FYI @larsoner @agramfort

For Megin this is ok. fNIRS community can decide.
There is only one new value; for the raw->amplitude only the name changes. This is fine if the meaning is the same. This would be true if the "raw signal" in (about all) systems indeed is the cw amplitude, and there is no need to have possibility to tell the difference. For example, that the raw signals of some systems might be something that need to be further processed to produce amplitude and phase.
I don't know about the conventions used in fNIRS, but for narrow band signal the amplitude can either mean the momentary amplitude (raw signal probably is that), or the envelope of the signal (that is, the ampitude of the analytic signal). Maybe this should be stated in the description which one is meant.

There is only one new value; for the raw->amplitude only the name changes. This is fine if the meaning is the same. This would be true if the "raw signal" in (about all) systems indeed is the cw amplitude, and there is no need to have possibility to tell the difference. For example, that the raw signals of some systems might be something that need to be further processed to produce amplitude and phase.

The two new ones are:

fnirs_fd_ac_amplitude   304  "fNIRS frequency domain AC amplitude"
fnirs_fd_phase          305  "fNIRS frequency domain phase"

And I actually initially had the same thought as you, that there could somehow be a translation between the frequency domain amplitude and phase and the time domain / continuous wave data. However, from what I understand, they are actually distinct in terms of the data they provide:

• For the frequency domain, data are initially acquired at a very high rate (MHz) to match a high stimulation rate, and every 4 ms (one sample period for the data on disk, i.e., 250 Hz sample rate) the amplitude and phase at multiple stimulation frequencies (in MHz somewhere) is extracted via FFT, and the data saved to disk are only these amplitude and phases every 4 ms (i.e., 250 Hz effective sample rate)
• For the continuous wave, data are simply measured at some rate like 250 Hz; this ends up being equivalent to the first term in the FFT above (as the value for the 4 ms sample period is simply the average over the time window, and this is the same as a DC term of a FFT)

This is explained better in mne-tools/mne-python#7912 (comment) and the comment that follows it, but basically, the meaning of "raw" is now more specific to mean the continuous wave / DC value that you get by just A/D converting the data. Frequency domain time-varying amplitude and phase of the MHz frequencies could also be considered different "raw" forms/aspects of the measured data, but they seem sufficiently different to warrant a new coil type (like mag vs grad). They are complementary and both can and will be used to estimate the existing other fNIRS types we have (optical density, hbo, and hbr).

Does this make sense?

For narrow band signal the amplitude can either mean the momentary amplitude (raw signal probably is that), or the envelope of the signal (that is, the ampitude of the analytic signal). Maybe this should be stated in the description which one is meant.

The signal formerly called "raw" that we now call "continuous wave amplitude" is indeed just the momentary amplitude, so I don't think we need to add anything (if it were the Hilbert envelope, indeed something like "envelope amplitude" would make more sense).

I entered through the diff and the second change slipped my eye. Should have read the comment thread bit better.
Yes, your explanation makes sense. And the cw / fd distinction is quite clear now when I understand bit better what this is all about. So fine for me.

Okay great, thanks @mkajola @rob-luke @agramfort for the quick work!

Thanks @mkajola, sorry I wasn’t clearer in the description. Thanks @larsoner for clarifying above.