LASHiS

LASHiS performs longitudinal automatic segmentation of hippocampal subfields from in vivo MRI to enable volumetric and morphometric analysis of hippocampal subregions across time.

Key Features:

• Longitudinal segmentation pipeline: Automatically segments hippocampal formation subfields across multiple time points to support longitudinal analyses.
• Joint-label fusion label propagation: Propagates labels from cross-sectionally labeled time point scans using joint-label fusion.
• Single subject template (SST) alignment: Applies joint-label fusion to a non-linearly realigned single subject template to obtain unbiased segmentation across individual time points.
• Multi-contrast segmentation: Utilizes multi-contrast MRI information during the segmentation process.
• Multi-field strength support: Processes high-resolution MRI acquired at 3 Tesla and 7 Tesla.
• Volumetric and morphometric outputs: Produces volumetric and morphometric measures for hippocampal subfields.
• Performance validation: Evaluated against three other approaches using volume similarity coefficients, Dice coefficients, and longitudinal Bayesian Linear Mixed Effects analyses.
• Test-retest reliability: Demonstrates improved test-retest reliability as evidenced by higher volume similarity and Dice coefficients.
• ANTs-based foundation: Built upon adaptations from the ANTs Longitudinal Cortical Thickness pipeline.
• Image orientation requirement: Requires z-oriented image slice direction (e.g., dimensions like 400x400x30).
• Reduced inter-subject variability: Enhances sensitivity in longitudinal studies by reducing inter-subject variability and increasing effect-sensitivity relative to cross-sectional designs.

Scientific Applications:

• Biomarker identification: Identification of sensitive biomarkers for neuropsychiatric disorders and Alzheimer's disease via subfield-specific volumetric and morphometric changes.
• Hippocampal subregion studies: Investigation of functional specialization and region-specific changes within hippocampal subfields.
• Longitudinal cohort analysis: Tracking hippocampal atrophy and morphological change across repeated in vivo MRI acquisitions.
• Method benchmarking: Comparative evaluation of segmentation approaches and assessment of test-retest reliability.

Methodology:

LASHiS propagates labels from cross-sectionally labeled time point scans via joint-label fusion applied to a non-linearly realigned single subject template, utilizes multi-contrast MRI information, and incorporates adaptations from the ANTs Longitudinal Cortical Thickness pipeline; it requires z-oriented image slice direction (e.g., 400x400x30).