[Insight-users] Pros and cons of the MeanSquaresImageToImageMetric ?

[motes motes]

The MeanSquaresImageToImageMetric also know as the SSD (sum of squared
differences) is often a good similarity measure between images. I have
read in various articles that is not recommended for images with large
intensity differences but they don't explain why. Are there any
documentation anywhere that explains the reason for this?