Mbirn: Support structural and functional MRI Calibration of multi-site stud

From NAMIC Wiki
Jump to: navigation, search
Home < Mbirn: Support structural and functional MRI Calibration of multi-site stud

Participants:

  • BWH/GCRC Principal Investigator: Reisa Sperling, MD, PhD
  • BWH/BIRN: Steve Pieper, PhD, Kelly Zou, PhD, Sandy Wells, PhD
  • MGH/BIRN: Doug Greve, PhD, Brad Dickerson, MD

Goal:
To validate functional MRI for potential use in clinical trials (and early diagnosis) for Alzheimer's Disease. The first task is to demonstrate test-retest reliability and cross-platform (GE and Siemens) reproducibility, and then to do a longitudinal study to examine the natural history of fMRI activation patterns in normal aging and early Alzheimer's subjects. The functional piece is primary, but it is also necessary to study comparable structural data, since anatomic regions of interest are used for the functional data, so there is a need to compare activation patterns with rates of atrophy using the structural data.


The problems fall into three categories:

  • Test-retest reliability. A group consisting of Steve Pieper, Doug Greve, Brad Dickerson and Sandy Wells is addressing this. This group has started meeting to go over the technical issues. The question is can the fBIRN Reliability analysis recently accepted for publication in Radiology be applied on a smaller scale for this study. Initial meetings have been inconclusive on this point, though there is a lot of benefit to discussing the ways in which the fBIRN reliability methods can transition to a real clinical setting. [How is this relevant?: Doug Greve has been analyzing subjects matching of name-face pairs, displayed both repetitively and non-repetitively in the scanner. He has analyzed the pilot data and generated power curves. "Power" in this context means the ability to detect an effect for a given number of subjects at a given threshold.]
  • Calibration effects between GE and Siemens scanners. Jorge Jovicich is interested in helping with the reproducibility issues with structural scans, and is gathering information on which protocols will be used on the two different scanners as well as on getting sample data. Ongoing work at MGH with 3T protocols for T1-weighted images that give good morphometry may be of help. He has suggested the following option: Have short (30 min) 1.5T structural sessions for morphometry/surface generation that you then use on the 3T fMRI. This is a little cumbersome because one needs two scans, but it will yield more fMRI time (more runs to choose from in the event of motion effects), and ensure cortical and subcortical morphometry data that you can relate to many other 1.5T studies. Also, Reisa has measured the T2* for the same phantom at both sites and come up with the same number, which is very encouraging. She is still working on some other measures, that will be done by this week. There are plans to investigate differences relative to other scanning parameters, and to also test for smoothness at a zero flip angle.
  • File formatting (Jonathan Sacks). There is quite a long list of file system incompatibilities which have made working with images from the two sites difficult: file format/different FOV/wrong slice thickness/slice orientation in reverse order etc. So far Doug Greve and others have corrected these manually at MGH. There would be much benefit to storing the files in the BIRN XML structure. The actual file format itself could automatically be converted to the desired one as part of the upload to the XML structure. The other conditions could be trapped automatically at upload time and either marked as different in the XML structure or fixed manually, if possible, and then uploaded. In summary, the XML structure provides an environment where image files can be uniformly categorized and standardized where possible.

Note on funding: The expectation is that funding will be obtained through the R01 mechanism. However, ample credit would be give to BIRN, and in particular, this would be treated as a BIRN/GCRC prototype. There is strong consensus amongst the investigators that this is very much the kind of project that will benefit from BIRN input.
There has recently been a site visit that resulted in very positive feedback, and the fMRI project underpinning the preliminary data for the proposed R01 project was rated outstanding.