Block wise bad channels interpolation before ICA ?

Hi, all.

Suppose I have a 10-block experiment and corresponding EEG file. In certain blocks, maybe 2 or 3, some channels became bad for some mysterious reasons. Is it OK to interpolate the bad channels within that specific block and then concatenate the whole ten blocks to create a longer file and run ICA with the longer file ?

That should be fine, as long as the number of components is below the minimum rank of your data.