Boundary-based registration improves sensitivity for detecting hypoperfusion in sporadic frontotemporal lobar degeneration

Sylvia Mihailescu, Quinn Hlava, Philip A. Cook, Maria Luisa Mandelli, Suzee E. Lee, Bradley F. Boeve, Bradford C. Dickerson, Maria Luisa Gorno‐Tempini, Emily Rogalskı, Murray Grossman, James C. Gee, Corey T. McMillan, Christopher A. OlmFrontiers in Neurology 2024 · DOI

Abstract

Arterial Spin Labeling (ASL) is sensitive to distinct regions of hypoperfusion in patient participants relative to controls, and in patients with sporadic frontotemporal lobar degeneration (sFTLD)-TDP relative to sFTLD-tau. Furthermore, decreasing perfusion is associated with increasing disease severity, particularly in sFTLD-TDP. This study demonstrates that boundary-based registration (BBR) can adequately register ASL-T1w images for both controls and patients, improving sensitivity for detecting regional hypoperfusion compared to traditional methods.

Introduction

Frontotemporal lobar degeneration (FTLD) is a heterogeneous group of neurodegenerative disorders characterized by progressive atrophy of the frontal and temporal lobes. Pathologically, FTLD is most commonly associated with the accumulation of either tau or TDP-43 proteins. Distinguishing these subtypes in vivo remains a significant challenge, as clinical phenotypes often overlap. Arterial Spin Labeling (ASL) MRI offers a non-invasive method to measure cerebral blood flow (CBF), which may serve as a "leading indicator" of neurodegeneration, often showing changes before significant structural atrophy occurs.

However, ASL data typically suffers from low signal-to-noise ratios and low spatial resolution. A critical technical hurdle in ASL analysis is the precise registration of low-resolution functional maps to high-resolution structural T1-weighted (T1w) images. Traditional manual registration is labor-intensive and subjective, while standard automated rigid-body methods often fail to account for geometric distortions. This study evaluates the efficacy of an automated boundary-based registration (BBR) technique in improving alignment and sensitivity for detecting disease-specific hypoperfusion.

Methods and Study Population

The researchers utilized data from the Connectomic Imaging in Familial and Sporadic Frontotemporal Degeneration project (FTDHCP). The study compared three distinct registration approaches:

  1. Manual Registration: A benchmark where a human expert overlays the images.
  2. Manual + BBR: Manual initialization followed by BBR refinement.
  3. FLIRT + BBR: A fully automated pipeline using FMRIB’s Linear Image Registration Tool (FLIRT) followed by BBR.

BBR is a non-rigid transformation method that focuses on the boundary between white and gray matter, allowing for more degrees of freedom to correct for ASL-specific distortions. The ASLPrep pipeline was used for preprocessing, and alignment quality was quantified using Pearson’s Correlation Coefficients (CC) across 219 cortical regions.

Figure 1: Participant inclusion flowchart detailing the selection of controls and sFTLD subtypes.

The final cohort consisted of 50 cognitively normal controls, 21 patients with likely sFTLD-tau, and 14 patients with likely sFTLD-TDP, all confirmed to be non-genetic through negative mutation testing.

Results

Improved Alignment and Efficiency

Both BBR-supplemented methods (Manual+BBR and FLIRT+BBR) achieved significantly higher correlation coefficients (p < 0.001) compared to manual registration alone. This indicates superior anatomical alignment. Furthermore, the automated FLIRT+BBR pipeline reduced processing time from approximately 30 minutes per subject (manual) to 15 minutes, with the added benefit of parallel processing capabilities.

Sensitivity to Pathological Subtypes

ASL-MRI successfully differentiated between sFTLD-tau and sFTLD-TDP. Patients with the TDP subtype exhibited more extensive hypoperfusion in temporal and frontal regions compared to the tau group. Crucially, BBR-based methods demonstrated higher sensitivity in detecting these subtle regional differences.

Figure 6: Voxel-wise comparison of hypoperfusion across registration methods and disease subtypes, highlighting the increased sensitivity of BBR.

Perfusion vs. Atrophy

The study found that hypoperfusion often preceded detectable atrophy in several brain regions, reinforcing the idea that ASL is a leading indicator of neurodegeneration. In the sFTLD-TDP group, lower CBF in the right temporal and orbitofrontal regions correlated significantly with higher disease severity scores (FTLD-CDR), suggesting that ASL-derived markers have direct clinical relevance.

Discussion and Limitations

The implementation of automated BBR represents a significant advancement in ASL-MRI processing for neurodegenerative research. By providing a reliable, high-throughput alternative to manual registration, BBR facilitates larger-scale studies and improves the detection of early physiological changes in FTLD.

A primary limitation of this study is that pathological subtypes were determined by clinical phenotype rather than autopsy confirmation. While these phenotypes are highly predictive, they are not definitive. Additionally, the sFTLD-TDP sample size was relatively small. Future research should aim to validate these findings in pathologically confirmed cohorts and explore the relationship between perfusion changes and white matter integrity using diffusion-weighted imaging.

Summary updated

Abstract

ASL is sensitive to distinct regions of hypoperfusion in patient participants relative to controls, and in patients with sFTLD-TDP relative to sFTLD-tau, and decreasing perfusion is associated with increasing disease severity, at least in sFTLD-TDP. BBR can register ASL-T1w images adequately for controls and patients.
BibTeX
@inproceedings{mihailescu2024boundary,
  author = {Sylvia Mihailescu and Quinn Hlava and Philip A. Cook and Maria Luisa Mandelli and Suzee E. Lee and Bradley F. Boeve and Bradford C. Dickerson and Maria Luisa Gorno‐Tempini and Emily Rogalskı and Murray Grossman and James C. Gee and Corey T. McMillan and Christopher A. Olm},
  title = {{Boundary-based registration improves sensitivity for detecting hypoperfusion in sporadic frontotemporal lobar degeneration}},
  booktitle = {Frontiers in Neurology},
  year = {2024},
  doi = {10.3389/fneur.2024.1452944},
}
Figure