Specification of a PET time window for MR coregistration

[bilgelm]

Rather than using the average of the full dynamic PET, using the average of the first few frames might yield better MR coregistration since earlier frames reflect anatomy better.

This optimal window could be automatically detected (by finding the frame with the peak counts (?)) or be specified by the user.

Dynamic PET could be useful for this purpose. It performs a series of checks on the frame timing information and includes functions to extract a given time window and calculate a weighted average (dynamic_mean).