Old News
We're so busy there hasn't been time to mention that this year's maintenance cycle went smoothly, and everything's working fine. Lots of users, lots of new results, and lots of updated software. MEG is cool beans, so come down to the lab and learn about the brain.
tako:/data7 is full and we've switched over to /data8. ~/data symbolic links have been updated and there's a ~/data7 link to the old location.
A 3dcalc command has been added to the MEG Analysis pages for converting SAM pseudo-F ratios to log ratios. See
here.
Happy Turkey Day, or Duck Day, or Ham Day, or Eggplant Lasagna Day, as the case may be. By the way, /data6 on Tako is full and most people are pushing data to /data7 now by default. ~/data symbolic links have been adjusted to reflect this and ~/data6 links have been made available.
Richard Coppola, director of the NIMH MEG Core Facility, displayed astonishing electromechanical talent by successfully repairing the power supply. The system is up and running. Click
here for pictures.
It turns out that the reason the new power supplies failed is that one of the cables connecting the supply to the rack is an old version that is no longer used. Now that we know this, we are getting one of the new supplies (which had been shipped back), and we will reconfigure the cable. We should have all of this done Monday morning. We're sorry for the inconvenience.
The 3.3 volt power supply in our MEG rack started to fail. We replaced the entire power supply, which includes a 5 volt supply. Guess what? The new 5 volt supply was DOA. So we switched back to the old power supply, which, though it is failing, is at least doing so intermittently. The new 5 volt supply is completely dead.
Hey, it's time for the NIMH Retreat again! Check out our poster (link at the right). See you in scenic Gettysburg, PA!
The warmup and cooldown was successful, and we are resuming normal operations on Tuesday.
Last week before we get our vacuum pumped. There's a new page for a new app,
3ddipole. It's not so much of an analysis tool as a thing to help spiff up a presentation. Feel free to add your favorite AFNI/SUMA results picture, time-frequency plot, or a clever paragraph or two that'll make the reviewers go "oooh!"
Not so much new to report. The power failure yesterday didn't affect us much, we were just down for a few hours. A reminder: don't schedule slots unless you have confirmed subjects. Use the "hold approval" feature if you just want to reserve the time.
AFNI wizard Bob Cox has modified the command
3dpc to perform PCA noise/dimension reduction on 3d+time datasets. This is particularly useful for cleaning up SAM sliding window data. A description of how to do this is posted here:
PCA.
Happy Thanksgiving. Tako's /data5 is full, so we're switching to /data6. Symbolic links have been redirected.
The DLP projector is back, and works.
New versions of
samslide and
StockwellDs were uploaded today.
samslide has some new features (thanks Tyler!) that make it easier to use on regular machines.
StockwellDs 1.4 does normalizations in a more sane way, and the
-B default was changed (see the usage message).
The DLP projector died. No clue what happened, some electronics failure apparently. It's out for repair.
A minor bug in the
-B processing of
StockwellDs was fixed. The current version is 1.2.
The final software release from CTF is now online, version 5.4.
Another successful warmup, no problems. Thanks to Slobodan for doing it slowly and carefully. The two new 3T magnets across the hall have no effect on us; our noise levels now are lower than they were before the warmup. We have lots of disk space on the new RAID array, that will last for years.
Scheduled maintenance is going well. The new RAID array has been installed, and Tako is back up. Update: 8/24 Hmm. No, Tako is not back up. We think it might be cabling problems, and are reconfiguring things. Stay tuned. OK, it turns out, you should not mix 320 and 160 speed SCSI devices on the same card or bus.
Matlab 7.3 (R2006b) has been installed on the stimulus computer (under WinXP, not Win98). The Instrumentation Toolbox is now available.
Two new mailing lists of note: the
Biomag list has been moved from the old server at CTF to a Yahoo group, and a new
CTF users group has been formed as well.
The
Stockwell software has been updated to work with fftw3 and numpy, and there's a new command-line tool called
StockwellDs that makes pretty pictures from MEG data.
randpermute has been updated to handle Z-scores such as you might get from 3dWilcoxon (an appropriate test for T-F data).
Kurage died. The server has been replaced, but not everything is working yet. The website is partially restored. Work continues. Update: Most things are working now.
Tako is being upgraded. Host keys have changed. You'll have to edit your .ssh/known_hosts file, and remove the line for tako.
The ongoing construction for the installation of the new magnet continues. We apologize for the sudden power outages and downtime. Meanwhile we are readying a new RAID server, which will have lots of capacity. Also, initial results from head motion correction are in, and they look very good.
Happy new year. It's the year of the pig (wild pig or boar, actually) now. We have installed
VisionEgg on the stimulus computer (Ika). This is a Python package for displaying visual stimuli. Since it's written in Python, it's very easy to use, and free.
A bug in Presentation that caused an access violation on Windows 98 when the scenario tab was clicked has been fixed.
Brainhull 1.7 is now available. This version now uses numpy rather than the old numarray. hull2suma has been updated and a new command to extract fiducials from .mri files, readmrifid.py, has been added.
