You certainly can do that, however the way we do variance normalization is to compute an estimate of the unstructured (i.e. random) noise variance and make this measure 1 everywhere in the brain. The effect of this is to make false positives equally likely across the entire grayordinates space.
Peace, Matt. From: Chao Zhang <[email protected]<mailto:[email protected]>> Date: Wednesday, July 13, 2016 at 3:05 PM To: Matt Glasser <[email protected]<mailto:[email protected]>> Subject: Re: [HCP-Users] fMRI time series intensity normalization Hi Matt, Thanks for the answer. When I did the extraction of ROI time series, before averaging the time series for voxels within the ROI, I did subtract the mean and then divided by the standard deviation (z score). So I wonder which one is more reasonable, demean vs z score? Doing it by the z score scheme would assume that each voxel's time series have the same standard deviation. I checked whether this is true for the FIX data (for five subjects) and a typical range for the time series standard deviation is tens to one/two thousand. In the preprocessing, after the intensity normalization to 10000 of mean, are these variations in standard deviation of time series true (biological) or can we ignore those? Best, Chao On Mon, Jul 11, 2016 at 7:48 PM, Glasser, Matthew <[email protected]<mailto:[email protected]>> wrote: This is why people typical subtract the mean image across time for many kinds of analyses (or if they concatenate multiple runs). Peace, Matt. From: Chao Zhang <[email protected]<mailto:[email protected]>> Date: Monday, July 11, 2016 at 5:57 PM To: Matt Glasser <[email protected]<mailto:[email protected]>> Subject: Re: [HCP-Users] fMRI time series intensity normalization Hi Matt, Sorry I did it wrong. Now the mean of all voxels within one frame is 10000. What I am concerning about is the mean values of each voxel across time (mean of the time series) are very different across the brain. So if one ROI includes voxels with very different baselines, can I calculate the average time series directly or do I need to do some normalization first? For example, I attached one brain image showing the map of mean values of each time series, the circle represent one ROI/region (include voxels with very different brightness/mean value) for which I want to get the representative time series, can I do the average directly? Thanks, Best, Chao On Mon, Jul 11, 2016 at 6:01 PM, Glasser, Matthew <[email protected]<mailto:[email protected]>> wrote: I don’t understand how you are computing the mean. The mean across the entire 4D dataset inside the brain mask will be 10000. Peace, Matt. From: <[email protected]<mailto:[email protected]>> on behalf of Chao Zhang <[email protected]<mailto:[email protected]>> Date: Monday, July 11, 2016 at 4:42 PM To: "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>> Subject: [HCP-Users] fMRI time series intensity normalization Hi, I am using the FIX fMRI dataset. However I have a question about the time series preprocessing: Step 6 in fMRIVolume pipeline said that '6. Intensity normalization to mean of 10000 (like in FEAT) and bias field removal. Brain mask based on FreeSurfer segmentation.'. But the mean values are not 10000, e.g. I checked one subject for which the mean ranges from -115 to 27000. So this becomes an issue when deriving the ROI/regional time series, what I did was first convert to z-score for each time series and then calculate the average across time series within the ROI. I came to recognize that this may not be correct. Another idea is to keep all original data and do the average within the ROIs. However, if one ROI is very big, then the large difference of the baselines (the mean values) of different voxels may make this choice not valid. I wonder why the the mean value is not actually 10000 and what is the correct way to derive the ROI time series. Thanks very much, Best, Chao _______________________________________________ HCP-Users mailing list [email protected]<mailto:[email protected]> http://lists.humanconnectome.org/mailman/listinfo/hcp-users ________________________________ The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. ________________________________ The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. ________________________________ The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. _______________________________________________ HCP-Users mailing list [email protected] http://lists.humanconnectome.org/mailman/listinfo/hcp-users
