Revision as of 13:52, 18 July 2015 by Rjpatruno (Talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

To use mrAlignMI (mrAlign Mutual Information) in matlab run:

> mrAlignMI

First, you will be asked whether to correct for intensity uniformity. This is recommended for surface-coil data but not for regular head-coil data sets.

Second, the program checks whether an existing transformation exists. If it finds one it will ask you whether it should start the aligment from this existing alignment. This is useful to refine an existing estimate. In this case the optimization search will start and end with smaller sampling steps.

When finished mrAlignMI will put up images that overlays the two volumes. These overlay images can be used to validate how well it did. mrAlignMI will open at least three windows. The first one shows the high-resolution anatomy (green) overlaid with the inplane data (red), before "inital coarse alignment" and after alignment "final alignment". The second window shows the same information exept that you can flip slices and evaluate the alignment. The third window allows you to save the alignment in the mrSESSION.mat file.

We find that in most cases the alignment is succesful on the first run. However, the alignment can fail. Typically, if the alignment failed the two volumes will be clearly misaligned. In this case, specifying a different (closer) starting position for the optimization may help. You can specify a differt starting position by giving an inital manual 6 parameter rigid body transformation:

 manual_xfm = [x-tranlation y-translation z-translation x-rotation y-rotation z-rotation]

For example, if you ran mrAlignMI and you found that the end result was misaligned, say the inplane was lower than the high-resolution anatomy. You can specify a initial rough alignment (defined in mm) that moves the inplanes 10mm up:

 manual_xfm = [0 0 10 0 0 0];"

and re-run mrAlignMI:


Alternatively, you can modify the transformation that the mrAlignMI estimated. This transformation is shown in the last line of output of mrAlignMI and looks something like this:

 3.181     -3.466     22.43     -02396   0.04858     0.0536    | -1.0682

If you think this alignment is resonable but too low you can copy and modify it and rerun mrAlignMI:

 manual_xfm = [3.181     -3.466     22.43+10     -02396   0.04858     0.0536];
Personal tools