Related Experiment Video
Updated: Aug 31, 2026
Whole-brain Segmentation and Change-point Analysis of Anatomical Brain MRI—Application in Premanifest Huntington's Disease
Published on: June 9, 2018
Event-based modeling of T1-weighted MRI is related to pathology in frontotemporal lobar degeneration due to tau and
Christopher A Olm1, Sarah E Burke2, Claire Peterson3
1Penn Image Computing and Science Laboratory, Department of Radiology, University of Pennsylvania, Philadelphia, PA, United States; Penn Frontotemporal Degeneration Center, Department of Neurology, University of Pennsylvania, Philadelphia, PA, United States.
Background:
In previous studies of patients with frontotemporal lobar degeneration due to tau (FTLD-tau) and FTLD due to TDP (FTLD-TDP), cortical volumes derived from T1-weighted MRI have been used to identify a sequence of volume loss according to arbitrary volumetric criteria. Event-based modeling (EBM) is a probabilistic, generative machine learning model that determines the characteristic sequence of changes, or "events", occurring during disease progression. EBM also estimates an individual patient's disease "stage" by identifying which events have already occurred. In the present study, we use an EBM analysis to derive stages of regional anatomic atrophy in FTLD-tau and FTLD-TDP, and validated these stages against pathologic burden.
Methods:
Sporadic autopsy-confirmed patients with FTLD-tau (N = 42) and FTLD-TDP (N = 21), and 167 healthy controls with available T1-weighted images were identified. A subset of patients had quantitative digital histopathology of cortex performed at autopsy (FTLD-tau = 30, FTLD-TDP = 17). MRI images were processed, producing regional measures of cortical volumes. K-means clustering was used to find cortical regions with similar amounts of GM volume changes (n = 5 clusters). EBM was used to determine the characteristic sequence of cortical atrophy of identified clusters in autopsy-confirmed FTLD-tau and FTLD-TDP, and estimate each patient's disease stage by cortical volume biomarkers. Linear regressions related pathologic burden to EBM-estimated disease stages.
Results:
EBM for cortical volume biomarkers generated statistically robust characteristic sequences of cortical atrophy in each group of patients. Cortical volume-based EBM-estimated disease stage was associated with pathologic burden in FTLD-tau (R2 = 0.16, p = 0.017) and FTLD-TDP (R2 = 0.51, p = 0.0008).
Conclusions:
We provide evidence that EBM can identify sequences of pathologically-confirmed cortical atrophy in sporadic FTLD-tau and FTLD-TDP.
Insights
Event-based modeling (EBM) successfully identified characteristic atrophy sequences in frontotemporal lobar degeneration (FTLD-tau) and FTLD-TDP. EBM disease staging correlated with pathologic burden, validating its use in these neurodegenerative diseases.
Area of Science:
- Neuroimaging and Machine Learning in Neurodegenerative Diseases
- Computational Neuroscience and Disease Progression Modeling
Background:
- Previous studies on frontotemporal lobar degeneration (FTLD-tau and FTLD-TDP) used arbitrary volumetric criteria from MRI to assess atrophy sequences.
- Event-based modeling (EBM) is a machine learning approach that identifies characteristic disease progression sequences and estimates individual disease stages.
Purpose of the Study:
- To apply EBM to derive stages of regional anatomic atrophy in FTLD-tau and FTLD-TDP.
- To validate EBM-derived disease stages against post-mortem pathologic burden in these FTLD subtypes.
Main Methods:
- Analyzed T1-weighted MRI from autopsy-confirmed FTLD-tau (N=42) and FTLD-TDP (N=21) patients, and 167 controls.
- Used k-means clustering to identify 5 clusters of cortical regions with similar gray matter volume changes.
- Applied EBM to determine atrophy sequences and estimate disease stages, then correlated these with digital histopathology data (FTLD-tau N=30, FTLD-TDP N=17).
Main Results:
- EBM generated statistically robust characteristic sequences of cortical atrophy for both FTLD-tau and FTLD-TDP.
- EBM-estimated disease stage based on cortical volume showed significant association with pathologic burden in FTLD-tau (R²=0.16, p=0.017).
- A stronger association was found between EBM-estimated disease stage and pathologic burden in FTLD-TDP (R²=0.51, p=0.0008).
Conclusions:
- This study provides evidence that EBM can effectively identify pathological sequences of cortical atrophy in sporadic FTLD-tau and FTLD-TDP.
- EBM offers a validated method for staging disease progression based on neuroimaging biomarkers in FTLD.
Related Concept Videos
01:24Magnetic Resonance Imaging
01:14Brain Imaging
These technologies include computerized axial tomography (CAT or CT scans), positron-emission tomography (PET scans), magnetic resonance imaging (MRI), functional magnetic resonance imaging (fMRI), and Transcranial Magnetic Stimulation (TMS).