We installed a new DLP projector. The light curve's rise time with the old projector was something like 30 plus or minus 12 milliseconds, the DLP projector's time is about 5 ms, so visual stimulus timings will be more accurate.
Version 5.2 of the MEG Analysis Software is here! This long awaited update includes the changes necessary for continuous head localization, and also some speedups in the SAM software. Get it here:
CTFSoftwareInstallation. Note that we have not yet switched over to the new Acq, but we will soon, and any datasets collected from that point on will require this new set of tools.
We're up and running.
One of the MEG amplifier cards died. We're having a new one sent out, which should be here Tuesday. We're down until then.
Back from Biomag '06, and our warmup/cooldown cycle went smoothly this time, no broken squids, yay.
Well, time to start on that poster. The MEG lab is very clean now, and we have a new computer, named "isoginchaku", which means "sea anemone" in Japanese.
Biomag is just around the corner; 9 weeks. Is your poster done yet? Mine isn't.
E-prime has been installed on the Stimulus computer.
Happy spring. Cherries, yay! Fred will present SAMerf techniques, Thursday at 2. Also, if your AFNI is more than two weeks old, you are a dinosaur, I have heard.
We've scheduled a
user's group meeting for March 23.
Note: this is two weeks later than the originally scheduled time.
Happy New Year. Presentation has been updated to version 9.90, and the license updated. The user interface has changed a little, but don't panic, everything's still there.
There's a new version of
randpermute which has a debugging option, and a new example was added to the page.
Check out the new
@auto_tlrc script in AFNI. It works pretty well. You'll need to use it instead of
adwarp if you go this route, currently. Remember to run it on the orthogonalized BRIK, and you'll probably want
-no_avoid_eyes. The
brainhull page has been updated to reflect this and other changes in the current version of AFNI.
Also,
hilbertDs was updated to work with 5.0 datasets.
There's a new version of
randpermute called
randpermute.py (it's a Python script). This version is much more capable than the old one; it can do 3dANOVAs as well as 3dttests. It hasn't been tested much on some of the more complex analyses, so please let me know if you have any trouble with it.
Our warmup/cooldown cycle is complete, and the system is again available for recording. We lost another SQUID.
It was one of the MROs, MRO13 to be exact. Oh well, we still have 273 left. Low frequency noise has been reduced. We also got several firmware upgrades, some new electronics, and a few loose screws tightened. We upgraded Kani and Uni, did some housecleaning, and Ika's display can now operate in Dual Head mode (so, for example, it's possible for the subject to see stimuli on one screen, while you see a status display and control the experiment from the other screen). As for Tako, it turns out that /data2's raid array controller was damaged by the power failure. None of the disks will format, and the array can't be rebuilt. The unit is basically dead.
It seems that one of the disks in the raid array was already flaky, and then the power failure pushed another one over the edge, resulting in a failure of the entire array. I'm still trying to figure out exactly what went wrong, but to make a long story short, /data2 is gone.
I have a list of affected directories, but since you all already have hard backups of your data, you shouldn't need this, right? /data2 was only an emergency backup we provided as a convenience. It contained data ranging from 20-Nov-2002 through 31-Aug-2004. We never said we'd keep it forever…
This week is our annual maintenance. We'll warm the system up, and pump on the vacuum for a few days (the maintenance is actually on the vacuum-—the rest of the system doesn't really need anything). In addition we'll be making a few changes around the lab, including OS upgrades on Kani and Uni, and putting a dual-head display card in Ika (the stimulus computer). Update: Tako was accidentally unplugged, and is temporarily off-line while I run consistency checks on the raid arrays, which were under heavy load at the time. /data3 appears fine, but /data2 is still off-line due to errors.
The
Software page now has a copy of
thresholdDetect2 that works with 5.0 datasets. As you may recall, this version features an improved thresholding algorithm that works with inverted signals, proper derivative thresholding, and bandpass smoothing. It's statically linked so no library worries. A few other old tools were added to the
Software page as well.
We're off to Toronto for HBM 2005. Also, on the
ctf2st page you can find a new accelerated version of the Stockwell transform for Octave, and the
Software page has a link to some Octave code that'll let you use FieldTrip.
The SMI Eye Tracker is now working in interactive mode with Presentation, under Windows XP only. What this means is that your Presentation script can now get real-time eye position data, so for example, you can trigger a trial only after the subject is fixated, or send codes to the MEG system depending on eye position (or pupil diameter). Calibration can also be triggered from Presentation, or you can do it beforehand. Eye position and pupil diameter analog data continue to be available on three ADCs for off-line processing.
The
Ctf2st utility has been modified to run a bit faster. Also, the prestim window for normalization no longer actually has to be 'prestim'.
fred
The
Ctf2st utility now includes a t-test option for use with multiple datasets. The Matlab statistics toolbox is required.
fred
newDs2 has been updated to work with the 5.0 CTF tools.
brainhull 1.6 is available. The
inflateSpheres and
fiddist utilities are included, and the webpage and documentation have been updated.
