spatial specificity and sensitivity to signal can be attained with a bespoke neonatal preprocessing pipeline through optimised motion and distortion correction, ICA-based denoising, and haemodynamic modelling. The improved sensitivity and specificity, made possible with this extended dHCP pipeline, will be paramount in making further progress in our understanding of the development of sensory processing in the infant brain.

Details DOI

YNIMG Journal 2019 Journal Article

Probabilistic TFCE: A generalized combination of cluster size and voxel intensity to increase statistical power

  • Tamás Spisák
  • Zsófia Spisák
  • Matthias Zunhammer
  • Ulrike Bingel
  • Stephen Smith
  • Thomas Nichols
  • Tamás Kincses

The threshold-free cluster enhancement (TFCE) approach integrates cluster information into voxel-wise statistical inference to enhance detectability of neuroimaging signal. Despite the significantly increased sensitivity, the application of TFCE is limited by several factors: (i) generalisation to data structures, like brain network connectivity data is not trivial, (ii) TFCE values are in an arbitrary unit, therefore, P-values can only be obtained by a computationally demanding permutation-test. Here, we introduce a probabilistic approach for TFCE (pTFCE), that gives a simple general framework for topology-based belief boosting. The core of pTFCE is a conditional probability, calculated based on Bayes' rule, from the probability of voxel intensity and the threshold-wise likelihood function of the measured cluster size. In this paper, we provide an estimation of these distributions based on Gaussian Random Field theory. The conditional probabilities are then aggregated across cluster-forming thresholds by a novel incremental aggregation method. pTFCE is validated on simulated and real fMRI data. The results suggest that pTFCE is more robust to various ground truth shapes and provides a stricter control over cluster “leaking” than TFCE and, in many realistic cases, further improves its sensitivity. Correction for multiple comparisons can be trivially performed on the enhanced P-values, without the need for permutation testing, thus pTFCE is well-suitable for the improvement of statistical inference in any neuroimaging workflow. Implementation of pTFCE is available at https: //spisakt. github. io/pTFCE.

Details DOI

YNIMG Journal 2019 Journal Article

Using GPUs to accelerate computational diffusion MRI: From microstructure estimation to tractography and connectomes

  • Moises Hernandez-Fernandez
  • Istvan Reguly
  • Saad Jbabdi
  • Mike Giles
  • Stephen Smith
  • Stamatios N. Sotiropoulos

The great potential of computational diffusion MRI (dMRI) relies on indirect inference of tissue microstructure and brain connections, since modelling and tractography frameworks map diffusion measurements to neuroanatomical features. This mapping however can be computationally highly expensive, particularly given the trend of increasing dataset sizes and the complexity in biophysical modelling. Limitations on computing resources can restrict data exploration and methodology development. A step forward is to take advantage of the computational power offered by recent parallel computing architectures, especially Graphics Processing Units (GPUs). GPUs are massive parallel processors that offer trillions of floating point operations per second, and have made possible the solution of computationally-intensive scientific problems that were intractable before. However, they are not inherently suited for all problems. Here, we present two different frameworks for accelerating dMRI computations using GPUs that cover the most typical dMRI applications: a framework for performing biophysical modelling and microstructure estimation, and a second framework for performing tractography and long-range connectivity estimation. The former provides a front-end and automatically generates a GPU executable file from a user-specified biophysical model, allowing accelerated non-linear model fitting in both deterministic and stochastic ways (Bayesian inference). The latter performs probabilistic tractography, can generate whole-brain connectomes and supports new functionality for imposing anatomical constraints, such as inherent consideration of surface meshes (GIFTI files) along with volumetric images. We validate the frameworks against well-established CPU-based implementations and we show that despite the very different challenges for parallelising these problems, a single GPU achieves better performance than 200 CPU cores thanks to our parallel designs.

Details DOI

AAAI Conference 2018 Conference Paper

Expressive Real-Time Intersection Scheduling

  • Rick Goldstein
  • Stephen Smith

We present Expressive Real-time Intersection Scheduling (ERIS), a schedule-driven control strategy for adaptive intersection control to reduce traffic congestion. ERIS maintains separate estimates for each lane approaching a traffic intersection allowing it to more accurately estimate the effects of scheduling decisions than previous schedule-driven approaches. We present a detailed description of the search space and A* search heuristic employed by ERIS to make scheduling decisions in real-time (every second). As a result of its increased expressiveness, ERIS outperforms a less expressive schedule-driven approach and a fully-actuated control method in a variety of simulated traffic environments.

