How to Understand the Stability Data

Tabular Data:

The tabulated data indicate statistics on the Peak to Peak fluctuations in an RoI in the center of each slice.

Column 1:

Peak to Peak intensity variation

Column 2:

Average intensity

Column 3:

Standard deviation (in percent)

Column 4:

The fourth is the correlation between this slice and the average of the others (to find evidence of outlier behavior)

Column 5:

Ratio of the SD to the correlation. If the SD is very high and the correlation is low, we worry about this location

Column 6:

Slope of the linear regression of the signal in time and its 95% confidence limits. This tests for drift

Column 7:

Probability of non-zero slope. This might be considered (crudely) the probability that that the 'drift' might contaminate your statistics

Images:

• The first and second images show the peak to peak and standard deviation by location. Asymmetries might suggest position drift. The edges will usually be hot spots, as will the air bubble at the top of the phantom.
• The third and fourth images show the maximum and minimum intensities over time.

Charts:

The colored graphs are the time-intensity plot (the raw data for which is linked to the page).

• The upper left is the raw intensity for three different locations.
• The upper right is the low frequency ('drift') component of the same data, and
• the middle left is the raw data less the low frequency components.
• The middle right is the average across all slices of the smoothed and raw time course
• The bottom left is a scatter of the middle slice versus the two end slices.

The idea of the last two data sets is to look for outlier behavior. The slices should look very similar and any low frequency variation should be highly correlated. If it is not, the implication might be that there is some sort of position-dependent spatial shift.

The source code for P2P can be found at our web-based distribution point.

This page is maintained by Mark Cohen [updated 4/5/04]