Functional Brain Imaging

A number of techniques are now established for non-invasive mapping of responses to sensory and cognitive stimulation of the human brain. These provide exciting opportunities for vision scientists to record the activity of brain areas involved in visual processes.

The technologies, which include functional magnetic resonance imaging (fMRI) and magnetoencephalograms (MEG), are based on different physical principles, but have in common the need to be performed within an electrically shielded environment. This means that most of the vision researcher's customary equipment should not be placed close to the imaging system: stimuli cannot be displayed on a CRT-based monitor and responses cannot be recorded using standard analogue or digital electronics.

Presenting Stimuli

Visual stimulation using computer graphics and CRT-based displays is incompatible with the sensitive environments required by functional imaging techniques.

• The display and driver electronics generate excessive quantities of radiation that interfere with the signals being collected
• The output from the CRT (a magnetically deflected system) is affected by the high magnetic fields needed for operating an MRI scanner
• Magnetic fields extend beyond the confines of the usual screened rooms and prevent the useful operation of CRTs even outside them

Using Projectors

A simple solution to the stimulus presentation problem is to use a projector: the stimulus is projected on to a screen located near the subject, typically at one end of a scanner. The subject, who is normally recumbent and therefore looking up, can view the screen with the help of a mirror mounted on the head coil. The projector can be used either inside or outside the shielded room. An ideal solution places it inline with a back-projection screen to alleviate spatial distortion (the keystone correction electronically implemented by the projector introduces other non-desirable artefacts).

There are two types of portable projector technology: those that use liquid crystal panels as the light modulating elements and those that use 'digital' micro mirrors. Both technologies rely on electrical rather than magnetic phenomena for their operation. Micro mirror-based portable projectors generate colour using sequential frames and a colour wheel rather than a set of fixed dichroic filters. This solution can create synchronisation problems so currently we suggest using LCD-based projectors.

LCD projectors

Modern LCD projectors contain a single high-intensity discharge lamp and three LCD spatial light modulators. The light is split into three coloured components, modulated by the LCD panels and then recombined optically before being launched via the projection lens onto the screen. The light source is essentially a well-controlled static device that adds no temporal components to the image. However, regular calibration of the light source is necessary to accommodate changes in luminance and colour temperature over time.

From a vision science perspective, the LCD spatial light modulators and their controlling circuitry provide the interesting properties of the device.

• Each LCD panel has an intrinsic resolution; the surface comprises a fixed array of elements with a defined spacing. Current resolutions range from 0.5 to 0.75 million pixels and each of the pixels can be updated at about 60 Hz.
• The internal electronics digitise the analogue signal, then compress and interpolate the data spatially and temporally to produce a data stream that it can display. If the frame rate of the input signal is too high, the projector will drop frames to provide the correct average rate. This causes problems for dynamic stimuli, particularly if the objects are moving in space.
• It is important to run a projector in its native mode, that is at a spatial resolution equal to the native resolution and at a frame rate within its direct capability.
• The intervention of the signal processing circuits in the video chain imposes a substantial delay between the arrival of the video input and the appearance of the light on the screen. This can be as much as 15 ms and makes the use of projectors in image stabilisation schemes more difficult.
• If it is necessary to know exactly when the stimulus onset occurs then the delay time must either be characterised (which is easy to do provided the display is locked to the source) or a small photocell system devised to feedback the light output directly into a data acquisition system.

The diagram above shows how the light output from an area on a projector screen near the centre varies with time when the patch is illuminated for two frames before being turned off. This can be compared to the light from a CRT when driven by the same video signal. The additional time delay imposed by the projector's electronics is also clearly visible.

¹ based on Epson EMP500

Temporal Synchronisation & Triggering

Projectors are compatible with many video sources but this does not mean that all combinations are suitable for vision research. For maximum display accuracy and ease of control we advise driving projectors with a Visual Stimulus Generator like the ViSaGe. Unlike other sources of video data, the ViSaGe provides accurate timing control, automatic gamma correction and experimental versatility.

If you are using an additional data acquisition system for recording eye movements or evoked responses, it is easy to take a digital trigger signal generated by the VSG and record this on a spare channel to provide a time marker synchronised to the stimulus onset. The recorded data can subsequently be correlated offline with scanner images after the experiment.

In all vision science applications it is important to ensure that the stimulus presented by the display device really has the required profile and also remains the same from trial to trial. Projector bulbs have finite lifetimes (about 2000 hours) and during this period it is reasonable to expect significant changes in colour and luminance. Cambridge Research Systems offers several calibration tools: the OptiCAL photometer, the ColorCAL colorimeter and the SpectroCAL vision science meter. Both solutions are fully integrated with our Visual Stimulus Generators and can be used to automatically generate the required lookup tables and colour coordinates. Using either of these devices is fast and makes display calibration simple.

Generating Stimuli

Visual stimuli can be generated by writing a computer program in a low-level language and then presented using a standard graphics card. While this approach is very powerful, it does require an intimate understanding of graphics technology to synchronise the display.

