skip to content

Mayneord Phillips Summer Schools

 

2001 Summer School - Physics and the Brain - Abstracts

  MONDAY 9 JULY:INTRODUCTION, MEASURING, MODELLING

Mapping the Mind Rita Carter

Until recently attempts to pin down the physical basis of mind were hampered by the fact that brain functions - unlike those of many other organs are not visible to the naked eye. The only way scientists could identify the brain areas associated with perceptions, emotions, thoughts and memories, was by observing people with distinct mental dysfunctions and matching these to damaged tissue. This usually meant waiting until the afflicted individual was dead. Hence, although brain "maps" had been drawn up for millennia, the first accurate siting of functional regions (Broca's and Wernicke's language areas) was not achieved until the 20th century .

Today neuroscientists can look into a living, working brain, and observe precisely what is happening, where, when a person feels, thinks or does a particular thing. This insight is provided by imaging techniques such as PET, fMRI and MEG. It has already made it possible to create a map of the mind which is so subtle and detailed that even the most abstract and previously "mysterious" human qualities - the sense of self; morality, religion - have been pinned down to the workings of particular neuronal groups.

The implications of this cartographical breakthrough are immense, because once the precise physical location of a mental function is determined, it is possible, in principle, to manipulate it. Indeed, the technology for doing this is already available and in use for treating conditions like depression and obsessive-compulsive disorder. In future it could be used to mould human minds in any way we please....

EEG/EP Colin Barber (Nottingham)

The electroencephalogram was, in the words of Grey Walter, the original "window on the brain" and has been a source of functional information for the best part of a century. It is, though, somewhat of a frosted glass window since the low conductivity of the skull "smears" the electric field representation and makes source localisation very difficult.

Nonetheless it provides useful localisation in, for example, diagnosis of epilepsy and it has exquisite time resolution, permitting us to plot brain activity on a millisecond scale. Further, use of the simple technique of synchronous averaging allows time-locked activity (evoked potentials, EPs) to be extracted and this has proven to be a powerful tool in the study of sensory processing. By suitable experimental design, potentials relating to cognitive processing (event-related potentials, ERPs) can be examined.

This presentation will review the contribution of EEG and EPs to the understanding of brain function and discuss, with example from vision and hearing, how modern signal processing techniques can extend their application.

PET/SPECT Tim Fryer (Cambridge)

Radioactive tracers that result in photon emission can be used to produce three-dimensional images of brain physiology, biochemistry and pharmacology. In single photon emission tomography (SPECT) the tracer emits single photons, whereas positron emission tomography (PET) utilises positron-emitting tracers that subsequently result in photon emission from electron-positron annihilation. The fundamental principles of radioactive tracer imaging will be given, including tracer kinetics, gamma ray detection and tomographic image reconstruction. The imaging characteristics and clinical uses of the two techniques will then be discussed.

Modelling: introduction Roger Orpwood (Bath)

The following topics will be covered.

The importance and role of modelling brain activity for our understanding of brain function, and for exploring what goes wrong in illness.

Modelling of synaptic communication and the role of receptors.

Modelling of other cell membrane ion channels.

Compartmental modelling of whole neurons, including the involvement of plastic receptors and of active conduction.

Local network modelling based on understanding of local connectivity in the cerebral cortex.

Modelling hierarchies of cerebral networks and its implication for higher brain functions.

This presentation will give a broad overview of the various levels of brain modelling that are possible.

back to page top

  TUESDAY 10 JULY: MEASURING, INTEGRATING, MODELLING

MEG Matti Hämäläinen (Helsinki)

Magnetoencephalography (MEG) is a unique functional imaging tool providing an uncompromised time resolution combined with a reasonably good spatial discrimination of neural current sources. The method is totally non-invasive and the studies can be done either in seated or supine positions. Furthermore, the equipment does not produce any acoustical noise, which makes the method particularly appealing for auditory studies.

This presentation is a review of various technical and methodological aspects of MEG. The talk comprises a brief discussion of the generation of cerebral magnetic fields, an overview of the construction and technical features of a modern MEG/EEG system, followed by an evaluation of available modelling techniques.

Several examples of actual MEG studies are included to illustrate the capabilities of the method in the study of the dynamics of the human brain.

MRI (fMRI) Gareth Barker (London)

The basics of nuclear magnetic resonance (NMR), including the concept of resonance and the Larmor equation will be introduced. A brief overview will be given of the spin warp MR imaging technique will be given, including the concepts of slice selection, frequency encoding and phase encoding. Examples will be shown of conventional, structural, MR images and the concepts of proton density, and T1 and T2 relaxation times, and their effect on image contrast will be introduced. Fast imaging techniques, in particular echo planar imaging (EPI), will then be discussed.

