How to properly remove cosmic spikes from Raman data?

[PjotrMeyvisch]

Hi all,
I'm having difficulties figuring out which values to choose for the cutoff, threshold and distance to average parameters to remove cosmic spikes from Raman data ("Spike Removal" in Preprocess Spectra widget). I often end up removing some spikes but also much broader bands originating from the samples (attached screenshot shows 532 nm Raman data of fossilized organic matter with strong G [~1600 1/cm] and D bands [~1350 1/cm]).

Is there any specific processing, other than baseline correction and some sort of normalization, that is advisable to do prior to spike removal that would increase the performance of this algorithm? I'm under the impression that the intensities of the spikes are too low compared to those of the real features to allow a clear distinction by the algorithm.

Since the spike removal algorithm, based on Whitaker and Hayes (2018), uses modified z-scores to distinguish spikes from real features, is it possible within Quasar to view these z-scores to better determine a threshold value?

Any tips and tricks are much appreciated!
Kind regards,
Pjotr Meyvisch