The VSG Software Library for the ViSaGe offers a much easier approach and includes a suite of stimulus demonstrations specifically designed for functional imaging applications. These cover many of the common paradigms, including stimuli for retinotopic mapping. All aspects of these stimuli can be defined from simple on screen controls, so often no further programming will be needed. If more complex stimuli are required, full source code for the demonstration programs is provided to illustrate how to use the VSG Software Library and transparently handle procedures like:

• Driving the projector in its native mode
• Activating colour and luminance calibration
• Synchronising the stimulus and display
• Generating trigger and marker information

Example Visual Stimuli

Retinotopic Mapping - this stimulus enables mapping of both angle and eccentricity. Used with cortical flattening software (for example: Dale AM, Fiscl B, Serano MI (1999) Cortical surface-based analysis. NeuroImage 9:179-194), the cortical representation of retinotopic space can be determined. This enables visual areas responding to specific stimuli to be located (Engals SA, Glover GH, Wandell BA (1997) Retinotopic organisation in Human Visual Cortex and the Spatial Precision of Functional MRI. Cerabral Cortex 7:181-192). The VSG Software Library includes a stimulus demonstration, which shows how to implement retinotopic mapping using the ViSaGe.

Motion - both first- and second-order motion stimuli are available. Choose from radial or linear luminance encoded motion. Second order-motion stimuli are encoded purely by contrast or colour changes in a field of dynamic random noise (Smith AT, et al (1998) The Processing of First- and Second-Order motion in human visual cortex assessed by functional magnetic resonance imaging. J of Neuroscience 18(10): 3816-3830).

Colour & Contrast - isoluminant grating, chequerboard and windmill stimuli.

Picture Display - the ViSaGe can be used to display photographic images as well as mathematical stimuli. Common graphics file formats like Windows bitmap (BMP) can easily be imported. The VSG Software Library includes a dedicated image file demonstration utility. This uses an optimised memory buffer to ensure accurate timing and provides an external trigger facility to synchronise data collection apparatus.

Recording Eye Movements

In fMRI, the distribution of BOLD (blood oxygen level dependent) responses depend on the type of eye movement performed, so to reliably classify brain areas it is important to carefully measure the actual eye movements. Eye movements are either spontaneous in nature or they are visually guided. The latter fall into two general classes: saccadic eye movements which are ballistic and direct the fovea to a target or location of interest, and pursuit eye movements which are smooth continuous tracking movements, when the eyes follow a moving target.

MR-Eyetracker

Some conventional eye-trackers are not fast enough to distinguish different types of movement and most incorporate ferrous materials and electronics that are incompatible with magnetic environments. The MR-Eyetracker is a contact-free solution that is easy to operate and is designed specifically for use in fMRI: it uses fibre optics and has high spatial and temporal resolution easily capable of discriminating saccades and pursuit.

Measurement Technique

Fibre optic cables guide infra-red light from a source located outside the scanner into the head coil and onto the eye of the subject. The light is reflected by the eye along the edge of the iris and is transferred back to infra-red sensitive diodes (limbus reflection). These photodiodes are coupled to an amplifier, which provides independent estimates of the horizontal and vertical positions in one eye. The system works with a daylight suppression technique, to prohibit interference from ambient light.

The fibre optic cables are encased in black malleable plastic for easy handling and their long length prevents the MR-Eyetracker Control Unit from interfering with the magnetic field. The ends of the cables are mounted into a custom eyepiece that is attached directly to the head coil. Just a few centimetres of additional space are required to mount the eyepiece, since it is designed to be lowered into the window of the head coil.

A mirror and mounting plate are also available for the head coil so that the subject can view a stimulus projected on to a screen located near the scanner.

Calibrating the Eye Position

Calibration of the MR-Eyetracker is a simple process that takes less than five minutes.

• The subject looks at a fixation point on the screen while the experimenter centres a red/green bar displayed at the bottom of the pupil and adjusts the zero-offset
• Subsequent measurements are made with a stimulus displayed at known horizontal and vertical angles

The fixation and stimulus can be generated with a visual stimulus generator like the ViSaGe and the vsgEyetrace software, or any other computer graphics system.

Capturing and Analysing Data

A two-channel data acquisition system is required to capture the separate horizontal and vertical eye position signals. An application such as National Instrument's LabVIEW, or a user-provided program can be used to display and process the signal. Alternatively, the vsgEyetrace software for the ViSaGe provides an integrated data acquisition and visual stimulator solution for common ocular-motor tasks. The scanner should be programmed to provide a TTL-compatible pulse at the beginning of each volume acquisition to synchronize both the stimulation and the acquisition programs. This is required to later classify the brain areas based on the processed eye position signals.

Example Data

In this example, the subject fixated at a point (FP) in the centre of the display (position 0/0). After one second the FP disappeared and a target (T) was turned on at position x/y, for 1.5 s duration. The subject performed a saccade from FP to T.

The stimulus was generated using a ViSaGe and back projected using an LCD projector onto a transluminent screen.

MR-Eyetracker Specification

Further Information

The MR-Eyetracker was developed by H.Kimmig, M.W. Greenlee, F.Huethe and T.Mergner at the Department of Neurology, University of Freiburg, Germany.

An earlier version of the system is described in: Kimmig, H., Greenlee, M. W., Huethe, F., & Mergner, T. (1999). MR-Eyetracker: A new method for eye movement recording in functional magnetic resonance imaging (fMRI). Experimental Brain Research, 126; 443-449.

Useful Information

• Functional Magnetic Resonance Imaging - Mark S. Cohen

Copyright © Cambridge Research Systems Ltd. Click for details.

^ Back to top