cntaverage - Single Subject ERP Calculation

Contents

Synopsis

-----------------------------------------------------------------------------
EEP 3.1  Max-Planck-Institute of Cognitive Neuroscience 1996-99
cntaverage 3.22   (OSF1 V4.0 alpha)           Thu Aug 19 14:09:57 1999
-----------------------------------------------------------------------------
standard:
  cntaverage <source cnt> <cfg> [<trg in>] [<rej in>]
             [-p <output path>/[<new code>]] [-c]

eog compensation:
  cntaverage -e [-a] <source cnt> <cfg> [<cls trg in>] [<pfc in>]
             [-p <output path>/[<new code>]] [-c]

options:
  -p ...     output file name
             all text before the last "/" is interpreted as a path
             which will be created if needed
             the rest, if exists, is the new subject code
  -c         calc covariance matrices
  -e         enable eog compensation mode
  -a         apply the eog compensation to eog channels

;-----------------------------------------------------------------------------
;Example configuration file for cntaverage
;-----------------------------------------------------------------------------

Conditions:             ; list of conditions to average

10                      ; condition code (read from .trg file)
Cond1 (GREEN)           ; condition label / color

20,2A=A                 ; multiple trigger codes may be summed up into 
Group_1 (SKY)           ; one average
                        ; '=..' assigns the new condition code

                 ; "condition code" (first line) should be a 2 character
                 ; code, it appears in the output .avr filename
                 ; "condition label" (second line) is a max. 10 char.
                 ; identifier which is stored in the file, it appears
                 ; in ERP plot legends


Channels:               ; list of channel labels
FP1                     ; (may be a subset of all channels available)
FP2                     ; if no "Channels" section is spezified, all channels 
F7                      ; are averaged
F3
FZ
EOGV
EOGH

Window:                 ; averaging window in msec - relative to trigger
-200..1500

Baseline:               ; definition of baseline
-300..0                 ; (may extend beyond or be completely outside of
                        ; averaging window, if required..)
                        ; you can turn off the baseline calculation by
                        ; writing "ABS" instead of the window borders

; the baseline area reference trigger can be distinct from the
; average window reference

;Reference displacement: -1         ; you can specify a trigger index offset

;Scanning for triggers:  9 9 15 15  ; or a trigger value for each condition
                                    ; (the reference displacement value indicates
                                    ; the direction for baseline trigger search)

Rejection:              ; range where rejection marks are being checked for
-200..1500              ; (rejection may temporarily be disabled by placing
                        ; the keyword 'OFF' behind 'Rejection:')

BLUE, GREEN, CYAN, RED, MAGENTA, YELLOW, WHITE, BLACK, STEEL, SKY, MINT, SEA, LEAVES, OLIVE, SIENNA, LIGHTGREEN, OCHRE, APRICOT, ORANGE, CRIMSON, ROSE, PINK, PURPLE, LILAC, AUBERGINE, PLUM, UV

Description

Input Trial Selection/Processing

It uses its configuration file and the wealth of signal/trial classifcation lists which you got from the preceding evaluation steps (cntreject, xcnt, trged, xeog).

In detail, the following happens for each condition which is defined in the .cfg file:

Standard:

1. All triggers which are to be averaged for this condition are located in .trg. The absolute trial epochs are computed from the (absolute) trigger latency and the (relative) window size definitions in .cfg.
2. If a .rej file is present and a rejection checking window is defined in the .cfg, these epochs are checked for overlapping with any of the rejected signal epochs. Trials with such overlaps are classified as "rejected".
3. All trial epochs (the smallest epoch which covers the averaging window and the baseline window, if any, and the rejection window, if any) are checked for containing discontinuity ("__") or DC-reset ("Rs") triggers within their limits. Such trials are "rejected" too.
4. The nonrejected trials are loaded. If requested, a baseline compensation value is computed for each trial/channel as the mean of the signal values within the baseline window. This value is substracted from the signal within the averaging window.
5. The means/variances of the corresponding sample points of the averaging window of all nonrejected trials are computed. That's it. The means are believed to represent the Event Related Potential for one subject in one condition.
6. The computed means, variances, trial numbers and some additional informations are stored as .avr file for subsequent evaluations.

1. All triggers which are to be averaged for this condition are located in .trg. The absolute trial epochs are computed from the (absolute) trigger latency and the (relative) window size definitions in .cfg.
2. A classification attribute for each "trial triggering trigger" is expected and read from .trg. (The trigger file with this attribute comes from xeog's "Save Trial Classification" function). It is an error if no classification can be found for one of the trials.
3. Trials with a "m" or "b" classification are passed through the EOG compensation step. See the algorithm for details.
4. Further processing is done as in the standard mode starting with step 4.). The "u" trials and the corrected "m" and "b" trials are nonrejected now, the "r" trials are rejected.

./foo.cnt
./foo.trg
./foo.rej
../cfg/average.cfg

Conditions: 

da,di,dum=d1
Daddel (GREEN)

li,la,lei=l1
Laber (RED)

./test/foo.bsl     - baseline value listing
./test/food1.avr   - ERP for subject foo in condition "Daddel"(d1)
./test/fool1.avr   - ERP for subject foo in condition "Laber"(l1)

./test/pa53.bsl    - baseline value listing
./test/pa53d1.avr  - ERP for subject pa53, formerly known as "foo",
                     in condition "Daddel"(d1)
./test/pa53l1.avr  - ERP for subject pa53, formerly known as "foo",
                     in condition "Laber"(l1)