PDF Details

YNICL Journal 2018 Journal Article

Stratification of MDD and GAD patients by resting state brain connectivity predicts cognitive bias

  • Janine D. Bijsterbosch
  • Tahereh L. Ansari
  • Stephen Smith
  • Oliver Gauld
  • Ondrej Zika
  • Sirius Boessenkool
  • Michael Browning
  • Andrea Reinecke

Patients with Generalized Anxiety Disorder (GAD) and Major Depressive Disorder (MDD) show between-group comorbidity and symptom overlap, and within-group heterogeneity. Resting state functional connectivity might provide an alternate, biologically informed means by which to stratify patients with GAD or MDD. Resting state functional magnetic resonance imaging data were acquired from 23 adults with GAD, 21 adults with MDD, and 27 healthy adult control participants. We investigated whether within- or between-network connectivity indices from five resting state networks predicted scores on continuous measures of depression and anxiety. Successful predictors were used to stratify participants into two new groups. We examined whether this stratification predicted attentional bias towards threat and whether this varied between patients and controls. Depression scores were linked to elevated connectivity within a limbic network including the amygdala, hippocampus, VMPFC and subgenual ACC. Patients with GAD or MDD with high limbic connectivity showed poorer performance on an attention-to-threat task than patients with low limbic connectivity. No parallel effect was observed for control participants, resulting in an interaction of clinical status by resting state group. Our findings provide initial evidence for the external validity of stratification of MDD and GAD patients by functional connectivity markers. This stratification cuts across diagnostic boundaries and might valuably inform future intervention studies. Our findings also highlight that biomarkers of interest can have different cognitive correlates in individuals with versus without clinically significant symptomatology. This might reflect protective influences leading to resilience in some individuals but not others.

Details DOI

YNIMG Journal 2017 Journal Article

Distinct resting-state functional connections associated with episodic and visuospatial memory in older adults

  • Sana Suri
  • Anya Topiwala
  • Nicola Filippini
  • Enikő Zsoldos
  • Abda Mahmood
  • Claire E. Sexton
  • Archana Singh-Manoux
  • Mika Kivimäki

Episodic and spatial memory are commonly impaired in ageing and Alzheimer's disease. Volumetric and task-based functional magnetic resonance imaging (fMRI) studies suggest a preferential involvement of the medial temporal lobe (MTL), particularly the hippocampus, in episodic and spatial memory processing. The present study examined how these two memory types were related in terms of their associated resting-state functional architecture. 3T multiband resting state fMRI scans from 497 participants (60–82 years old) of the cross-sectional Whitehall II Imaging sub-study were analysed using an unbiased, data-driven network-modelling technique (FSLNets). Factor analysis was performed on the cognitive battery; the Hopkins Verbal Learning test and Rey-Osterreith Complex Figure test factors were used to assess verbal and visuospatial memory respectively. We present a map of the macroscopic functional connectome for the Whitehall II Imaging sub-study, comprising 58 functionally distinct nodes clustered into five major resting-state networks. Within this map we identified distinct functional connections associated with verbal and visuospatial memory. Functional anticorrelation between the hippocampal formation and the frontal pole was significantly associated with better verbal memory in an age-dependent manner. In contrast, hippocampus–motor and parietal–motor functional connections were associated with visuospatial memory independently of age. These relationships were not driven by grey matter volume and were unique to the respective memory domain. Our findings provide new insights into current models of brain-behaviour interactions, and suggest that while both episodic and visuospatial memory engage MTL nodes of the default mode network, the two memory domains differ in terms of the associated functional connections between the MTL and other resting-state brain networks.

