 |
 |
Hi, have been looking at this in a bit more detail, some info below. First of all, a correction -- the second stage of PNM generally runs OK (produces a number rather than NaN at least). Not sure why I got NaNs previously. However, the output of stage 1 has changed as I described (from decimal to hex). The effects of this change which I have found are: 1. The firefox page you can use to check the peak detection just doesn't work -- the page loads but no data is displayed. I can load the data in matlab (and parse the hex numbers) and the output looks OK (i.e. there are cardiac traces and peak detection where it should be). 2. I have run Pnm now on some data I collected and processed last year (when Pnm output decimal data) and then compared the regressors which were produced to what I am getting now-- the regressors are completely different (i.e. using fslmaths to subtract one from the other shows they are not identical). This is occurs whether the analysis includes the removal of erroneous peaks or not (i.e. whether I run stage two with or without the --carddel etc) So, although I was wrong about stage 2 producing NaNs, the behaviour of the program has changed as have the regressors it is producing. Apologies for misdirection with the NaN thing, but would be great if you have any suggestions about this Mike