As many of you already know, we have on-line within the BIC a data base of 150 normal subjects that were scanned here at the MNI. Many of you use all or some of this data, so the following message is very important because IT WILL AFFECT YOUR ANALYSIS.
For each subject, the data base contains the original T1, T2 and PD data sets in 'native' space and the final T1, T2 and PD registered together in stereotaxic space. In addition, there are classified data (grey-matter, white-matter and CSF), surface objects extracted from the final T1 data sets that model the cortical surface, non-linear deformation fields, automatic structure segmentations, and automatic sulcal segmentations.
We are going to update the data base shortly in order to address a basic problem when comparing two groups of subjects (e.g., the normal ICBM group and a 2nd group from a patient population). Comparisons must be done not only with MRI data acquired with the same protocol, but also with exactly the same analysis software. Otherwise it is not possible to determine whether any group differences are due to software differences or to real biological differences. This has not been strictly correct in the past, since the dozens of software programs used in the analysis have been updated in a uncoordinated fashion. Thus it has been impossible to reproduce exactly an analysis done in the past. While differences observed would be small, we could not guarantee that the each of the programs used in an analysis remained stable after the completion of that analysis.
Over the past few months, we have been re-designing and re-implementing many of the programs required to generate the ICBM data base. Now, software and data base updates will be done in a very controlled fashion. In particular, we have built a "software quarantine" so that we can re-create the exact execution environment used to produce any new version of the ICBM data base. This will allow programs that are used on a daily basis to be updated when bugs are found and fixed.
N.B. This means that the daily execution environment will be exactly as it is now. However, when data from a particular project is to be compared with the ICBM data base, the "quarantined software" execution environment should be used for analysis. We will describe how this is to be done later.
We are now ready to implement the quarantine. This will involve replacement of the current ICBM data base with its first quarantined version.
Since this data base is used by many researchers (for example, many projects involve painting anatomical structures on the T1-weighted stereotaxic images), we have made every effort to ensure that any changes to the data base have a minimal impact on you, the users. However, improvements to the many diffferent algorithms used to generate all of the derived data (ie, the final stereotaxic T1-weighted MRI volume) has changed the data slightly and therefore will affect any analysis using the ICBM MRI data base. The main consequence of these changes is that it is not possible to directly compare the following between old and new data bases:
The good news is that is is not necessary to throw away all painted structures and start again from scratch. The changes in the transformations to stereotaxic space are on the order of 1-2 pixels at the cortex, so this means that it is ONLY NECESSARY to VALIDATE and VERIFY any painting already done with the new T1-weighted volumes by loading up your existing painted volumes and verifying them with Display.
The existing database will remain on-line is its current location until December 10th, 1997 (/data/icbm/mri1/subjects). Afterwards, this data base will be REMOVED.
The new database is accessible as of TODAY (November 28) from /data/icbm/mri1/new_icbm/subjects. On December 17, the new data base will be moved to the original data base location in /data/icbm/mri1/subjects.
It is strongly suggested that you stop all work on the old data base and move to the new one immediately. Remember that the HBM98 abstract deadline is Jan 31, 1998 - so it is imparitive that you plan for this data base change.