Calculate Crossbleed Correction Factor

• Calculation of amount of spectral cross bleed between two images
• Ideally recording of a single fluorophore in multiple channels is used, but the algorithm, based on cross correlation calculation of images can calculate cross bleed coefficient when multiple fluorophores are present, but in distinct pattern in the image to be corrected. However the channel that's crossbleed is calculated (Image B) has to be clean single fluorophore fluorescence).

Basic usage:

• Two images (two different channels) have to be open and linked.
• Time lapses can be used, in this case all frames are taken into account
• Open the Calculate Crossbleed Correction Factor from the Tools menu.
• To calculate the contribution of the fluorescnece signal of the fluorophore present image B into image A set the names of these image windows in the dialog.
• Use the Calculate button for calculation
• The  result can be directly entered into a cross bleed correction coefficient matrix in the Preferences dialog.

Theory:

The pixelwise cross-correlation of two images is (close to) zero, when images are showing completely different structures. If cross bleed occurs, the cross-correlation is nonzero. By solving the equation that the cross correlation (which is analogous to product) of ImageA-x*ImageB equals zero, x can be calculated as the cross-correlation coefficient over the auto-correlation coefficient of ImageB.

Cross bleed correction coefficient matrix

The cross bleed correction coefficient matrix matrix is declared as kernel in the Preferences dialog.
Each row corresponds to a channel, and tells that which fluorophores (columns) have how much contribution to this channel. The diagonal is by definition ones.

Example:

Let's say you want to spectrally unmix fluorophores A and B, which are recorded as Channel 1 (Ch1) and Channel 2 (Ch2), respectively.
i.) You have the possibility of staining your sample with either with A or with B:

1. Using your acquisition software record both Ch1 and Ch2 in the sample stained by A and the in an other sample stained by B.

2. In the Image Analyst MKII load the images and subtract background. It is very important to perform background subtraction in the same way when performing cross bleed calculation and when performing the actual unimixing, e.g. at the same percentile.

3. The first row of the matrix describes Ch1.
   Fluorophore A contributes by definition by 1 to Ch1. This is in the top left corner of the matrix.
   To calculate the contribution of B to Ch1, use the images acquired with fluorophore B only and set Ch1 as Image A and Ch2 as Image B in the Calculate Crossbleed Correction Factor dialog. Press Calculate, and enter this value in the top right corner of the matrix.

4. The second row of the matrix describes Ch2.
   Fluorophore B contributes by definition by 1 to Ch2. This is in the bottom right corner of the matrix.
   To calculate the contribution of A to Ch2, use the images acquired with fluorophore A only and set Ch2 as Image A and Ch1 as Image B in the Calculate Crossbleed Correction Factor dialog. Press Calculate, and enter this value in the bottom left corner of the matrix.

ii.) If you don't have samples separately stained, but the fluorophores are staining distinct structures:

1. If only Ch1 contains cross bleed form fluorophore B, but not the Ch2 form fluorophore A, the above calculation can be performed.
2. The algorithm currently in use is not suitable for calculation of cross bleed when it is mutual.

Alternatively use Blind Spectral Unmixing with NMF (nonnegative matrix factorization) to calculate crossbleed coefficients between several cross-talking channels.