Tor,
This looks wonderful, but...
>??? Undefined function or variable 'resample'.
>
>Error in ==>
>C:\MATLAB6p1\toolbox\OptimizeDesign10\core_functions\designvector2model
.m
>On line 33 ==> model = resample(model,1,TR*10);
>
>Error in ==> C:\MATLAB6p1\toolbox\OptimizeDesign10\optimizeGA.m
>On line 441 ==> model =
>designvector2model(stimList,ISI,HRF,TR,numsamps,nonlinthreshold,S);
>
>Error in ==>
>C:\MATLAB6p1\toolbox\OptimizeDesign10\example_scripts\ga_example_script
.m
>On line 220 ==> M = optimizeGA(GA);
Does this mean the signal processing toolbox is required to run the cool
genetic algorithm stuff?
Best wishes, Geraint
__________________________________________
Geraint Rees MRCP PhD
Institute of Cognitive Neuroscience,
University College London,
17 Queen Square,
London WC1N 3AR
phone +44-20-7679-5496
fax +44-20-7813-1420
web http://www.fil.ion.ucl.ac.uk/~grees
|-----Original Message-----
|From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]]
On
|Behalf Of Tor Dessart Wager
|Sent: Monday, September 09, 2002 5:42 PM
|To: [log in to unmask]
|Subject: Announcing release of genetic algorithm for experimental
design
|
|Dear SPM list,
|
|We'd like to announce the release of a new toolbox for experimental
design
|in fMRI. The beta version of the toolbox is a set of Matlab routines
that
|implement a genetic algorithm for experimental design. The algorithm
can
|be used to optimize:
|
|a) The efficiency of estimating one or more contrasts
|
|b) The efficiency of estimating hemodynamic response functions
(HRF)
| for different trial types
|
|c) Counterbalancing the frequency with which each trial type
follows
| each other one, n steps back in time (useful for, e.g.,
selective
| averaging analysis)
|d) Linear combinations of these specified by the user
|
|The algorithm works by optimizing the order of presentation of
different
|types of events, including rest intervals. The calculation of
efficiency
|can take into account nonlinearity in BOLD response (with a very crude
|model), high- and low-pass filtering, inclusion of rest or probe
intervals
|periodically throughout the run, and scanner noise autocorrelation. In
|addition, there is some provision for analysis of events whose identity
|depends on the nature of previous events and mixed block/event-related
|designs.
|
|Also included is a copy of a recent draft of the paper describing the
|genetic algorithm, which is in press at Neuroimage.
|
|We have seen substantial improvement in design efficiency using the
|toolbox (approx. 6 x for contrast estimation, 1.5 x for HRF estimation,
|and (3 x / 1.25 x for simultaneous optimization of both).
|
|The functions and documentation can be downloaded from:
|
|http://www.lsa.umich.edu/psych/research&labs/jjonides/download.html
|
|although they may become available soon from the spm toolboxes website.
|Some experience with Matlab may be very helpful in using the functions.
|To install after downloading, you will need to unzip the gz
(unix/linux)
|or zip (windows) file and put the directories it contains in your
Matlab
|path. To get started, see the documentation file
|Genetic_Algorithm_readme.rtf or .doc.
|
|Thank you,
|
|Tor Wager
|Tom Nichols
|
|
|_____________________________
|Tor Wager
|Department of Psychology
|University of Michigan
|Cognition and Perception Area
|525 East University
|Ann Arbor, MI 48109-1109
|
|Office: 734-936-1295
|Home: 734-995-8975
|Email: [log in to unmask]
|_____________________________
|