1) Run art.m program on the QADemo/Artifact data:
 
a) Type art at your matlab command prompt: >> art
 
b) How many Sessions: Enter “1” Note: It’s best to run art.m on a Session by Session (run) basis (Session is SPM term/lingo for run) because there are large intensity changes between sessions and hence the time series variance calculation will be artificially inflated which would minimize the sensitivity to capture within session outliers.
 
c) Which Global mean to use? Select “Regular” Note: the star indicates the default suggestion, similar to SPM. This will calculate global mean intensity for each image.
 
d) Select type of motion parameters: Select “txt, SPM” Note: You can also use FSL or Siemens text files.
 
e) Navigate to the image data in your QA demo Artifact demo and select sw* images. Note: There are shortcuts that help you filter data. See the ? for full list.
 
f) Right mouse click and “select all”
 
g) Select movement parameters: “ rp_2brun1_001.txt”
 
h) Drop first scan? “no” Note: This is used when the first time point has a drastically different intensity than the rest of the scans. Now, the art program is reading in the time series data.
 
 
2) View Data Note: You can immediately see that the data set is quite NOISY.
 
a) Change thresholds: You can change the thresholds to capture the worst offending artifactual time points. Note that there are a large # of outliers with the typical default thresholds. You can see that the artifacts are caused by large motion in the middle of the session. [2, .5 .01] are standard thresholds which would correspond to 2 std for time series and .5 mm translation and little less than a degree for rotation ( 1 deg = .0174 rad) Change these thresholds to [2 10 .1] to capture the worst outliers.
 
b) Check range on top xy plot of time series: The range on the top XY plot is large, anything over 50 is typically pretty noisy.
 
c) Note that there is motion summary saved to your matlab workspace From this review of the data you can see that the data are especially noisy from time time points 78-87, you can choose to view these time points via the SPM check reg functionality (may select up to 15 images) or you can choose to view these data via a movie.
 
Susan Whitfield-Gabrieli MAR09