Hi Francois,
Recently I have adapted the processes: process_extract_max and process_extract_values to define the latency edges using the mid-range of the local peak. In the process process_extract_values I used this to reshape the pre-selected time-window(e.g: 300-600ms), so the mean could be computed in a less biased time-frame. This worked well for computing the mean amplitude. I used the process_extract_max to compute onset and offset of the amplitude wave also using the mid-range, but including nearest-neighbour interpolation to define the precise time-points. I couldn’t manage to include these processes well into separate options in the GUI of Brainstorm’s “extract values” menu. Also, I’m new to matlab, so my code is not really elegant and probably inefficient. But maybe you can fix this if you consider that it is worth to include these measures as an alternative to purely eye-balled time-windows. I thought it might be good alternative to areal measures (e.g. fractional peak), because it keeps a linear treatment of the wave. I compared the measures with DTW differences and it seems to work fairly well. Although it wasn’t a formal comparison, the onset and offset measures I managed to obtained were close to fractional peak precision. However, I only managed to make the code work at a single electrode level.
If you think this might be interesting for Brainstorm I’ve attached the files with the adapted code. I imagine you’re quite busy, so this is probably a nuisance, but maybe it might be useful for someone in the future. For process_extract_max, changes are from line 109 to 163. For process_extract_values, changes are from line 592 to 656.
Hoping you might find this interesting or useful. All best wishes,
Simon