Christopher G. Schwarz
@chrisgschwarz.bsky.social
Changing the face of neuroimaging research
The Mayo Clinic Study of Aging have de-identified, de-faced, and released another 3 years of data, making a total of >13 years of longitudinal imaging data with >14,000 images and >25,000 observations/visits available for qualified researchers to download: www.gaaindata.org/partner/MCSA
www.gaaindata.org
October 22, 2025 at 2:21 PM
The Mayo Clinic Study of Aging have de-identified, de-faced, and released another 3 years of data, making a total of >13 years of longitudinal imaging data with >14,000 images and >25,000 observations/visits available for qualified researchers to download: www.gaaindata.org/partner/MCSA
New: www.sciencedirect.com/science/arti.... We found free, open-source face recognition can reidentify non-defaced brain MRI with up to 59% accuracy, vs. two commercial packages at 92% and 98%. This demonstrates re-identification feasibility across multiple and free face rec programs.
Measuring the Potential Risk of Re-identification of Imaging Research Participants from Open-Source Automated Face Recognition Software
In recent years facial recognition software has gone from an area of research to widespread adoption and broad public availability. Open-source face r…
www.sciencedirect.com
September 23, 2025 at 2:28 PM
New: www.sciencedirect.com/science/arti.... We found free, open-source face recognition can reidentify non-defaced brain MRI with up to 59% accuracy, vs. two commercial packages at 92% and 98%. This demonstrates re-identification feasibility across multiple and free face rec programs.
Do you need help using SCAN imaging data in your research? Join us today (1pm ET, 10am PT) at SCAN office hours, where Tessa Harrison and will I try to answer all your questions. SCAN is an open dataset with >10000 MRI and PET studies from ADRCs across the US acquired with standardized protocols.
June 3, 2025 at 1:14 PM
Do you need help using SCAN imaging data in your research? Join us today (1pm ET, 10am PT) at SCAN office hours, where Tessa Harrison and will I try to answer all your questions. SCAN is an open dataset with >10000 MRI and PET studies from ADRCs across the US acquired with standardized protocols.
We have released MCALT 1.5, adding functionality to measure regional SUVR for input PET files, jointly with T1w MRI already processed with MCALT. It also adds the TauTemporalMeta atlas, which are the exact regions used in various Jack et al. tau PET publications. www.nitrc.org/projects/mca...
NITRC: Mayo Clinic Adult Lifespan Template and Atlases: Tool/Resource Info
www.nitrc.org
April 29, 2025 at 9:06 PM
We have released MCALT 1.5, adding functionality to measure regional SUVR for input PET files, jointly with T1w MRI already processed with MCALT. It also adds the TauTemporalMeta atlas, which are the exact regions used in various Jack et al. tau PET publications. www.nitrc.org/projects/mca...
Are you at #ADPD2025? Come and see me today or tomorrow at poster 391, to hear about the clinical and imaging data we've shared from MCSA on GAAIN, or whatever else you are interested in!
April 4, 2025 at 7:25 AM
Are you at #ADPD2025? Come and see me today or tomorrow at poster 391, to hear about the clinical and imaging data we've shared from MCSA on GAAIN, or whatever else you are interested in!
Another reason for technologies like de-facing/re-facing: data that are more thoroughly de-identified can be shared more widely and with fewer restrictions, vs. the "direct request" peer-to-peer model, which takes an average of 8 months(!), according to this important work.
New paper in Imaging Neuroscience by Anita S. Jwa, Martin Norgaard, and Russell A. Poldrack:
Can I have your data? Recommendations and practical tips for sharing neuroimaging data upon a direct personal request
doi.org/10.1162/imag...
Can I have your data? Recommendations and practical tips for sharing neuroimaging data upon a direct personal request
doi.org/10.1162/imag...
March 20, 2025 at 1:31 PM
Another reason for technologies like de-facing/re-facing: data that are more thoroughly de-identified can be shared more widely and with fewer restrictions, vs. the "direct request" peer-to-peer model, which takes an average of 8 months(!), according to this important work.
