Recent versions of 3ddeconvolve seem to recommend higher and higher values of polort and barf up warning messages if the one on the command-line is deemed to be too small. I wish this could change for the following reasons.

1. The BOLD signal has low-frequency components for event-related designs. If jittered ITIs are optimized with a program such as optseq2, ITI's tend to be ordered such that short ITIs clump together, creating low-frequency peaks and valleys in the BOLD response (Liu Neuroimage 13, 759-773, 2001). The Legendre polynomials specified by polort will suck up some of this signal, thereby reducing estimates of the BOLD signal itself. This suggests that polort should be kept as small as possible.

2. I assume that the purpose of the polynomial regressors is to account for low-frequency signal changes due to scanner instabilities. This, however, is site- and protocol-specific. One source of low-frequency noise is interfering RF, and this depends on the quality of the screen room and the RF environment. Our scanner has a screen room that is much better than the specification (as measured by GE field engineers). Our QA scans show a linear drift term but no higher order terms. Another site on campus with a newer screen room of a different design has significant terms up to fourth order.

Another source of instability can be the EPI protocol. For some reason, EPI's acquired with parallel imaging and acceleration factors of two have significant third and fourth order polynomial terms while those acquired with full k-space do not.

Since we have a tight screen room and usually acquire EPIs with full k-space, it seems to me that the appropriate choice of polort is one. A lab sited next to the radome of an air force base with a screen room installed by WalMart would probably want a much higher value. Perhaps recommending polort=4 is good idea if a blanket recommendation is needed, but it shouldn't be required.

Unfortunately, I have difficulty persuading people in my lab to actually use polort=1 because, as Guys With a Website, you have the full authority of the internet behind you and all I have is experience and hard data behind me.

Perhaps I am missing something, but it would be nice if you you could lighten up on the warning messages in future releases.

Thanks,
John