We have been trying to process the PPMI structural and diffusion MRI data, and found matching different MRI sessions of the same subject to not always be so straightforward. For instance, to correct for susceptibility-induced distortions in diffusion MR images (via FSL’s topup), there doesn’t seem to be a reference linking the images with opposite phase encoding directions, and using the image names is not always reliable (besides being labor-intensive). Similarly, corresponding diffusion images to the correct same-session structural T1-weighted image is not so straightforward due to the lack of necessary image-corresponding information (we are using DICOM date/time stamps). Is there a reference, such as a CSV table, providing these pieces of information?
Hello Iman. Unfortunately, I’m not knowledgeable in this topic to provide you with an straightforward answer on how to deal with your specific problem, BUT, I know that the PPMI team is working on providing this type of MRI-extracted data inside LONI, and maybe they will someday provide the types of imaging you are interested in?
The reason I know this is because my PhD uses structural MRI data extracted from PPMI and we wanted to provide this data to be included inside LONI, as we’ve discovered it wasn’t there, then we talked with the researcher responsible for that, which is Barbara Marebwa (an Associate Director at MJFF), and she explained that they have their own team working on that and will be implementing this sometime soon.
Nonetheless, I hope that a neuroimaging expert is able to answer your specific question here.
Thank you, Daniel. I actually discussed this with Barbara a few weeks ago, and she did mention that the PPMI curating team is working on providing a comprehensive dataset. I am eager to see what that dataset will contain!
Hi Iman, great to see you in the community and thank you for your question!
I’m wondering if @AmgadDroby would have any insight regarding structured metadata enabling the matching of structural and diffusion MRI data in PPMI?
We also have some new members in our community with imaging expertise from the MJFF MRI Working Group (@martab , @whiter , @LauraJonkman) who may be able to provide updates/guidance!
The DTI sequence in PPMI has undergone multiple iterations and has been modified over time. Additionally, there may be some degree of variation between centers, depending on the site from which the data is being used.
Check for Metadata Documentation: PPMI should have an associated metadata document, often in CSV format, that provides image acquisition details, including scan type, session, and phase encoding information. Search the PPMI data repository for such a file, or reach out to their data team for clarification. This could significantly streamline your session matching.
Automated DICOM Parsing: Instead of manually checking DICOM timestamps, consider using a DICOM parsing tool that can extract and match relevant tags (like SeriesInstanceUID, StudyDate, etc.) to automate the matching process. Tools like dcm2niix or Python packages such as pydicom can assist in this.
SL provides tools like fslmerge and topup that expect phase encoding direction to be specified. Look for a PPMI-specific protocol or pipeline for processing dMRI in FSL. Alternatively, investigate if the BIDS (Brain Imaging Data Structure) format has been implemented in the dataset, as it standardizes how MRI data are organized.
If no obvious reference table exists and you are still struggling, it’s worth contacting PPMI’s support team. They may have an internal solution or more guidance on how to match the MRI images across sessions.
I hope this is helpful, and I’d be glad to hear more details to assist you further in addressing your specific challenge.