Details DOI

YNIMG Journal 2017 Journal Article

Investigations into within- and between-subject resting-state amplitude variations

  • Janine Bijsterbosch
  • Samuel Harrison
  • Eugene Duff
  • Fidel Alfaro-Almagro
  • Mark Woolrich
  • Stephen Smith

The amplitudes of spontaneous fluctuations in brain activity may be a significant source of within-subject and between-subject variability, and this variability is likely to be carried through into functional connectivity (FC) estimates (whether directly or indirectly). Therefore, improving our understanding of amplitude fluctuations over the course of a resting state scan and variation in amplitude across individuals is of great relevance to the interpretation of FC findings. We investigate resting state amplitudes in two large-scale studies (HCP and UK Biobank), with the aim of determining between-subject and within-subject variability. Between-subject clustering distinguished between two groups of brain networks whose amplitude variation across subjects were highly correlated with each other, revealing a clear distinction between primary sensory and motor regions (‘primary sensory/motor cluster’) and cognitive networks. Within subjects, all networks in the primary sensory/motor cluster showed a consistent increase in amplitudes from the start to the end of the scan. In addition to the strong increases in primary sensory/motor amplitude, a large number of changes in FC were found when comparing the two scans acquired on the same day (HCP data). Additive signal change analysis confirmed that all of the observed FC changes could be fully explained by changes in amplitude. Between-subject correlations in UK Biobank data showed a negative correlation between primary sensory/motor amplitude and average sleep duration, suggesting a role of arousal. Our findings additionally reveal complex relationships between amplitude and head motion. These results suggest that network amplitude is a source of significant variability both across subjects, and within subjects on a within-session timescale. Future rfMRI studies may benefit from obtaining arousal-related (self report) measures, and may wish to consider the influence of amplitude changes on measures of (dynamic) functional connectivity.

Details DOI

YNIMG Journal 2014 Journal Article

Optimizing full-brain coverage in human brain MRI through population distributions of brain size

  • Maarten Mennes
  • Mark Jenkinson
  • Romain Valabregue
  • Jan K. Buitelaar
  • Christian Beckmann
  • Stephen Smith

When defining an MRI protocol, brain researchers need to set multiple interdependent parameters that define repetition time (TR), voxel size, field-of-view (FOV), etc. Typically, researchers aim to image the full brain, making the expected FOV an important parameter to consider. Especially in 2D-EPI sequences, non-wasteful FOV settings are important to achieve the best temporal and spatial resolution. In practice, however, imperfect FOV size estimation often results in partial brain coverage for a significant number of participants per study, or, alternatively, an unnecessarily large voxel-size or number of slices to guarantee full brain coverage. To provide normative FOV guidelines we estimated population distributions of brain size in the x-, y-, and z-direction using data from 14, 781 individuals. Our results indicated that 11mm in the z-direction differentiate between obtaining full brain coverage for 90% vs. 99. 9% of participants. Importantly, we observed that rotating the FOV to optimally cover the brain, and thus minimize the number of slices needed, effectively reduces the required inferior–superior FOV size by ~5%. For a typical adult imaging study, 99. 9% of the population can be imaged with full brain coverage when using an inferior–superior FOV of 142mm, assuming optimal slice orientation and minimal within-scan head motion. By providing population distributions for brain size in the x-, y-, and z-direction we improve the potential for obtaining full brain coverage, especially in 2D-EPI sequences used in most functional and diffusion MRI studies. We further enable optimization of related imaging parameters including the number of slices, TR and total acquisition time.

Details DOI

AAMAS Conference 2012 Conference Paper

Coordinated Look-Ahead Scheduling for Real-Time Traffic Signal Control

  • Xiao-Feng Xie
  • Stephen Smith
  • Gregory J. Barlow

We take an agent-based approach to real-time traffic signal control based on coordinated look-ahead scheduling. At each decision point, each agent constructs a schedule that optimizes movement of the currently approaching traffic through its intersection. For strengthening its local view, each agent queries the scheduled outflows from its direct upstream neighbors to obtain an optimistic observation, which is capable of incorporating non-local impacts from indirect neighbors. We summarize results on a road network of tightly-coupled intersections that demonstrate the ability of our approach. 1

PDF

AAAI Conference 2012 Conference Paper

Incremental Management of Oversubscribed Vehicle Schedules in Dynamic Dial-A-Ride Problems

  • Zachary Rubinstein
  • Stephen Smith
  • Laura Barbulescu

In this paper, we consider the problem of feasibly integrating new pick-up and delivery requests into existing vehicle itineraries in a dynamic, dial-a-ride problem (DARP) setting. Generalizing from previous work in oversubscribed task scheduling, we define a controlled iterative repair search procedure for finding an alternative set of vehicle itineraries in which the overall solution has been feasibly extended to include newly received requests. We first evaluate the performance of this technique on a set of DARP feasibility benchmark problems from the literature. We then consider its use on a real-world DARP problem, where it is necessary to accommodate all requests and constraints must be relaxed when a request cannot be feasibly integrated. For this latter analysis, we introduce a constraint relaxation post processing step and consider the performance impact of using our controlled iterative search over the current greedy search approach.

