Thresholding the EEG sources @ group-level analysis

Hellow EEG experts!
I come up with somehow a fundamental question in interpreting the group-average EEG sources specially after some observations in my data at hand.
For a single subject, we may simply threshold the source amplitudes to make an estimation of active regions.
In a group-level analysis, we may conduct t-tests and threshold the p-values of the sources.
In my data, I would get a good result if I set a statistical threshold of 0.05, and then set an amplitude threshold on the survived sources.
As EEG source localization is somehow noisy both due to raw data itself as well as due to the under-determined inverse solution, it seems wise for me to do such a two-level thresholding in a group-level analysis. Has such a similar concept been discussed in the literature?

Best regards

What matters is that you describe your approach so that it can be understood and replicated by other researchers. In principle, you need to set a p-value threshold first before your run analyses and then interpret the resulting maps without further post-processing. But again, if you opt to adda another layer of thresholding, you need to provide all the details and then indicate that the effects observed are with a type-I error probability < 0.05.