This screen provides information
on pixel statistics.
The shortcut key is (ctrl)-T (or Apple-T on Macintosh).
In the default configuration, the program calculates the
number of pixels with total counts in each of 20 bins, ranging from the least intense
to the most intense. Both the total number of pixels, and the percentage of pixels, in
each bin are calculated. This is done both for the entire data set, and for the
region selected in the Plot Panel. This allows the user, for example,
to exclude strong scattering near the beam stop from the statistics.
If desired, new bin ranges can be entered by editing the low and high ranges
for each bin. The bins do not have to be in order, and they can even overlap.
To recalculcate the statistics using your new, edited ranges, click on the Recalculate
button. To restore the default binning ranges, click on the Autocalc button.
Boxes at the bottom show the minimum, maximum, and average counts per pixel in the
entire data set, and for the
region selected in the >Plot Panel. These are straightforward. There is also a box labeled
"Dispersion". This is more complicated. It calculates
D = AVERAGE [ (yi - Ei) 2 / Ei ]
where yi is the actual counts in pixel i, Ei is the
average of the counts in the four surrounding pixels (i.e., a measure of
the "expected" counts), and the average is done over pixels with 50 or
more counts. This parameter has various interpretations:
- Suppose first that the data are fairly flat and featureless (for example,
produced by a "flood field" calibration source) and that the
counts in each pixel are actually equal to the number of photons that hit
that part of the detector. In that case, Ei should actually be
the expected number of counts in the pixel in question, and if Poisson
statistics hold then D should be equal to 1.
- Suppose instead that the signal is still featureless, as above, but
the counts reported for each pixel are only proportional to the number of
photons that hit that pixel, yi = a Ni. In that
case you can show that the parameter D is actually a measurement of the
proportionality constant "a".
- On the other hand, if there is curvature to the data (for example, if
the data set contains Bragg peaks, not unknown for x-ray data sets!) then
D will be greater than 1 even if Poisson statistics hold. In this case
the significance of D is somewhat less clear--it is a measure of the
curvature, or second derivative, of the data set.
In Batch mode, see the CALCSTATISTICS, NUMSTATBINS, SETSTATBIN, and SAVESTATISTICS commands.
|