MR-Eyetracker FAQ

1. How does it work?
2. What is the limbus?
3. How can I integrate the MR-Eyetracker with my existing stimulator and data collection equipment?
4. What is the sampling rate?
5. What about software?
6. How do I stabilize head position?
7. Why do I need to stabilize the head?
8. Can I have longer fibre optic cables?
9. Where should I site the Control Unit?
10. Will it work with non-humans?
11. Will CRS demonstrate the MR-Eyetracker system in our scanner?
12. Can CRS provide training/installation?
13. Is the output linear?
14. Can I measure vertical movements simultaneously to horizontal?
15. CRS quote resolution better than 0.25°. What can I expect to get?
16. Where is the camera mounted?
17. Can I monitor fixation/direction of gaze/reading?
18. You say a lot about what I can't do with it, what use is it?
19. Will it interfere with my MRI/MEG machine?
20. What effect does the MR-Eyetracker have on signal to noise ratio?
21. Do you have k-space data?
22. What scanners can it be used with?
23. I have an xyz scanner that is not on your list of mounts. Can you make me a custom mount?
24. Can it be used outside of the scanner?
25. Why does the mount cost extra?

1. How does it work?

The MR-Eyetracker converts eye position into an analog voltage. It uses the limbus tracking technique. Limbus trackers record eye movements by measuring the differential reflectance between the boundary of the sclera and the iris viewed through the cornea.

In a conventional limbus tracker, the eye is illuminated by infrared light, usually coming from a light emitting diode (LED). The amount of light reflected back from the eye's surface is measured with a photodiode and this gives the position of the eye. As the cornea and iris reflect much less infrared light than the sclera does, the intensity of the reflected light is proportional to the relative amounts of sclera and cornea within the acceptance angle of the detector and hence eye position.

Rather than mounting the photo-diodes and LEDs directly at the eye, the MR-Eyetracker uses fibre optic cables to relay the infrared light to and from the eye. This enables the electronics to be remotely located, minimising tracker and scanner interaction. See reference 1 for a review of eye tracking techniques.

This is a simplified block diagram of the circuitry. The fibres in each fibre optic bundle are split into two. One half is coupled to an LED for transmission of the infrared light, the other half to a photodiode for capture of the infrared light reflected back from the eye. At the eyepiece, fibres from both halves are uniformly interleaved. The two LEDs are switched on and off (chopped) at 10 kHz. The signal current resulting from the reflected light is converted to a voltage and buffered. The sum and difference of the signals from each side of the eye is derived. The difference is proportional to horizontal eye position and sum to vertical position. Subtracting the signal when illumination is off from the signal when illumination is on demodulates the chopped signals. This compensates for ambient infrared illumination.

2. What is the limbus?

The limbus is the margin between the cornea and the sclera.

3. How can I integrate the MR-Eyetracker with my existing stimulator and data collection equipment?

The MR-Eyetracker is an analog output eye tracker. As such it can be connected to any suitable data acquisition system. The output voltage range is ± 5 V full-scale and will interface to many readily available data acquisition systems.

4. What is the sampling rate?

The MR-Eyetracker is an analog device and produces a continuous voltage dependant upon eye position. Internally, the dark compensation circuitry is chopped and samples the illuminated and dark phases at 10 kHz. However this is not discernable in analog output. The Control Unit contains filters with a bandwidth of 500 Hz. This is well matched to a Nyquist sampling frequency of 1 k samples/second, giving millisecond temporal resolution. However it should be noted that even very fast saccades do not contain frequencies greater that 110 Hz and that this resolution is convenient rather than necessary.

5. What about software?

As noted in FAQ 1, the MR-Eyetracker is an analog output device for connection to a data acquisition device. Any of the popular data acquisition and display packages can be used with it, e.g. LabView. However you will have to develop some specific routines for calibration if measurements in actual visual angle are required.

Cambridge Research Systems supplies the vsgEyetrace program for those customers using the MR-Eyetracker with the ViSaGe. vsgEyetrace uses the ViSaGe for both stimulation and data acquisition and it has the added advantage that the functions are tightly synchronized. Stimuli are provided for most common oculomotor tasks and the data is displayed in real time as scrolling graph and can be saved to disk. Functions for calibration to visual angle and MR-Eyetracker set-up are included. This software costs £300 (US$450) and can be downloaded from our web site (you will need to contact our Sales team for an unlock code). The program was developed in the popular Borland Delphi language and full source is provided. It is designed to form the basis of your own dedicated software routines and to allow measurements to be made 'straight out of the box'.

6. How do I stabilize head position?

The usual head clamping arrangement found on many head coils does not usually provide adequate stabilization. Two stabilization methods are common; the vacuum head cap and the bite bar.