MCSA have uploaded another 4877 de-identified, de-faced follow-up MRI and PET. There are now >10 years of longitudinal imaging, with >12,000 images (T1, FLAIR, DTI, PiB PET) from >1800 participants, available for all qualified researchers to download from GAAIN.org. www.gaaindata.org/partner/MCSA
www.gaaindata.org
March 14, 2025 at 4:01 PM
MCSA have uploaded another 4877 de-identified, de-faced follow-up MRI and PET. There are now >10 years of longitudinal imaging, with >12,000 images (T1, FLAIR, DTI, PiB PET) from >1800 participants, available for all qualified researchers to download from GAAIN.org. www.gaaindata.org/partner/MCSA
Reposted by Christopher G. Schwarz
Fonov and Collins investigate optimal sample sizes for creating representative average anatomical brain templates: doi.org/10.52294/001...
@fmri-today.bsky.social @mallarchak.bsky.social @ohbmofficial.bsky.social
@fmri-today.bsky.social @mallarchak.bsky.social @ohbmofficial.bsky.social
March 10, 2025 at 3:58 PM
Fonov and Collins investigate optimal sample sizes for creating representative average anatomical brain templates: doi.org/10.52294/001...
@fmri-today.bsky.social @mallarchak.bsky.social @ohbmofficial.bsky.social
@fmri-today.bsky.social @mallarchak.bsky.social @ohbmofficial.bsky.social
As everyone submits your AAIC abstracts, remember to pay attention to the filenames of your figure and table uploads. Unless they changed it from previous years, reviewers can see those. I have seen a lot of author names leak through like "LastName_Figure1.png".
January 27, 2025 at 5:27 PM
As everyone submits your AAIC abstracts, remember to pay attention to the filenames of your figure and table uploads. Unless they changed it from previous years, reviewers can see those. I have seen a lot of author names leak through like "LastName_Figure1.png".
Reposted by Christopher G. Schwarz
Happy 30th birthday year to SPM!! 🎈🍰
To get the party started, we are delighted to announce SPM 25.01, a major new version of the open-source neuroimaging software.
Read the preprint to discover what's new arxiv.org/pdf/2501.12081 and download SPM for free from Github github.com/spm/spm/rele...
To get the party started, we are delighted to announce SPM 25.01, a major new version of the open-source neuroimaging software.
Read the preprint to discover what's new arxiv.org/pdf/2501.12081 and download SPM for free from Github github.com/spm/spm/rele...
arxiv.org
January 24, 2025 at 3:56 PM
Happy 30th birthday year to SPM!! 🎈🍰
To get the party started, we are delighted to announce SPM 25.01, a major new version of the open-source neuroimaging software.
Read the preprint to discover what's new arxiv.org/pdf/2501.12081 and download SPM for free from Github github.com/spm/spm/rele...
To get the party started, we are delighted to announce SPM 25.01, a major new version of the open-source neuroimaging software.
Read the preprint to discover what's new arxiv.org/pdf/2501.12081 and download SPM for free from Github github.com/spm/spm/rele...
Good morning from Human Amyloid Imaging 2025 in beautiful San Juan! You can come see me at poster 79 on Thursday or 130 on Friday, but I am here all conference and always happy to talk about PET quantification or defacing!
January 15, 2025 at 4:05 PM
Good morning from Human Amyloid Imaging 2025 in beautiful San Juan! You can come see me at poster 79 on Thursday or 130 on Friday, but I am here all conference and always happy to talk about PET quantification or defacing!
I am officially seeking a postdoc to work with us on software development for de-facing and PET quantification. Please pass along to any eligible candidates you may know. jobs.mayoclinic.org/job/rocheste...
January 3, 2025 at 4:05 PM
I am officially seeking a postdoc to work with us on software development for de-facing and PET quantification. Please pass along to any eligible candidates you may know. jobs.mayoclinic.org/job/rocheste...
mri_reface 0.3.5 is now released www.nitrc.org/projects/mri.... Major changes include support for ASL and PD MRI, early/full-dynamic PET, NAV4594 and PI2620 PET, a Mac ARM build, and many other requested enhancements/fixes. This is a recommended upgrade that can be applied in-place.
NITRC: mri_reface: Tool/Resource Info
www.nitrc.org
December 12, 2024 at 3:24 PM
mri_reface 0.3.5 is now released www.nitrc.org/projects/mri.... Major changes include support for ASL and PD MRI, early/full-dynamic PET, NAV4594 and PI2620 PET, a Mac ARM build, and many other requested enhancements/fixes. This is a recommended upgrade that can be applied in-place.