PDF Details

YNIMG Journal 2011 Journal Article

DTI measures in crossing-fibre areas: Increased diffusion anisotropy reveals early white matter alteration in MCI and mild Alzheimer's disease

  • Gwenaëlle Douaud
  • Saâd Jbabdi
  • Timothy E.J. Behrens
  • Ricarda A. Menke
  • Achim Gass
  • Andreas U. Monsch
  • Anil Rao
  • Brandon Whitcher

Though mild cognitive impairment is an intermediate clinical state between healthy aging and Alzheimer's disease (AD), there are very few whole-brain voxel-wise diffusion MRI studies directly comparing changes in healthy control, mild cognitive impairment (MCI) and AD subjects. Here we report whole-brain findings from a comprehensive study of diffusion tensor indices and probabilistic tractography obtained in a very large population of healthy controls, MCI and probable AD subjects. As expected from the literature, all diffusion indices converged to show that the cingulum bundle, the uncinate fasciculus, the entire corpus callosum and the superior longitudinal fasciculus are the most affected white matter tracts in AD. Significant differences between MCI and AD were essentially confined to the corpus callosum. More importantly, we introduce for the first time in a degenerative disorder an application of a recently developed tensor index, the “mode” of anisotropy, as well as probabilistic crossing-fibre tractography. The mode of anisotropy specifies the type of anisotropy as a continuous measure reflecting differences in shape of the diffusion tensor ranging from planar (e. g. , in regions of crossing fibres from two fibre populations of similar density or regions of “kissing” fibres) to linear (e. g. , in regions where one fibre population orientation predominates), while probabilistic crossing-fibre tractography allows to accurately trace pathways from a crossing-fibre region. Remarkably, when looking for whole-brain diffusion differences between MCI patients and healthy subjects, the only region with significant abnormalities was a region of crossing fibres in the centrum semiovale, showing an increased mode of anisotropy. The only white matter region demonstrating a significant difference in correlations between neuropsychological scores and a diffusion measure (mode of anisotropy) across the three groups was the same region of crossing fibres. Further examination using probabilistic tractography established explicitly and quantitatively that this previously unreported increase of mode and co-localised increase of fractional anisotropy was explained by a relative preservation of motor-related projection fibres (at this early stage of the disease) crossing the association fibres of the superior longitudinal fasciculus. These findings emphasise the benefit of looking at the more complex regions in which spared and affected pathways are crossing to detect very early alterations of the white matter that could not be detected in regions consisting of one fibre population only. Finally, the methods used in this study may have general applicability for other degenerative disorders and, beyond the clinical sphere, they could contribute to a better quantification and understanding of subtle effects generated by normal processes such as visuospatial attention or motor learning.

Details DOI

AAMAS Conference 2010 Conference Paper

Distributed Coordination of Mobile Agent Teams: The Advantage of Planning Ahead

  • Laura Barbulescu
  • Zachary Rubinstein
  • Stephen Smith
  • Terry Zimmerman

We consider the problem of coordinating a team of agentsengaged in executing a set of inter-dependent, geographicallydispersed tasks in an oversubscribed and uncertain environment. In such domains, where there are sequence-dependentsetup activities (e. g. , travel), we argue that there is inherent leverage to having agents maintain advance schedules. In the distributed problem solving setting we consider, eachagent begins with a task itinerary, and, as execution unfolds and dynamics ensue (e. g. , tasks fail, new tasks are discovered, etc. ), agents must coordinate to extend and revisetheir plans accordingly. The team objective is to maximizethe utility accrued from executed actions over a given timehorizon. Our approach to solving this problem is based ondistributed management of agent schedules. We describe anagent architecture that uses the synergy between intra-agentscheduling and inter-agent coordination to promote task allocation decisions that minimize travel time and maximizetime available for utility-acrruing activities. Experimentalresults are presented that compare our agent's performanceto that of an agent using an intelligent dispatching strategypreviously shown to outperform our approach on synthetic, stateless, utility maximization problems. Across a range ofproblems involving a mix of situated and non-situated tasksour advance scheduling approach dominates this same dispatch strategy. Finally, we report performance results withan extension of the system on a limited set of field test experiments.

