Dr Kevin Murphy

Wellcome Trust Career Development Fellow

Research group:
029 206 88743
CUBRIC, Maindy Road

Research summary

The broad theme of my research has been to understand variability and noise in FMRI data with the aim of improving our understanding of the relationship between neuronal activity and BOLD signal changes measured by FMRI. Currently, I am working on a Wellcome Trust-funded fellowship aimed at quantifying vascular influences on neurovascular coupling. For FMRI to reach its maximum potential, neurovascular coupling changes must be understood and appropriately considered. Using ageing as a model of altered vasculature, I am developing the necessary neuroimaging tools to assess cerebrovascular health. The resulting neurovascular coupling corrections promise to substantially enhance the utility of FMRI in both healthy and clinical groups, benefiting neurological health in an ageing population and translating to any patients with altered vascular dynamics.

Selected publications (2014 onwards)


Full list of publications


Research topics and related papers

Quantifying vascular influences on neurovascular coupling with fMRI
(Funding: Wellcome Trust Research Career Development Fellowship)

Neuroimaging is a research area that uses scanners to investigate the workings of the brain. FMRI is one technique that measures changes in blood flow related to brain activity. When neurons are stimulated, they consume oxygen which causes blood flow changes nearby. This can be used to determine which areas of the brain perform which functions. However, this FMRI BOLD signal is not a direct measure of neural activity. When comparing groups, the implicit assumption that neurovascular coupling properties are equivalent is made. Any alterations in these cerebrovascular dynamics lead to BOLD differences that are unrelated to neuronal activity. For example, normal ageing can cause blood vessels to stiffen. This alters the vessels’ response by causing signal differences that are unrelated to brain activity. Interpreting results can be difficult since we can’t tell if signal differences originate from neurons or from vessels. Using an elderly group as a model for altered coupling, this project is investigating such BOLD differences and developing suitable correction strategies.

The project has three main objectives that are incorporated into an FMRI-cerebrovascular modelling framework:
1) Quantify the influence of dynamic cerebral autoregulation on the BOLD signal to account for blood pressure-related perfusion fluctuations;
2) Account for vascular reactivity-related changes in BOLD signals, thus providing a novel marker of cerebrovascular health;
3) Translate these cerebrovascular models into young/elderly group comparisons by testing their ability to characterise group differences that are present in MEG-derived neuronal activity measures.

In both young and elderly people, I am examining blood flow (CBF) which is affected by blood pressure and carbon dioxide (CO2) levels. By altering arterial CO2 levels and measuring blood pressure along with pulse wave velocity as a marker of systemic tone, I am developing correction strategies that account for neurovascular coupling with age. Direct measures of neural activity using MEG will be used for comparisons.

Since normal ageing and many clinical conditions are accompanied by alterations in microvasculature, FMRI will be unreliable unless appropriate corrections for neurovascular coupling differences are made. Modelling strategies that address these concerns not only increase the robustness of normal ageing FMRI studies but translate to many clinical conditions that are more frequent in the ageing population such as hypertension, dementia and diabetes.

Optimising Pharmacological FMRI for Drug Development
(Funding: Pfizer Ltd.)

FMRI of the human brain is a promising technique for assessing measures of CNS drug action in pharmacological development. The most ubiquitous method for mapping functional activity measures the blood oxygenation level dependent (BOLD) signal. Direct measurement of cerebral perfusion/cerebral blood flow (CBF) can also be performed using arterial spin labelling (ASL). This project optimises FMRI experimental procedures for pharmacological investigations in humans, with particular reference to pain. In particular the procedures investigated aim to improve the sensitivity to drug effects while controlling for potential confounds to data interpretation in a pharmacological FMRI study.

It has been demonstrated previously using FMRI that if a drug reduces pain, brain areas engaged in processing pain signals will become less active. The extent of this change could give an objective measure of drug performance and could aid decision making when considering the choice to proceed with the development of a compound.

The BOLD signal is an indirect measure of brain activity that is dependent on blood properties. Changes in breathing rate and heart rate caused by pain will modify the FMRI signal compared to non-painful conditions. This causes interpretability problems for FMRI results. Similarly, certain drugs can affect the brain’s blood vessels themselves often causing increases in cerebral blood flow and/or changing the vessel’s ability to respond (vascular reactivity), thus further altering the signal. The aim of this current project is to independently measure breathing and cardiac traces along with CBF and vascular reactivity to remove these confounds from the FMRI signal. The goal is to improve the reliability of FMRI in pain-related drug studies.

