KGS Lab Manual

How to...

• Mount the RAID / your home directory on a Windows box
• Run Real-time fMRI
• Use the High-Res Surface Coil (RessCoil)
• Make a Bite Bar
• Do Manual Shimming
• Do Subslice Prescription

Mount the RAID / your home directory on a Windows box

Machines in the KGS lab

Here is a partial list of machines in the Grill-Spector Lab, and the rooms in which they're located.

The main machine is moach, which is the server for the local lab. Most lab members' home directories is on some partition of moach. The other big machine to know is white, which is the Wandell Lab main server, but which we share for many purposes (including mounting the RAID as well as mounting drives off these linux boxes on Windows computers). When you connect to either white or moach (or eynaim), either by using SSH or by sitting down at the linux workstations, you should be able to see more or less the same setup: the same drives being mounted, and the same programs installed (e.g., MATLAB uses a common license manager on white). The RAID partitions available through this setup are: /biac1/kgs/ (500 Gb), /biac2/kgs/ (1TB), and /biac2/kgs2 (1 TB)

Mounting a drive

The recommended way to mount a directory from these sytems on Windows is through white. (Some lab members mount via the BIAC machine sns2, in the BIAC workroom, but this is more complicated and not recommended). On your Windows machine:

1. Test that you can connect to white:
   1. Under the Start Menu, select "Run ..."
   2. Type "explorer" at the Run prompt. A Windows Explorer window will open up.
   3. Enter "\\white.stanford.edu\biac1-kgs\" in the address bar. It's important to use back-slashes (\) rather than forward slashes (/) -- the former is for connecting to drives, the latter for connecting to web pages using protocols like HTTP.
   4. If things are working properly, the window will prompt you for your username and password enter them. (You can go ahead and save them, if they're the username/password you plan on using for mounting the machine.)
   5. If the drive successfully connects, you'll see a list of directories such as "3Danat" and "mri". Great! You can now mount the drive; go to step 2 (note you haven't mounted it yet, just connected to it.)
   6. If you can't connect to it, try contacting Rory, or else Martin Frost (action AT biac) if you have a serious problem.
2. Either by right-clicking on your "My Computer" icon, or by using the Tools menu in a window, select "Map Network Drive..."
3. A dialog should pop up. Select the drive letter you'd like to use (like Z: or X: or, what the heck, K:). Then enter the path you'd like to mount. Some common ones:
   • First RAID partition: \\white.stanford.edu\biac1-kgs
   • Second RAID partition: \\white.stanford.edu\biac2-kgs
   • Third RAID partition: \\white.stanford.edu\biac1-kgs\biac3
   • Home directory: First you'll need to log into moach/white/eynaim. On Windows, you can connect securely using the PuTTY program: http://www.chiark.greenend.org.uk/~sgtatham/putty/download.html
     1. Connect to white.stanford.edu and log in.
     2. Before changing directory, type "pwd". This will give you your home directory path. Note this path.
     3. Then, cd to /biac2/kgs/. Use the following command:
        ln -s [home directory path] [my user name].
        e.g. ln -s /moach/users/kwyjibo/ kwyjibo
     4. This will create a link to your home directory on moach in /biac2/kgs/. You can now go back to Windows proper (exit putty) and map this through white.
     5. In the "Map Network Drive" dialog, enter \\white.stanford.edu\biac2-kgs\kwyjibo [i.e., your account name]
4. Click ok. Voila! You should be mounted.
5. An aside: it's sometimes helpful if you use the same username on your Windows box as on moach/white. But if you want to use a different one e.g. for security reasons, then you can click on the "Connect using a different username/password" link in the network mapping dialog. I've heard reports of this being sometimes flakey, but I've tried it and it seems to work nicely.

Run Real-time fMRI

This is used to see the results of the functional scans as they happen. This can be quite useful to do several things: for instance, to check for scanner spiking noise, check that localizer reference scans are working, or confirm that the prescribed inplanes contain the functional regions you're interested in.
The main instructions for running this are on David Ress' page .
However, there are a few notes here that may be of relevance to people in Kalanit's lab:

• The machine, rtviz, that we use to run the real-time software, is also used by Chris DeCharms and his coworkers. When they use it, they boot the machine into Windows mode. If you are scanning and want to use this machine, and notice it's in Windows, mode, you can reboot the machine. (Someone may be logged in, but if you've reserved the scanner time, you're justified in making use of the machine and restarting.)
  You can reboot it by doing the following (also posted on the side of the monitor):
  1. Restart the computer (In Windows, select the prompt in the lower left corner)
  2. Once the machine restarts and the white-text-on-black-background comes up, press F1 (may need to press a couple of times) to get into the BIOS manager.
  3. Go under Boot -> Hard Disk Drives and hit Enter on the first Hard Disk listed.
  4. A window will pop up with 2 options. Select the Seagate drive to boot Linux. (The IBM drive will boot Windows).
  5. Hit F10 to exit, and Yes to accept/save the changes.
• It's important to read through Ress' instructions before you go, but if you just want to try it out, there are very well-written Quick-start instructions . Remember to Create P-mag files and log out of the machine before you finish. The real-time code will also create P*.img files along with P*.7 and P*.mag files; these can be discarded after you're done.
• The analysis software is prinicpally oriented towards applying a sinusoid, a la mrVista. However, for event-related experiments, or block-design experiments which don't alternate blocks in a predictable manner, you can use the bandpass analysis option to get a good sense of where your ROIs are. For block designs, a good frequency to include in is that of your blocks -- e.g., if your blocks are 16 seconds long, make sure the range you input into this analysis includes 1/16 Hz. The defaults provided on the instructions page are useful.