One of the most fascinating structural imaging techniques is diffusion tensor imaging (DTI), which has the potential to demonstrate, in vivo, the position and orientation of white matter tracts. The use of magnetic field gradients to create MR images whose intensity and contrast depend on the diffusion of water within and around tissues will be explained, and the concepts of isotropic and non-isotropic diffusion will be introduced. Measurement of the full diffusion tensor will be described, and the parameters that can be derived from this, including the mean diffusivity (representing the magnitude of water diffusion) and various anisotropy measures (representing the ïdirectionalityÍ of diffusion) will be defined. Examples of images of these parameters will be shown, and the problem of displaying and interpreting the large amount of data created will be discussed. Finally, the use of ïtractographyÍ to map putative connections between brain regions will be demonstrated.

While DTI has the potential to reveal structural information about brain tissue, functional MRI (fMRI) has, as its name suggests, the ability to details of brain function. The concept of BOLD (Blood Oxygen Level Dependent) contrast, on which most current fMRI techniques depend will be introduced, and the haemodynamic response, which links this MR contrast change to underlying physiological changes, will be discussed. Different experimental paradigms (eg block designs, event related designs) will be described and the processing steps required to produce the final functional image (including factors such as the choice between parametric and non-parametric statistics) will be discussed. Examples will be shown of simple fMRI experiments, and a number of clinical applications will be described.

Integrating Harm-Jan Wieringa (Hannover)

We will have a look at the reasons for integrating different kinds of information (different modalities), and in particular, but not exclusive, the use of MRI with EEG and MEG. We will look at obtaining the necessary data, the (pre-) processing of that data and the techniques for integration and also for the visualization of the results. An important aspect to consider is the errors that we make in combining different techniques, so that we can be careful not to suggest an accuracy that is not there. This will also show the limitations for future improvements.

Modelling Vision Hanspeter Mallot (Tübingen)

What does it mean, to see? was the opening question of David Marr's famous book on vision published almost 20 years ago. In his answer he stated that vision is an information-processing task taking images as its input and producing as its output a description, or representation of the world. Thus, vision inverts optics in that it leads from the images back to (a description of) the world. This view is still largely valid, even though the perspective has been broadend during the last decade by asking what exactly is represented and for what purpose.

In the course, I will discuss biological vision as an information-processing task. The first part will focus on the basics of image generation and image processing, i.e. the process that is to be inverted in "inverse optics" and the tools are can be used to do so. The second part will present an example from early vision, i.e. the perception of depth from stereoscopic images. In the final part, I will then proceed to the recognition of three-dimensional, a field where David Marr's representational approach has been challenged by recent psychophysical findings.

back to page top

  WEDNESDAY 11 JULY: MIMICKING

Cochlear Implantation Steve Mason (Nottingham)

Cochlear implantation is an innovative and effective form of management of profoundly deafened patients who obtain little or no benefit from conventional amplification using acoustical hearing aids. Electrical stimulation of the peripheral auditory nerve by-passes the damaged hair cells in the cochlea and provides access to speech sounds and the acoustical environment. In recent years the number of very young children receiving cochlear implants has increased significantly as the efficacy and benefits of early implantation become apparent. Children younger than 5 years of age demonstrate a remarkable ability to adapt to this form of stimulation due to plasticity in the auditory system.

We will explore the historical development of the cochear implant from the time in 1957 when Djourno and Eyries first reported a device that could directly stimulate the auditory nerve, through to modern day devices and potential developments in the future. The effects of electrical stimulation on the auditory system can be studied extensively using electrophysiological techniques. Early reports of these electrically evoked potentials were presented in guinea pigs by Meikle in 1977 and in humans by Starr and Brackmann,1979. Modern day practice enables us to record from many different levels of the auditory pathway, similar to those associated with conventional acoustical stimulation: electric response audiometry (ERA).

At the level of the inner ear we can observe the compound auditory nerve action potential (ECAP) as it leaves the cochlea using a technique known as Neural Response Telemetry (NRT). Central perception and cognition can be accessed objectively using the cortical response and event related potentials. We will explore the tonotopic organisation and functioning of the cochlea and how it influences the coding strategies of the electrical stimuli adopted by the implant system. Evoked potentials will be presented that track the transmission of this information from the cochlea, through the brainstem pathways and its subsequent processing in the brain.

Retinal Implantation Eberhart Zrenner (Tübingen)

There are presently several concepts to restore vision in blind or visually impaired persons by implanting electronic devices into the eye or the visual cortex in order to evoke useful visual sensations.

Sub-retinal micro-photodiode arrays have a number of advantages: degenerated photoreceptors are replaced by MPDAs; the remaining neuronal network of the retina can be utilized for signal processing; positioning and fixation of the MPDAs in the sub-retinal space is relatively easy; no external camera and external image processing is required; eye movements can be used for localization of objects. A sub-retinal prosthesis requires functioning optics and a preserved optic nerve. Targeted diseases are retinitis pigmentosa, and in a later state possibly also age-related macular degeneration (AMD).

Since 1995 our consortium has produced several prototypes of sub-retinal silicon chips, consisting of hundreds to thousands of micro-photodiodes with an active area of 20µmx20µm to 200µm x 200µm, equipped with microelectrodes (gold or titanium nitride) in a monopolar or bipolar fashion arranged in arrays, round or square with several millimetres in diameter and 50µm thickness.

