There are two IDL routines that were used for de-spiking SOHO/CDS data, and they are found to give good results for EIS data. The routines are:
clean_exposure - works on 2D images (i.e., exposure images)
new_spike - works on 3D data-cubes, making use of neighbouring exposure comparisons to help flag CRs
Both routines do a good job of removing of CRs, with clean_exposure slightly more thorough than new_spike, but new_spike is less likely to remove real data (high velocity events were sometimes partially flagged as CRs by clean_exposure).
A wrapper routine called eis_despike is used to call new_spike for EIS data. It is called as, e.g.,
IDL> eis_despike, wd, wd_clean, swtch=spikes
where wd is the original data window, wd_clean is the cleaned window, and spikes is an array containing the locations of the spikes.
If you eis_prep on your data, then it automatically removes CRs with eis_despike unless the /nocr keyword is given.
If you are worried about whether real data are being flagged as CRs, then use the routine eis_spike_viewer.
If you find an example where real data has been removed then please contact Peter Young.
There are usually too many cosmic rays during SAA periods for the de-spiking routines to work effectively and so you should generally ignore SAA periods in your data. A possible exception is if you use very short exposure times, but this has not been checked.