3.1.2 Channel Registration
NanoPyx's channel registration works by aligning multiple imaging channels to correct for chromatic aberrations and channel misalignments, ensuring that features captured in different channels overlap accurately. The process involves selecting a reference channel and comparing it with the other channels. The algorithm identifies corresponding features or landmarks in each channel and calculates the elastic transformations needed to align them with the reference channel. Once the transformations are computed, they are applied to the non-reference channels, aligning them precisely with the reference channel. This ensures that multi-channel images are correctly registered, allowing for accurate colocalization and analysis of different molecular or structural components captured in various channels.
Reference: Romain F Laine et al 2019 J. Phys. D: Appl. Phys. 52 163001
-
Image Stack: The image stack to be processed, required to have shape: (channel, rows, columns).
-
Reference Channel: Channel to be used as reference. Defined the image stack index of the channel (0 based). The reference channel will not be adjusted, only the other ones will be transformed to align with the reference channel.
-
Maximum Shift (pxs): Maximum shift in pixels that is expected between reference and each other channel. Will work as a hard limit for the elastic transformation.
-
Blocks per Axis: Number of blocks on which each axis is divided in order to calculate local correction vectors. Each pixel of the non-reference channels will then be corrected by weighting the distance of each vector to each pixel and applying the corresponding transformation. Often times bad alignment will be caused by using an incorrect number of blocks.
-
Minimum Similarity (0-1): A Pearson's Correlation Coefficient (PCC) is calculated for each block, this value removes from the elastic transformation calculation blocks where the normalized PCC is lower than the minimum.
-
Save Translation Masks: Whether to locally save the translation masks so that they can be later used to align other image stacks.
-
Apply: Whether to apply the calculated elastic transformation to the currently selected image stack.
-
Save Translation Masks Path: Path on where translation masks will be saved.
-
img: image to be corrected, required to have shape: (channel, rows, columns).
-
Path to Translation Masks: Path to the previously calculated translation masks. Should be in file format .tiff. Should be previously calculated using the Estimate Channel Alignment function.