HAI Book 2025 - Flipbook - Page 169
DelBene, Alexander
29
Performance of an MR-independent analysis pipeline for
quantification of tau abnormality using [18F]flortaucipir PET
Alexander DelBene1, Alexandra Gogola1, Alexandria Reese1, Cristy Matan1, Ming-Hsuan Chiang1,
Beth Snitz2, Ann Cohen3, Oscar Lopez2, Victor Villemagne3, Brian Lopresti1
1
Department of Radiology, University of Pittsburgh School of Medicine, Pittsburgh, PA, US
Department of Neurology, University of Pittsburgh School of Medicine, Pittsburgh, PA, US
3
Department of Psychiatry, University of Pittsburgh School of Medicine, Pittsburgh, PA, US
2
Background: Magnetic resonance (MR) imaging contraindications commonly limit participation in PET
neuroimaging studies because most analysis methods require structural MR. An analysis method that accurately
indexes the level of tau abnormality using only PET image data could broaden participant inclusion. CapAIBL, an
MR-independent method, was compared against SPM-generated CenTauRz (CTRz).
Methods: 150 participants (age 71.1±8.4 years, 44% female, 92% nHW) from the University of Pittsburgh ADRC
underwent 3T MR, amyloid [11C]PiB PET, and tau [18F]flortaucipir (FTP) PET scans. PET images were corrected for
motion and summed (PiB=50-70 min, FTP=80-100 min). FTP scans were processed using the CenTauR masks
through both the SPM and CapAIBL pipelines. Both pipelines were sampled in template space. In the SPM pipeline,
a linear PET to MR transformation is done before normalizing. Meta-Temporal (MT) FTP was expressed in CTRz. A
global CL g20 was used to determine A status for additional evaluation. Linear regressions determined the level of
agreement between the two pipelines.
Results: CapAIBL and SPM pipelines showed a high level of agreement (Figure 1) between MT CTRz scores across
the entire cohort (´=0.933, r2=0.976), and for subsets of A+ participants (´=0.954, r2=0.982, n=70) and Aparticipants (´=0.855, r2=0.927, n=80). Compared to CapAIBL, SPM-pipeline CTRz values showed a small, positive
bias in A- participants (0.08±0.88 CTRz) that grew more pronounced in A+ participants (0.43±0.91 CTRz; Figure 2).
However, comparing T status based on CTRz scores (T+: MT CTRzg2) between pipelines, this bias had a minimal
effect on T classification (97.4% concordance).
Conclusions: MR-less CapAIBL CTRz values show robust agreement with the MR-based SPM pipeline, particularly
when A´ pathology is present. Future work will evaluate the concordance when native-space FTP images are
analyzed through both pipelines, assess additional regions of interest, and determine if the FTP CTRz equations
require updating/refinement.
HAI2025 - 169