In vitro experiments with chicken and RCS rat retinae in a sandwich technique, in which recordings are made by means of multielectrode arrays either from the inner or the outer retina has revealed:

charge injections of about 0,4nC per electrode are sufficient to excite post-receptoral retinal neurons;

electrode distances of 50-150µm in the outer retina can be resolved in ganglion cell recordings;

retinae with completely degenerated photoreceptors (RCS rats, 160 days and older) can still be excited by subretinal electrodes in a proper spatially organized manner;

surface coating of MPDAs as e.g. with laminins improves cell adhesion and biocompatibility.

In vivo experiments have revealed:

inner retinal layers are well preserved in the central retina (as shown by comparative histological studies of human and animal forms of degenerative retinal disorders) even in patients with longstanding retinitis pigmentosa.

Two surgical approaches for safe introduction of the devices have been developed: ab interno: via the classical transvitreal acces to the retina, and ab externo: via a scleral flap near the lombus through the subretinal space (like in a tunnel) to the back of the eye.

inner retinal layers are well preserved after long term implantation of subretinal MPDAs in pigs (up to 16 months)

MPDAs remain fixated at stable subretinal positions and are well functioning up to 16 months as revealed by multifocal electroretinograms in both, rabbit and pigs.

MPDAs showed some damage of the silicon oxide surface of the implant

spatially sensitive electrically evoked cortical potentials can be recorded with mult-ielectrode and optical recording from the visual cortex of rabbit and pig following acute electrical subretinal stimulation via electrode foil strips.

Although stimulation characteristics, spatial and temporal properties of subretinal stimulation by microphotodiode arrays and biocompatibility are now well investigated and the possibility of eliciting cortical potentials transmitted via subretinal electrical stimulation devices has shown the feasibility of the subretinal approach, still a number of questions remain to be solved. How well is spatial resolution and feature localization maintained at the level of the visual cortex? How can long-term stability of silicon chips, the surface of which is affected after long-time implantation be achieved? Will retinal neurons tolerate a long-term electrical stimulation without morphological and/or functional alteration? What type of image can be perceived by blind patients through subretinal MPDAs?

Utilization of subretinal MPDAs in blind patients can therefore take place only after answers to these questions are available by further experiments in animals.

back to page top

  THURSDAY 12 JULY: BRIDGING BORDERS

PET/other Tim Fryer (Cambridge)

This presentation will be split into two main sections: PET and other imaging modalities (MRI,Computed Tomography CT, fMRI, SPECT), and PET with non-imaging techniques (EEG, cerebral probes, trans-cranial Doppler ultrasound, near infra-red spectroscopy, intra-cranial pressure monitors). PET images lack anatomical detail and the process of fusing the functional information provided by PET with anatomical images from MR or CT aids visual diagnosis and spatial normalisation, and facilitates the use of anatomical region-of-interest analysis and partial volume correction of PET images. Information from PET can also be combined with other functional images from fMRI and SPECT. The three-dimensional information provided by PET has also been used to verify focal or global monitors of brain function. This role is particularly important in the field of brain trauma (head injury, haemorrhage, stroke) where PET has been used in combination with EEG, cerebral probes, trans-cranial Doppler ultrasound, near infra-red spectroscopy and intra-cranial pressure monitors.

EEG/MEG Matti Hämäläinen (Helsinki), Colin Barber (Nottingham)

As we all know fro our school-day physics, electric and magnetic fields from a current-carrying conductor are orthogonal. This opens the possibility of their combined use to help overcome the severe problem of interpreting the field complexity introduced by the multiple folding of the cortex, and its great variation from one person to another.

This presentation will comprise two parts. In the first (HM) the fundamental theoretical differences, and their implications will be discussed. In the second (CB) one attempt to use their complementarities in the field of vision will be described.

fMRI/EEG Jack Belliveau (Boston)

To follow

Modelling/MimickingRobotics Hanspeter Mallot (Tübingen)

The purpose that brains have evolved for is the generation and control of behaviour. Therefore, successful theories of brain function should translate into control principles for autonomous agents or robots. The synthesis of a behavioural competence on an autonomous robot is a rigorous test for any theory of brain function.

In my talk, I will present an overview of mechanisms of animal navigation and their implementation into autonomous robots. Navigation mechanisms can be classified according to the memory demands of each mechanism. In path integration, a powerful mechanism found in many animals, the vector to the starting point is stored and continuously updated without memory of the path travelled. Landmark navigation involves local position information stored in long-term memory. Landmarks may mark the goal (beacons), they may be guidance with respect to a goal, or sub-goals allowing the construction of routes. Finding ways in structured environments (such as indoor configurations of rooms and corridors) requires storage of the spatial relations of many landmark points. Such a memory allowing the generation of routes is often called a cognitive map. As one example of how to implement cognitive maps into autonomous robots, the view-graph approach will be discussed.

back to page top

About Us | Site Map | Contact Us | Copyright © 2008 The Mayneord Phillips Trust