Stimulus delivery in the CBU Tim Trio
Stimulus delivery computers
There are three stimulus delivery computers. All have identical software and receive the scanner trigger pulse and button box responses.
Software
The machines run Windows 7 . Installed software is Matlab 2014A, GStreamer V1.12.2, Sliksubversion, Eprime2 pro sp1, Cogent Toolbox versions 1.25, 1.32 and 1.33, Psychophysics Toolbox versions 3.0.12 and 3.0.14, Presentation version 18.2, Anaconda Python 2.7 with Psychopy and Pygaze.
Synchronising with scanner
From Matlab and Python National Instruments PCI 6503 card is used to capture all pulses (scanner synch pulse and button pulses alike). You can access its functions via ScannerSynch and MEGSynch (Matlab), or python_scansync (Python, MRI only). This is the recommended route.
From Eprime and Matlab, you may use the parallel port to synchronise, and to collect button box responses. See below for details.
Parallel port
There's now also the possibility of using the parallel port as an input/output device in your experiment. For the moment only the leftmost stimulus presentation machine has this option. All signals will be send to both the National Instruments PCI 6503 card and the parallel port.
BEWARE!! All signals coming from the parallel port are inverted as compared to the PIO signals. This means that a button press will show as a button release! This can be solved quite simply by setting the 'Invert' property in the port device properties dialog in E-Prime.
As the scanner pulse is inverted compared to the other buttons, this will now still be the wrong way around, sadly. The simplest way to solve this is to enable two 'Port' devices in E-Prime, one inverted for the buttons and one not inverted for the scanner pulses.
The parallel port can be used with E-Prime, without the need for extra libraries or explicit coding in your scripts. There's also a little test program on the machines, called 'PortTest', which will show all incoming signals from both PIO and parallel port.
Setting up the parallel port for E-Prime
Please note that we no longer provide official support for E-prime on the MRI. Johan Carlin may be able to provide an example script or two, but if you are starting a new project we strongly encourage you to move to Matlab/Psychtoolbox or Python/Psychopy.
NOTE: the following applies to EPrime2 only. It may be possible to do it in EPrime too, but this has not been tested. So if you want to use the ParallelPort as a trigger, please use EPrime2 (professional).
Configuring E-Prime is relatively easy (this was done with a release candidate for E-Prime2). Say you want to configure E-Prime to get input from the parallel port for both response box and scanner pule. First, enable in your project 2 ParallelPort devices (apprently EPrime is touchy about the names you give to devices, so if possible just leave the default names, i.e. ParallelPort and ParallelPort2). For both devices set the port mode to be on Input and the address to be &H379 (not 378 -- this will make the input numbering easier to use in E-Prime). The device you want to use for the ButtonBox should also be inverted (as mentioned above). Now that you have two parallel port devices enabled, all you need to know is that the scanner pulse will be received as a number 8, while the four fingers of the right hand will be (from index to pinky) 7,6,5,4.
At this point you can just set your starting screen to pause until an 8 is received from the appropriate parallelport device, and that will trigger your stimuli as soon as a scanner pulse comes in. However, make sure that screens that need response via button box do not allow the number 8, or each trigger will seem like a response to E-Prime and your data will be of little use!
Mimic machine
In the CBU computer user room there is a "CBU mimic" configured exactly as the stimulus delivery machines, and with a pulse generator to simulator the scanner trigger, and a response box. You must test your programs thoroughly on the mimic before running any experiment.
Pre-testing machine
There is another mimic machine in the imaging centre, which you may use for pre-testing in experiments that require a training session just before the scan. You'll then need to have two experiments attend.
Visual presentation: NNL LCD widescreen DEFAULT SETUP
Recommended resolution 1920x1080. You can also use 1024x768 but be careful to change the monitor aspect ratio appropriately using the physical buttons next to the display.
Refresh rate 60 Hz.
Latency Unknown at present. Assume similar to old projector.
Size Distance between subject and screen 1565mm
Width display on MRI screen = 698.4mm (25.16 degrees visual angle)
Width if using 4:3 aspect ratio = 522mm (18.94 degrees visual angle)
Height = 392.85mm (14.31 degrees visual angle)
Visual presentation: Christie video projector
Recommended resolution 1024x768
Refresh rate 60 Hz. Please do not change this parameter. Although the projector will accept inputs up to 100 Hz, it will not actually produce output at these higher rates, but instead, sub-sample down to a rate just above 60 Hz.
Latency: There is a fixed temporal delay between computer refresh and actual display of a little less than 20 ms. If you don't account for this, your RTs are probably around 20 ms shorter than you think they are.
Size: Distance between MRI screen and mid-point of mirror = 823mm Approx distance between eyes and mid-point of mirror ~ 90mm Width of full-screen display on MRI screen = 268mm
This means the full screen display has a visual angle of 2 * arctan(268/2/(823+90)) = 0.29 radians = 16.7 degrees.
There are Matlab functions available for transforming visual angle to stimulus size, or the other way around.
Using Psychtoolbox with the NNL LCD screen
Note that if you are using the new LCD screen with anything other than the native resolution (1920x1080) Psychtoolbox's sync tests will fail, causing some fairly cryptic crashes when you attempt to Screen('OpenWindow'). The workaround is to include Screen('Preference','SkipSyncTests',1) in your script before opening the screen. But before applying this, please be very sure that your experiment really doesn't require any kind of reasonable display timing. Speak to Johan Carlin about this issue if unsure.