Use the High-Res Surface Coil

[Coming up ...]

Make a Bite Bar

Bite bars can be very useful for reducing subject motion. But they can also make the scanning session like a trip to the dentist. Best to use your own judgement as to whether to use a bite bar. (For sure, they shouldn't be used during long 3-D volume anatomy scans, as they do make subjects tend to swallow more frequently, which can make the 3-D anat turn out funny).

To make a bite bar, you'll need a blank bite bar (a hard, black plastic doo-dad with a screw in the middle for attaching to the bite bar mount) and some bite bar material (generally gray-colored claylike wafers). Both of these can be found in the Grill-Spector drawer, but if you don't find it there, try the miscellaneous (aka Heeger lab) drawer, or ask to use one from the Wandell/Gabrieli drawers.

Making the bite bar:

1. As a precaution against leaving a mess, you may want to get a disposable blue mat from the side of the shelf above the lab drawers in the 3T suite.
2. Get a large cup and fill it with hot water from the cooler. (Push in the button on the back of the red-handled spigot). Careful, the water is quite hot.
3. You'll only need half of the clay wafers for a bite bar. If the wafer is long and rounded at both ends, break it in half. Dip the half you're going to use in the hot water for about 20-30 seconds, until it starts looking like silly putty.
4. Wrap the gooey material around the black bite bar blank. Try to make it U-shaped -- you just want enough material to fit around your teeth. If it starts to harden in a messy position, dip it back in water to soften it up again.
5. While the material is still somewhat soft, bite down strong enough to leave an impression of your teeth. When you use it, you won't need to bite down very hard, so make sure you'd be comfortable having your teeth loosely grip the bite bar.
6. Let it harden. It may be good to get a container (like a cup) that you can mark with your name. (No fun using someone else's bitebar. ;)

Attaching the bite bar:

• The mount is in the large tupperware container to the side of the magnet. It is a wooden U-shaped thing. You will also need the large screw and (optionally) some washers/spacers -- they're all in a plastic bag in the same container.
• Put the large screw in the top, with a washer if needed. Screw in the bite bar underneath, also using a washer as needed. No need to tighten it all the way yet.
• When the subject is lying with his/her head in the coil, carefully lower the mount on to the cradle. Let the subject tighten the sides and top of the bite bar when he/she is comfortable with it.

Do Manual Shimming

We don't usually perform shimming but Gary says we should. The point of shimming is to compensate for inhomogeneities in the magnetic field (mainly caused by the differences in magnetic susceptibility of the various tissue types in your head). It's not particularly difficult, but you might allow an extra 15 minutes the first time you do it.

A session takes slightly longer (about 5 min) with explicit shimming. You do two short scans. Autoshimming provides only a linear correction to the field; the explicit shimming does a higher order correction.

Two critical facts:

1. The purpose of writing down S/I (superior/inferior) numbers after the inplanes, transforming them with the shimcalc utility, and entering the transformed numbers into the shim scan protocol, is to prevent the table from moving between scans.
2. Autoshimming must be disabled (unselected) in the functional scans. Otherwise, this will override the higher order shim and you will have wasted your time.

Do Subslice prescription

We sometimes want to do an initial functional scan, and then choose a subset of slices for the following scans. To do this, you use a perl script called subslice. Here's an example:

> subslice
Enter anatomic prescription:
R/L values (start, stop): r0 r0
A/P values (start, stop): p45.2 a33.6
S/I values (start, stop): s33.5 i22.6
Slice thickness, spacing, num slices: 4 0 12
Enter desired functional prescription:
first slc #, last slc #, num slcs: 3 6 2
Here is the functional prescription:
R/L A/P S/I
Start R 0.0 P 30.9 S 23.3
End R 0.0 P 9.4 S 8.0
# slices: 2 Table delta: 10.2
Slice spacing: 8.0

This is for an oblique prescription, and it calls for a Table delta. If Signa actually performed this correctly, all would be well. As it is, you'll get the correct slice locations, but the table will move 10.2 mm even if you type in the value. For axial, sagittal or coronal, the Table delta works ok, and it will give the proper prescription without moving the table although there may be round off errors.

So here are the step by step instructions:

1. Prescribe your inplanes as usual
2. Run your initial functional scan as usual
3. Write doen the R/L, A/P, & S/I values
4. Copy the functional series (right click on the series name at the left, copy. Then right click, paste)
5. Run subslice
6. Enter the subset prescription explicitly (don't do graphic RX) Save series, etc.

   Last modified 2004-Jun-8 05:28 GMT.