PDF

YNIMG Journal 2009 Journal Article

An evaluation of four automatic methods of segmenting the subcortical structures in the brain

  • Kolawole Oluwole Babalola
  • Brian Patenaude
  • Paul Aljabar
  • Julia Schnabel
  • David Kennedy
  • William Crum
  • Stephen Smith
  • Tim Cootes

The automation of segmentation of subcortical structures in the brain is an active research area. We have comprehensively evaluated four novel methods of fully automated segmentation of subcortical structures using volumetric, spatial overlap and distance-based measures. Two methods are atlas-based — classifier fusion and labelling (CFL) and expectation–maximisation segmentation using a brain atlas (EMS), and two incorporate statistical models of shape and appearance — profile active appearance models (PAM) and Bayesian appearance models (BAM). Each method was applied to the segmentation of 18 subcortical structures in 270 subjects from a diverse pool varying in age, disease, sex and image acquisition parameters. Our results showed that all four methods perform on par with recently published methods. CFL performed better than the others according to all three classes of metrics. In summary over all structures, the ranking by the Dice coefficient was CFL, BAM, joint EMS and PAM. The Hausdorff distance ranked the methods as CFL, joint PAM and BAM, EMS, whilst percentage absolute volumetric difference ranked them as joint CFL and PAM, joint BAM and EMS. Furthermore, as we had four methods of performing segmentation, we investigated whether the results obtained by each method were more similar to each other than to the manual segmentations using Williams' Index. Reassuringly, the Williams' Index was close to 1 for most subjects (mean=1. 02, sd=0. 05), indicating better agreement of each method with the gold standard than with the other methods. However, 2% of cases (mainly amygdala and nucleus accumbens) had values outside 3 standard deviations of the mean.

Details DOI

AAMAS Conference 2008 Conference Paper

A Few Good Agents: Multi-Agent Social Learning

  • Jean Oh
  • Stephen Smith

In this paper, we investigate multi-agent learning (MAL) in a multi-agent resource selection problem (MARS) in which a large group of agents are competing for common resources. Since agents in such a setting are self-interested, MAL in MARS domains typically focuses on the convergence to a set of non-cooperative equilibria. As seen in the example of prisoner’s dilemma, however, selfish equilibria are not necessarily optimal with respect to the natural objective function of a target problem, e. g. , resource utilization in the case of MARS. Conversely, a centrally administered optimization of physically distributed agents is infeasible in many reallife applications such as transportation traffic problems. In order to explore the possibility for a middle ground solution, we analyze two types of costs for evaluating MAL algorithms in this context. The quality loss of a selfish algorithm can be quantitatively measured by the price of anarchy, i. e. , the ratio of the objective function value of a selfish solution to that of an optimal solution. Analogously, we introduce the price of monarchy of a learning algorithm to quantify the practical cost of coordination in terms of communication cost. We then introduce a multi-agent social learning approach named A Few Good Agents (AFGA) that motivates selfinterested agents to cooperate with one another to reduce the price of anarchy, while bounding the price of monarchy at the same time. A preliminary set of experiments on the El Farol bar problem, a simple example of MARS, show promising results.

PDF

YNIMG Journal 2008 Journal Article

Evidence for abnormalities of cortical development in adolescent-onset schizophrenia

  • Natalie L. Voets
  • Morgan G. Hough
  • Gwenaelle Douaud
  • Paul M. Matthews
  • Anthony James
  • Louise Winmill
  • Paula Webster
  • Stephen Smith

Voxel-Based Morphometry (VBM) identifies differences in grey matter brain structure in patients with schizophrenia relative to healthy controls, with particularly prominent differences found in patients with the more severe, adolescent-onset form of the disease. However, as VBM is sensitive to a combination of changes in grey matter thickness, intensity and folding, specific neuropathological interpretations are not possible. Here, we attempt to more precisely define cortical changes in 25 adolescent-onset schizophrenic patients and 25 age- and sex-matched healthy volunteers using Surface-Based Morphometry (SBM) to disambiguate the relative contributions of cortical thickness and surface area differences to changes in regional grey matter (GM) density measured with VBM. Cortical changes in schizophrenia were widespread, including particularly the prefrontal cortex and superior temporal gyrus. Nine regions of apparent reduction in GM density in patients relative to healthy matched controls were found using VBM that were not found with SBM-derived cortical thickness measures. In Regions of Interest (ROIs) derived from the VBM group results, we confirmed that local surface area differences accounted for these VBM changes. Our results emphasize widespread, but focally distinct cortical pathology in adolescent-onset schizophrenia. Evidence for changes in local surface area (as opposed to simply cortical thinning) is consistent with a neurodevelopmental contribution to the underlying neuropathology of the disease.