We've updated the
Ctf2st utility to work with version 3 of the FFTW libraries, which is what recent versions of Matlab use, so installation is now nearly painless. The installation instructions have been expanded, and precompiled versions for the NIH Biowulf system are available. Many thanks to the fearless beta testers!
The optical joystick is working well again. The power supply was replaced with a quieter one, and the optical sensors were realigned. They are very delicate! Mishandling of the joystick was at least partly responsible for the noise problems we've been having. Please be careful!
dozscore has been renamed
3dNormalize. It's still a script but now includes a little C utility called
stats that's handy on its own. Plus, 3dNormalize can do 3d+time BRIKs now, which means you can average normalized and Talairached
samslide movies across subjects.
The firmware upgrade fixed our remaining hardware issues. Yay! The system now boots cleanly every time, and multiple trial recordings with lots of stopping and starting are no longer a problem. Yay!
Two minor bug fixes were added to the ctf2st code. The scaling for averaging multiple subjects was incorrect and datasets with only virtual channels were unreadable. (Thanks JG & SJ!)
fred
We've scheduled a firmware upgrade for this weekend. Hopefully, this will take care of the last of our problems.
There's a note on the
brainhull page about multisphere inflation. If you use a brain model, instead of a head model, you need to inflate the spheres slightly, because
SAMsrc uses the spheres to define a clipping volume, and tries to account for the skull by stripping off an additional .5 cm.
Today's user group meeting has been cancelled. Most of the interesting lab news is on this page, anyway. Note: We are actively soliciting users who would like to give presentations on their results! If you have some results you'd like to share, please let us know.
Kurage has been upgraded to Fedora Core 3. If you notice any problems with the scheduling calendar or the website, please let me know.
tomh
A minor fix was added to the ctf2matlab code. A warning about reading an unrecognized data format was removed. (Thanks Masao!)
fred
After careful investigation, it appears that the old ctf to matlab code works fine with the new datasets. Please let us know if we are kidding ourselves. A slightly updated version of the ctf2matlab code has been posted.
Presentation has been updated to version 9.20 on Ika, and now runs cleanly on both Win98 and WinXP. Ika defaults to Win98, but you can use WinXP if you wish. The eye-tracker was wired in to the new ADC patch block; gaze x and y (in screen coordinates), and horizontal pupil diameter (uncalibrated) are available as analog signals, and the joystick firmware was updated.
The
CTFSoftwareInstallation page has been updated with instructions on what to do if you're running an older version of RedHat.
Most of the hardware problems have been licked. The EEG was fixed, the rack doesn't overheat, and it now boots clean every time. We still have one nagging crash left (possibly related to the earlier heat-related problem) that we hope to have fixed soon.
The website now sports W3C XHTML and CSS compliance links. If it doesn't render well, it's your browser's fault (IE). We recommend
Firefox.
Accounts on Tako have been restructured. User names are now the same as on the
calendar. The latest software is available; see
CTFSoftwareInstallation for more information.
VSM has sent another engineer to work on our hardware problems over the weekend. He'll be here until Monday. He also brought with him an official copy of the analysis software, which I'll be making available here, as soon as I finish writing the installation instructions. I'd like to take this opportunity to thank VSM for their quick response to our problems.
The problem may be heat related. In retrospect, I am perhaps not surprised. We've had heat problems before, very similar to this (repeated head loco failures) and the rack never did boot properly when it was hot.
I think I'm going to add smileys to the website…
Update: it's definitely heat related. When the rack overheats, collection fails. So, we've opened the rack doors, and things seem to be working. The engineers are still trying to track down exactly which components are overheating, but in the meantime, it appears to be working. Again, our apologies for any inconvenience these delays may have caused.
Well, we still have some hardware gremlins. We can record, but head localization is not working. It was working on Friday… Anyway, we are working with the VSM engineers to resolve this problem as quickly as possible.
The new hardware is up and running, and we are now in a testing phase. We have a new electronics interface box, which is going to require some changes in how things are connected to the MEG system. Please bear with us in the next week or two as we will all be learning how to use the new system at the same time!
Our hardware upgrade has run into a few snags, from being short one engineer, to missing parts, to firmware failures. We expect to be down the rest of the week. Sorry.
On a positive note, the new website is now live.
Happy New Year. 2005 is the Year of the Rooster.
Presentation on Ika (the stimulus computer) was upgraded to version 9.13. We switched to using a USB License Key; it must remain connected to Ika at all times. Do not remove it.
Design of the new website is almost complete. This site is a Wiki, being run by
PmWiki software. It is hoped that this will allow us to keep the website more up-to-date, and to allow a certain level of collaboration from the user community in the documentation both of lab procedures and data analysis.
The new hardware for our upgrade arrived. This upgrade will allow, among other things, head motion correction, solving one of our most annoying collection issues. The installation is currently scheduled to begin 11-Jan-2005. This is a significant hardware change, and may require some Presentation scripts to be changed slightly. More on this as it develops.
MEG system shipped from CTF in Vancouver. Time begins.