Throughout this project, FMRI tasks have been optimised to imitate chronic pain studies. Pharmacologically induced increases in CBF have been emulated using mild hypercapnia (breathing a low concentration of CO2 in air). Correction methods such as breath-hold normalisation were developed and optimised to reduce the noise in the FMRI signal caused by physiological changes. The repeatability of these techniques has been demonstrated. These correction methods can be used to either increase the sensitivity to brain activity differences or to reduce the number of subjects required in a study. For example, the technique for vascular reactivity normalisation can reduce the number of subjects required by 17% while maintaining statistical power.

Research group


Research Projects

Quantifying vascular influences on neurovascular coupling with fMRI
Optimising Pharmacological FMRI for Drug Development

Undergraduate education

Bachelor’s Degree – First Class B.A. (Mod.) in Theoretical Physics
1996 – 2000: School of Mathematics and Department of Physics, Trinity College Dublin

Postgraduate education

Wellcome Trust Research Career Development Fellow
Aug 2010 – Aug 2015: CUBRIC, School of Psychology, Cardiff University
Summary: Quantifying vascular influences on neurovascular coupling with FMRI

Post-Doctoral Research
Aug 2008 – Aug 2010: CUBRIC, School of Psychology, Cardiff University
Supervisor: Prof. Richard Wise
Summary: Developing physiological noise correction and vascular reactivity normalisation routines for use in studies of pain and drug administration

Feb 2005 – Aug 2008: Section on Functional Imaging Methods, Laboratory of Brain and Cognition, NIMH, NIH.
Supervisor: Dr. Peter Bandettini (Section Chief), Dr. Leslie Ungerleider (Lab Chief)
Summary: Research topics include removal of physiological noise from FMRI data, the TE-dependence of BOLD-related physiological noise, the impact of TSNR on signal detection, TSNR as a function of resolution and tissue type, investigation of the long term (>15s) evolution of the HRF, detection of transient effects of magnetization caused by neuronal firing.
Doctoral Degree – Ph.D.
2001 – 2004: Institute of Neuroscience and Department of Psychology, Trinity College Dublin
Supervisor: Prof. Hugh Garavan                                              
Thesis Title: Methodological issues in event-related FMRI
Summary: Development of mathematical and computational algorithms to optimise human brain mapping studies utilising functional magnetic resonance imaging. Determination of suitable experimental parameters for use in event-related FMRI studies of human cognition.

Master’s Degree – M.Sc. in High Performance Computing
2000 – 2001: School of Mathematics, Trinity College Dublin
Thesis Title: Building a High Performance Cluster
Summary: The study of parallel computer systems with a heavy emphasis on mathematical techniques. Topics of examination included MPI programming languages, stochastic modelling techniques, finite difference methods and hardware considerations. The thesis project consisted of designing, building and optimising a 32-node Beowulf cluster computer.

Awards/external committees

Wellcome Trust Research Career Development Fellowship, 2010 – 2015
Wellcome Trust Flexible Travel Award, 2008
Government of Ireland Scholarship, 2001 - 2004

Editorial Board Membership

NeuroImage, Oct 2011 – Sept 2014
Frontiers in Brain Imaging Methods, May 2012 – April 2015


International Society for Magnetic Resonance in Medicine (ISMRM)
Organization for Human Brain Mapping (OHBM)
Organising committee for 2009 British Chapter ISMRM meeting

Reviewing Experience

Experimental Brain Research
Human Brain Mapping
Magnetic Resonance in Medicine
Medical Physics
Transactions on Medical Imaging

Invited Talks

November 2011: University of Vermont, USA
November 2011: NYU, USA
October 2011: Hammersmith Hospital, Imperial College, London
April 2011: BNA 2011, Harrogate, UK
March 2011: IOP, King’s College London, UK
May 2010: ISMRM 2010, Stockholm, Sweden
April 2009: Center for Functional MRI, UCSD
July 2007: FIL, University College London, UK
July 2007: CUBRIC, University of Cardiff, Wales
August 2005: SFIM, National Institutes of Health, Bethesda, USA
Dec 2003: Medical Psychology, University of Tuebingen, Germany