Details DOI

YNIMG Journal 2007 Journal Article

Meaningful design and contrast estimability in FMRI

  • Stephen Smith
  • Mark Jenkinson
  • Christian Beckmann
  • Karla Miller
  • Mark Woolrich

Optimising the efficiency of an experimental design is known to be of great importance. However, existing methods for calculating design rank deficiency and contrast estimability (an important aspect of experimental design) relate to computational precision rather than image noise and are therefore not very meaningful. For example, a contrast between two experimental conditions may be mathematically “estimable” while requiring a huge differential BOLD response for statistical significance to be reached. In this paper we formulate standard efficiency equations in terms of required BOLD effect, and use this to generate measures of rank/estimability which are meaningful. This takes into account the strength and smoothness of the timeseries noise and is applicable to complex contrasts; we show how to re-express several regressors and an associated contrast vector as a single equivalent regressor, so that we can calculate the contrast's effective peak–peak height unambiguously. We also present some example results on typical designs, and characterise noise results from a range of typical FMRI acquisitions, in order to allow experimenters to apply efficiency estimation in advance of acquiring data.

Details DOI

YNIMG Journal 2002 Journal Article

Effects of Word Form on Brain Processing of Written Chinese

  • Shimin Fu
  • Yiping Chen
  • Stephen Smith
  • Susan Iversen
  • P.M. Matthews

Both logographic characters and alphabetic pinyins can be used to write words in Chinese. Here we use fMRI to address the question of whether the written form affects brain processing of a word. Fifteen healthy, right-handed, native Chinese-reading volunteers participated in our study and were asked to read silently either Chinese characters (8 subjects) or pinyins (7 subjects). The stimulus presentation rate was varied for both tasks to allow us to identify brain regions with word-load-dependent activation. Rate effects (fast minus slow presentations) for Chinese character reading were observed in striate and extrastriate visual cortex, superior parietal lobule, left posterior middle temporal gyrus, bilateral inferior temporal gyri, and bilateral superior frontal gyri. Rate effects for pinyin reading were observed in bilateral fusiform, lingual, and middle occipital gyri, bilateral superior parietal lobule/precuneus, left inferior parietal lobule, bilateral inferior temporal gyrus, left middle temporal gyrus, and left superior temporal gyrus. These results demonstrate that common regions of the brain are involved in reading both Chinese characters and pinyins, activated apparently independently of the surface form of the word. There also appear to be brain regions in which activation is dependent on word form. However, it is unlikely that these are entirely specific for a given word form; their activation more likely reflects relative functional specializations within broader networks for processing written language.

Details DOI

YNIMG Journal 2002 Journal Article

Improved Optimization for the Robust and Accurate Linear Registration and Motion Correction of Brain Images

  • Mark Jenkinson
  • Peter Bannister
  • Michael Brady
  • Stephen Smith

Linear registration and motion correction are important components of structural and functional brain image analysis. Most modern methods optimize some intensity-based cost function to determine the best registration. To date, little attention has been focused on the optimization method itself, even though the success of most registration methods hinges on the quality of this optimization. This paper examines the optimization process in detail and demonstrates that the commonly used multiresolution local optimization methods can, and do, get trapped in local minima. To address this problem, two approaches are taken: (1) to apodize the cost function and (2) to employ a novel hybrid global–local optimization method. This new optimization method is specifically designed for registering whole brain images. It substantially reduces the likelihood of producing misregistrations due to being trapped by local minima. The increased robustness of the method, compared to other commonly used methods, is demonstrated by a consistency test. In addition, the accuracy of the registration is demonstrated by a series of experiments with motion correction. These motion correction experiments also investigate how the results are affected by different cost functions and interpolation methods.

Details DOI