Processing

Once all the previous parameters have been defined, the alignment of the selected images is performed by pressing the REGISTRATION button.

An indicative value of the “goodness” of the alignment (score) is displayed in the application’s console (bottom left). This value is calculated based on the proportion of well-overlapping binarized images relative to the total overlapping area. The quality of the overlap can be assessed by viewing the Combined image in binary mode (with Apply Mask enabled). From a visual point of view, red and green zones indicate areas of non-overlap (equivalent to a ‘XOR’ operation) between the original binarized Fixed and Moving images.

Combined images after registration in gray (left) and binarization (right) mode respectively leading to a score of 95.8% with StackReg.

With each registration computation, the matrix corresponding to the affine transformation applied to the Moving image (regarding the transformations activated in the Options tab) is displayed in the terminal associated with the application’s execution:

[[  0.77767435   0.43040319 -24.41735391]
 [ -0.63508967   1.16203181  52.11186476]
 [  0.           0.           1.        ]]

The INV mode allows the roles of Fixed and Moving images to be reversed during the registration processing.

After registration, each of these images can be saved via the SAVE IMAGES button.

The alignment parameters (e.g., ROIs, thresholds, …) can also be saved in a .json file using SAVE PARAMS for later reuse with RELOAD PARAMS.

Note

To reduce computation times during the registration processing, high-resolution images are sub-sampled to a size of 512 pixels per dimension. This criterion can be adjusted through the Max image size parameter under the Registration section in the Options tab.