Hello ADNI experts. I am working with ADNI data longitudinally and I would like to receive advice regarding how to better estimate the dates of different assessments. I wanted to rely on the REGISTRY csv file, but it has inconsistencies (just to give an example, there is one ADNI2 subject that had visit m30 seven months before m24). Many of the assessments do not have EXAMDATE as well, and some VISCODE2 span assessments years apart. How to better proceed here? Thanks!
Month 30 is not a regularly scheduled clinic visit, so you may want to check to see if there are any actual measurements you are interested in linked to that visit. We generally recommend using the dates in the REGISTRY file as the exam dates, especially for the files that do not have the EXAMDATE specifically entered. VISCODE is what the site entered as the specific visit the information was being collected at, while VISCODE2 is derived to translate VISCODE into something more meaningful. Using the dates to estimate the time between visits (or from the initial visit) is recommended for longitudinal analysis, since there is a lot of variability across participants of when specific visits took place.
For the PET scans, the scan date is always in the DICOM headers. You can reliably calculate the longitudinal intervals from these. The scan dates also appear on the LONI website for every scan, whether you are looking at a list of original scans, or of the "pre-processed" scans.
