Stimulus presentation details
In general, we are trying to make the Stimulus Delivery as similar as possible to the MRI (see http://imaging.mrc-cbu.cam.ac.uk/imaging/CbuStimulusDelivery). The actual hardware used (computer, soundcard, videocard etc) is identical, as is almost all software. However, there are important differences, which we will expand here.
Computer & Software
The stimulus computer is a Dell PC, running Windows 7. The machine is not networked, so you need to copy your data on and off it using a USB memory stick. This can be tricky because of security settings that require all external data sources to be scanned for viruses before use. Use a memory stick with as little on it as possible if you can, otherwise the scanning might take very long.
Stimulus presentation software is either E-Prime 2, Matlab or Presentation. Input and output are all through the parallel port for E-Prime, and through a National Instruments PCI6503 Card for Matlab. Presentation can use both input/output methods.
Projector and screen
The visual presentation is done with a Panasonic PT-D7700 DLP projector outside of the MSR, projecting though a periscope into the MSR onto a back-projection screen that is placed in front of the chair. There's a pair of markers on the floor where the screen should be positioned for correct focus. The matt side of the screen should be facing the projector, as this serves to minimise reflections.
The projected image is approximately 37 by 49 cm and the distance from the top edge of the helmet, which is about a far as the eye of the avarage subject, to the screen is 129 cm. The preferred screen resolution is 1024 * 768 at 60Hz. It is strongly recommended to stick to this (see section about refresh rate below).
Important note on procedure
Note: Please NEVER turn the projector off. If the stimulation computer is turned on when the projector is off, it messes up the video settings as the computer will not see the projector anymore, so will set the LCD monitor to be the main one. It can, and should, be set to stand-by, using the 0 and 1 buttons on the remote. Just do not use the main switch on the side. There's a sign with some help on how to use the projector on the wall in the control room.
Using a light diode, we have measured a 34ms delay between the trigger codes sent from the stimulus PC (synchronised with a screen refresh) to, eg, the MEG machine, and the actual visual change on the screen in the MEG room. This time seems to reflect two screen refreshes (at 60Hz) that are buffered by the projector. This time will therefore need to be added to the times of trigger codes in the MEG data. The EN acquisition software will actually enable you to specify this delay for a visual event in the acquisition setup. When specifying this the corresponding trigger will be delayed by the correct amount of time, and will be in sync with the actual presentation again.
Maximum refresh rate
The maximum refresh rate that the projector will produce is 60 Hz. You should not try to produce displays at a higher rate. Although the projector will accept the signal from the computer at higher rates (75 Hz) and produce a display, it drops frames to downsample the output to 60 Hz, making the exact timing of the stimulus unpredictable.
We also found out that changing the resolution can affect the refresh rate. The default resolution on the stimulus presentation computer is 1024 * 768 at 60 Hz. When you tell E-Prime to use a resolution of 1280 * 1024, the refresh rate also changes to 75 Hz for no apparent reason. As mentioned above, the projector is not actually able to do this.
The advice is to stick to 1024 * 768 and 60 Hz and to not change these settings. In E-Prime, the advice would be to set the diplay object explicitly to this resolution and speed, and not to just use the Windows settings.
Also, you may want to have E-Prime display the refresh rate it has detected at the beginning of the experiment. Simply add a text object containing: "Refresh Rate: [Display.RefreshRate]"
The stimulus delivery computer has a SoundBlaster ZxR soundcard.
Up to December 2014 we used the Xonar Essence STX soundcard, which produced a latency of 33ms with a jitter of +/- 3ms. Since January 2015 the soundcard is a SoundBlaster ZxR which has a latency of 26ms with +/- 2ms jitter. These latencies are measured at the inserts, so include the delay caused by the 2.5 meter of tube the sound has to travel through.
The headphones supplied by Elekta Neuromag are not able to produce high quality sound. On top of that they aren't completely free of magnetic artifacts either. We do not use them anymore. Instead, we've installed a set of higher quality headphones outside of the MSR, with 2.5 meter of tube attached. These tubes enter the MSR on the right side of the MEG machine.
In addition the sound output of the stimulus computer can be recorded in the fif output file of the MEG acquisition system. To enable this you only have to add the channel 'MISC5' to your setup.
The headphones are in-ear types. Use a new pair of inserts for each subject, and dispose of any used inserts immediately. It is easier to insert the eartips before the volunteer goes into the MSR, as this can be fiddly.
It is very important that you make sure that your soundfiles are created as loud as is possible, without distorting the sound. Make sure you adjust your recording level correctly during the actual recording, and adjust the level in a sound editor if needed.
Measurements for the headphone system are here: MEG tube phones with silicone inserts.pdf. These still used the previous Xonar Essence STX soundcard, but we do not expect the new SoundBlaster ZxR card to have much of an effect on this.
A script, called 'HearingThreshold', is available on the desktop of the stimulus presentation computer to test if the headphones are inserted properly. The subject will need a pair of button boxes and when the script is started sounds are played to both ears for a couple of minutes while the subject is asked to press the button on the side of the sound when they hear one.
When the script terminates it will report two numbers, one for the right ear and one for the left ear. These numbers are the attenuation, in dB, that can be applied to the test tone before the subject is unable to hear it. Normal values are from 45 to 75 dB. The difference between the left and the right ear should ideally not be more than 10 dB. The default test tone used is a 1000 Hz pure tone of about 1 second at maximum loudness.