In this new study from Giessen, they tested whether 4 brain MRIs could be re-identified using only publicly available face recognition websites. They correctly matched 2 "within minutes", and it required NO domain expertise or special access at all. This really supports the need for de-facing! (1/n)
Re-identification of anonymised MRI head images with publicly available software: investigation of the current risk to patient privacy
Facial recognition software (FRS) has historically been perceived as lacking the capability to identify individuals from cross-sectional medical image…
www.sciencedirect.com
November 27, 2024 at 5:15 PM
In this new study from Giessen, they tested whether 4 brain MRIs could be re-identified using only publicly available face recognition websites. They correctly matched 2 "within minutes", and it required NO domain expertise or special access at all. This really supports the need for de-facing! (1/n)
A nice Bluesky thing: if you search "Feeds" (left column menu) for "OnlyPosts", you can add a custom feed that only shows posts by the accounts you follow, i.e. no likes and no retweets. Sometimes I want to see those, but sometimes I don't, and this is a feature I always wanted from Twitter.
November 26, 2024 at 3:33 PM
A nice Bluesky thing: if you search "Feeds" (left column menu) for "OnlyPosts", you can add a custom feed that only shows posts by the accounts you follow, i.e. no likes and no retweets. Sometimes I want to see those, but sometimes I don't, and this is a feature I always wanted from Twitter.
Our latest paper, describing the history and procedures for de-facing in ADNI, is now published in Alz & Dem:
alz-journals.onlinelibrary.wiley.com/doi/full/10....
This is part of a larger special issue on ADNI.
alz-journals.onlinelibrary.wiley.com/doi/full/10....
This is part of a larger special issue on ADNI.
Implementation and validation of face de‐identification (de‐facing) in ADNI4
INTRODUCTION Recent technological advances have increased the risk that de-identified brain images could be re-identified from face imagery. The Alzheimer's Disease Neuroimaging Initiative (ADNI) is...
alz-journals.onlinelibrary.wiley.com
October 18, 2024 at 4:32 PM
Our latest paper, describing the history and procedures for de-facing in ADNI, is now published in Alz & Dem:
alz-journals.onlinelibrary.wiley.com/doi/full/10....
This is part of a larger special issue on ADNI.
alz-journals.onlinelibrary.wiley.com/doi/full/10....
This is part of a larger special issue on ADNI.
We have uploaded ~1800 de-identified and de-faced Amyloid (PiB) PET studies to the Mayo Clinic Study of Aging shared dataset posted on GAAIN www.gaaindata.org/partner/MCSA. These join the ~1800 MRI studies already available. Follow-up imaging visits are coming soon. #datasharing #openscience
www.gaaindata.org
September 25, 2024 at 2:27 PM
We have uploaded ~1800 de-identified and de-faced Amyloid (PiB) PET studies to the Mayo Clinic Study of Aging shared dataset posted on GAAIN www.gaaindata.org/partner/MCSA. These join the ~1800 MRI studies already available. Follow-up imaging visits are coming soon. #datasharing #openscience
The Mayo Clinic Study of Aging have now uploaded 3588 MRIs (de-faced, of course!) and richly characterized, longitudinal clinical and demographic data from over 22,000 participants for public sharing on GAAIN.org ! This is only the beginning of our huge data sharing initiatives. Check it out!
www.gaaindata.org
April 5, 2024 at 10:21 PM
The Mayo Clinic Study of Aging have now uploaded 3588 MRIs (de-faced, of course!) and richly characterized, longitudinal clinical and demographic data from over 22,000 participants for public sharing on GAAIN.org ! This is only the beginning of our huge data sharing initiatives. Check it out!
Anyone at AD/PD? Visit my talk on (the expected lack of) effects of mri_reface on FDG markers of DLB on Thursday, 1:50pm session on advances in diagnosis and drug development, auditorium iii.
March 7, 2024 at 11:48 AM
Anyone at AD/PD? Visit my talk on (the expected lack of) effects of mri_reface on FDG markers of DLB on Thursday, 1:50pm session on advances in diagnosis and drug development, auditorium iii.
We've released mri_reface 0.3.3 with minor enhancements and bugfixes. This fixes behavior when input .nii files have inconsistent sform vs qform geometry, automatically splits dynamic PET images to allow registration across frames, and applies some minor bugfixes to nii2dicom.
NITRC: mri_reface: Tool/Resource Info
www.nitrc.org
November 20, 2023 at 8:52 PM
We've released mri_reface 0.3.3 with minor enhancements and bugfixes. This fixes behavior when input .nii files have inconsistent sform vs qform geometry, automatically splits dynamic PET images to allow registration across frames, and applies some minor bugfixes to nii2dicom.