help PCA find the right axes

[beanumber]

See (msschwartz21/deltascope#6)

Stretch along x, y, and z by different scalars so the PCA is more likely to find the right axes.

[beanumber]

@msschwartz21 what are the physical scales of the measuring device? How many microns for x, y, and z?

[msschwartz21]

The scalar which I used for pca is 1000x, 100y, 1z.

The voxel dimension measured by the microscope is 0.13 um x, 0.13 um y, and 0.21 um z.

[beanumber]

I'm putting this correction into the coordinates. So the originally coordinates will now be in microns, rather than voxels.

[beanumber]

52c8e8c seemed to result in some mild improvements. Note the curvature on AT_33 is much better!

file:///home/bbaumer/Dropbox/barresi/quadratic_range.png