}}}
Auditory presentation
When presenting auditory stimuli at the CBU, you have a choice of three different headphone systems. Unless otherwise specified, the radiographers will provide you with the Siemens tube phone system for communicating with your volunteer. However, for studies in which quality sound reproduction is required, there are two other systems available that will likely be preferred.
Whichever system you use, it is important that you use good quality audio recordings in which the stimuli span use the full dynamic range available. If you don't understand what this means, you should read this page.
Option 1: Siemens tube phones
For many experiments in the scanner that require sound, but not too much audio fidelity, the built-in Siemens tube phones are sufficient.
Frequency response
Sound level set to max on PC and 9 bars on the scanner. Tones at various frequencies were then presented. The levels (in dB SPL) of the primary (H1) and first and second harmonics were:
Freq (Hz) |
H1 |
H2 |
H3 |
250 |
92.97 |
73.86 |
74.39 |
500 |
98.07 |
74.34 |
65.58 |
1000 |
98.10 |
59.20 |
27.16 |
2000 |
68.61 |
32.10 |
9.38 |
3000 |
47.24 |
5.0 |
15.82 |
4000 |
49.99 |
5.05 |
13.53 |
6000 |
27.88 |
11.57 |
7.97 |
8000 |
23.03 |
8.50 |
7.78 |
The full frequency response as measured using periodic noise from a spectrum analyzer:
Headphone simulator/pre-emphasis tool Here is a tool that allows you to do two things: (1) approximately simulate what your stimuli will sound like in the scanner (2) pre-emphasise your stimuli, boosting the frequencies that will be reduced by the presentation system
A new tool for pre-emphasis has been produced by Christopher Long, which does not produce a subtle ringing artefact produced by the previous tool. This is described here
The old tool is described on this page:
Option 2: Etymotic ER3
This alternative pneumatic tube presentation system offers higher quality. It is being installed in Mar 2006.
Frequency response
Gary Chandler kindly made these measurements:
|
|
LEFT |
|
|
|
|
RIGHT |
|
|
Amplitude dB SPL |
|
|
|
Amplitude dB SPL |
|
||
Freq (Hz) |
H1 |
H2 |
H3 |
|
Freq (Hz) |
H1 |
H2 |
H3 |
250 |
107.42 |
32.60 |
55.82 |
|
250 |
106.34 |
44.63 |
57.80 |
500 |
101.45 |
47.69 |
61.46 |
|
500 |
100.60 |
48.40 |
62.30 |
1000 |
102.99 |
56.90 |
53.20 |
|
1000 |
101.78 |
57.40 |
53.25 |
2000 |
94.66 |
61.65 |
25.43 |
|
2000 |
94.20 |
61.10 |
26.60 |
3000 |
90.59 |
50.25 |
20.10 |
|
3000 |
90.30 |
51.20 |
22.95 |
4000 |
85.30 |
30.40 |
2.38 |
|
4000 |
85.06 |
33.50 |
2.79 |
6000 |
62.82 |
20.54 |
0.30 |
|
6000 |
62.27 |
18.24 |
2.41 |
8000 |
45.28 |
6.53 |
5.90 |
|
8000 |
47.40 |
5.45 |
5.04 |
Headphone simulator/pre-emphasis tool I have made a tool that allows you to do two things: (1) approximately simulate what your stimuli will sound like in the scanner (2) pre-emphasise your stimuli, boosting the frequencies that will be reduced by the presentation system This tool is described on this page
Option 3: NNL Electrostatic Headphones
If the previous two airtube systems will not satisfy your requirements, there is one other option available, a high quality electrostatic system from Nordic Neuro Labs. This has now passed safety evaluation and is available for use. This system provides high-quality stimulus delivery and is intended for studies in which optimal presentation of auditory stimuli is essential.
Since optimal audio reproduction is not compatible with the use of ear-plugs, the NNL headphones are typically used without any additional hearing protection. They will therefore not attenuate scanner noise as effectively as using the etymotic system in combination with ear defenders. Although experience has shown that using the NNL headphone without earplugs does not produce uncomfortable sound levels, it would generally be better to use the NNL headphones only for studies which use sparse imaging.
Button box
The default box has four buttons designed to be used from one hand or two.
Microphone
We have a Phonar optical noise cancelling microphone that allows clear recordings of spoken responses without much interference from the scanner noise.
Physiological measurements
Instructions for recording skin conductance can be found here: SCR_fMRI-manual.doc, and here: fMRI_compatible_SCR_recording-CBU.ppt (with thanks to Sonia Bishop)
The same hardware can probably be used for cardiac or respiratory monitoring.
Eye tracking
We have a 50 Hz SMI eye tracking system for MRI on site. See the Eye tracking Wiki for more information.
Points of contact
Gary Chandler, who set up almost all of this equipment, is the best person to ask if you'd like an update on any changes to the system, or if something doesn't work.
If there is a piece of equipment you'd like to see added, please contact Gary Chandler or Marta Correia.
If you would like access to the facility to install your programs or test the equipment, please contact mri.admin@mrc-cbu.cam.ac.uk .