Several manufacturers produce bite bars for use in the magnet and some groups have developed their own, though there are considerable practical problems in their use. Subjects tend to dislike the procedure, and compliance is often considerably reduced. The bite bar is seen as more invasive than the scanner itself. There are also safety issues. The subject's range of head movement is considerably limited by the headcoil and it is difficult for the subject to voluntarily release the bite. Aside from the feeling of claustrophobia there is a risk of gagging and choking. There is also the attendant risk of injury to the jaw and teeth from a rigidly mounted bite so some means of automatic release when extreme forces are applied should be provided. The choice of impression material should also be considered. The usual hard impression wax used for bite bars in the lab, should be avoided. This is extremely brittle and tends to easily chip. In the lab this is no problem but the supine position within the scanner poses a severe choking risk.

The vacuum head cap is a good compromise. It consists of a two-layered hood containing many soft polystyrene beads. The hood is placed around the subjects head and the beads distributed to form an even layer around the head. The air within the hood is evacuated with a simple hand pump. The beads collapse on themselves forming a rigid mould of the subject's head. Although this does not complete stop the subject from moving their head, there is only one position in which they are truly comfortable, which they tend to return to. We have found the Vacufix vacuum head hoods from B.u.W. Schmidt to be particularly good. These can be supplied as an optional accessory.

7. Why do I need to stabilize the head?

As the MR-Eyetracker is usually mounted onto the MRI head coil, rather than directly onto the subject's head, head movements relative to the sensor are possible. The MR-Eyetracker cannot distinguish movement of the eye due to head translation and rotation from that due solely to eye rotation. One millimeter of head movement is roughly equivalent to 5° of eye rotation. Consequently for accurate eye position measurements it is necessary to stabilize the head with respect to the MR-Eyetracker. For some applications small amounts of head movements can be tolerated. Signal deviations due to head movement can easily be distinguished from those due to eye movements from the velocity of the position shift. Head movements usually result in an overall DC shift in the recorded eye position. As the MR-Eyetracker is linear over its measurements range, for some types of experiments (e.g. saccadic gain) this can be cancelled by making measurements of relative eye position during experimental trials.

8. Can I have longer fibre optic cables?

As standard, 9 metres fibres are supplied. These are usually long enough for most installations. However it is possible to supply longer fibres, at least 15 metres, but custom manufactured fibres are considerably more expensive than those supplied as standard.

9. Where should I site the Control Unit?

Ideally the MR-Eyetracker Control Unit should be mounted outside of the Faraday room. The fibres can usually be conducted through a waveguide filter. Most installations usually have a suitable waveguide in the penetration panel. Some scientists prefer to mount the Control Unit in the scanner room itself. This is usually for reasons of convenience. Whilst aligning the MR-Eyetracker sensor, it is necessary to instruct the subject to make appropriate eye movements. It is much easier to give clear instructions to the subject and monitor compliance whilst in the same room. If the Control Unit is sited in the scanner room careful precautions must be taken. For safety, the Control Unit power supply must not be taken into the vicinity of the scanner as it contains ferrous parts. To avoid scan artefacts careful consideration must be given to electrical connection to the control box (see FAQ 19 for more details).

10. Will it work with non-humans?

We believe that it should be possible to use the MR-Eyetracker with any animal that has a clearly defined limbus and a suitably large eye. However to date we have no customers actually doing this. If you are interested in a non-human application, it may be possible to arrange a demonstration at your laboratory.

11. Will CRS demonstrate the MR-Eyetracker system in our scanner?

For standard scanner and head coils combinations it is possible to demonstrate the MR-Eyetracker at your facility using the Eyetrace software (please check FAQ 22 for a list of supported MRI scanners). However we will only do this when you are making an immediate fiscal decision. Alternatively, it is possible for us to arrange demonstrations in existing customer's laboratories.

12. Can CRS provide training/installation?

Yes, it is possible for one of our staff scientists or engineers to assist you in getting the MR-Eyetracker working with your equipment and to provide basic training in its operation. Pricing for this service starts from £1000 (US$1500). Please request a quotation by contacting our Sales team.

13. Is the output linear?

The MR-Eyetracker is a limbus tracker (see FAQ 1). It has all the advantages and disadvantages of this measurement technique. For eye movements on the cardinal axes (purely horizontal or purely vertical) the output voltage is a linear function of eye rotation. The output can be simply calibrated by making saccades to known locations and using linear regression of output voltage to target location.

For oblique eye movement with both a horizontal and vertical component, the transfer function is not simple. Firstly, as eye movements are measured on one cardinal axes a change is measured on the other axes. This crosstalk results in the two axes not being independent and is non linear due to the geometry of the measurement technique. For relatively small movements, < 10°, it is possible to linearise using a number of well established mathematical techniques. However for larger oblique eye movements measurements are confounded by the outputs being potentially ambiguous. That is to say, the same voltages can appear at the outputs for two different eye combinations of horizontal and vertical eye position.

