[Homer-users] gaussian GLM, odd beta pattern

[Defenderfer, Jessica]

        External Email - Use Caution        

Hello homer users,

Im using hmrDeconvHRF_DriftSS to regress my short channel and also estimate the HRF using idxbasis1 (series of gaussian betas, using parameters 0.5  0.5).  I am taking an average of the betas during the expected peak of activation. I continue to get really unexpected beta patterns that I don’t know how to explain. For example, the block averaged group results reveal that channel 3 during condition 2 demonstrates hemodynamic activation (oxy is positive, deoxy negative), but the betas suggest that oxy is negative and deoxy higher. Many channels that appear activated (block averaged) in EasyNIRS, have negative beta values. Can anyone help me better understand what is happening?

Thank you,
Jessica