14. Can I measure vertical movements simultaneously to horizontal?

Yes it is possible to simultaneous measure from horizontal and vertical movements simultaneously. Separate analog outputs are provided for this purpose. However there are limitations to the measurements that can be made (see FAQ 13 & FAQ 15).

15. CRS quote resolution better than 0.25°. What can I expect to get?

The resolution is dependant on having a large optical signal. This in turn is dependant on good alignment with the eye. If care is taken it is possible to record micro-saccades in the order of 10'. However this usually requires re-adjusting the sensor location whilst monitoring the output of the MR-Eyetracker. When only the visible alignment illumination is used, resolution of 0.25° is easily achievable in a short set-up time.

16. Where is the camera mounted?

The MR-Eyetracker does not use a video camera. See FAQ 1.

17. Can I monitor fixation/direction of gaze/reading?

As explained in FAQ 13, the MR-Eyetracker is non-linear when measuring oblique eye movements. Practical measurements are limited to within the central ± 10° field and even then oblique measurements must be linearised. Therefore it is not the eye tracking system of choice for visual scanning, reading or any type of measurement where horizontal and vertical components exist simultaneously.

18. You say a lot about what I can't do with it, what use is it?

The MR-Eyetracker was developed in conjunction with Dr Hubert Kimmig and Professor Mark Greenlee at the Department of Neurology, University of Freiburg. It was designed to meet the needs of oculomotor scientists that are not met by a video eye tracker. Although the video technique has superior linearity over the oculomotor range, limbus tracking has a number of advantages:

1. High sensitivity and low noise
2. Good DC stability
3. Wide bandwidth enabling accurate temporal and dynamic characterization of eye movements (e.g. saccade metric measurements)
4. Short delay between eye movement and measurement, which enables gaze contingent displays to be implemented

Scientists familiar with oculomotor research methodology appreciate the advantages of the technique and it is commonly employed outside of the scanner environment. Experimental protocols can be designed to minimize or overcome the linearity limitations. Often purely horizontal eye movements are adequate.

19. Will it interfere with my MRI/MEG machine?

The preferred location for the MR-Eyetracker Control Unit is outside of the scan room (see FAQ 9). In this configuration there is no possibility of imaging artefacts arising from the using the MR-Eyetracker. The headcoil mount is manufactured from non-metallic materials, which have been fully tested in the MRI environment. No radio frequency interference ('zippers' in the image) or susceptibility artefacts occur.

If circumstances dictate that the MR-Eyetracker is used inside the faraday room then interference free operation is possible if careful consideration is paid to the installation. The Control Unit is extremely well screened and has been fully tested for radio frequency emission and pickup. There are a number of working installations with this arrangement in regular use with no problems from associated artefacts.

A full discussion of radio frequency interference is beyond the scope of this FAQ but we will touch on the important points. The chief source of interference is not the Control Unit, but the interconnecting cables. A length of wire passed through a waveguide filter, not connected to anything, will introduce significant noise into the image. The length of cable between the penetration panel and the Control Unit should be as short as possible. There should be no cables which are not connected at both ends, e.g. the vertical channel output. Ferrite chokes should be used on all cables to filter high frequency noise. These are readily available from most electronic suppliers.

20. What effect does the MR-Eyetracker have on signal to noise ratio?

When properly installed the MR-Eyetracker should cause no noticeable decrease in signal to noise ratio.

21. Do you have k-space data?

We can provide k-space data measured from a water phantom with the MR-Eyetracker in place on the head coil and operating.

22. What scanners can it be used with?

The MR-Eyetracker is compatible with virtually all scanners that have open-face head coils. We can supply mounting systems off-the-shelf for:

Siemens

• Vision
• Sonata
• Symphony
• Allegra

GE

• Signa 1.5T
• Signa 3T

Philips

• Gyroscan

23. I have an xyz scanner that is not on your list of mounts. Can you make me a custom mount?

CRS can provide custom engineered solutions for non-standard scanners. To produce such a mount we require accurate measurement of the scanner, preferably manufacturers' mechanical drawings, and access to the scanner for development purposes. If the scanner is extremely unusual or a custom design, an extra charge will be made for development costs.

24. Can it be used outside of the scanner?

Yes, the MR-Eyetracker can be used as effectively outside the scanner as in, provided that proper head stabilization is used (see FAQ 6 & FAQ 7). However, due to its unique optical coupling, it is considerably more expensive than conventional limbus trackers, which will give similar performance.

25. Why does the mount cost extra?

The cost of head coil mounts reflect the complexity of design and cost of manufacture for a specific scanner. Also, where we do not have an off the shelf mounting solution, some customers prefer to purchase the MR-Eyetracker without a mount and develop their own.

References

1. Eye Tracking Techniques

Collewijn, H, Eye Movement Recording, in Carpenter R.H.S. and Robson J.G. ( Ed ), 1999, Vision research; A Practical Guide to Laboratory Methods, Oxford University Press, 245-285.

2. MR-Eyetracker History

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

^ Back to top