Related Experiment Video
Updated: Jun 22, 2026
09:06
Whole-brain Segmentation and Change-point Analysis of Anatomical Brain MRI—Application in Premanifest Huntington's Disease
Published on: June 9, 2018
Automated MRI-based classification of primary progressive aphasia variants
Stephen M Wilson1, Jennifer M Ogar, Victor Laluz
1Memory and Aging Center, Department of Neurology, University of California, San Francisco, CA, USA. swilson@memory.ucsf.edu
Neuroimage
|June 9, 2009
Summary
Automated MRI analysis accurately differentiates primary progressive aphasia (PPA) variants. Combining imaging and linguistic data improves diagnostic accuracy for progressive non-fluent aphasia and semantic dementia.
Area of Science:
- Neurology
- Neuroimaging
- Computational Neuroscience
Background:
- Primary progressive aphasia (PPA) is a neurodegenerative syndrome affecting speech and language.
- PPA presents with three main variants: progressive non-fluent aphasia (PNFA), semantic dementia (SD), and logopenic progressive aphasia (LPA).
- Accurate differential diagnosis is crucial for targeted therapies, but clinical evaluation and visual MRI interpretation can be challenging.
Purpose of the Study:
- To assess the efficacy of automated structural MRI analysis in differentiating PPA variants.
- To compare the diagnostic performance of imaging-only models versus combined imaging and linguistic models.
Main Methods:
- Structural T1 MRI scans from 86 PPA patients and 115 controls were analyzed.
- Grey matter images underwent principal component analysis (PCA) for feature extraction.
- Linear support vector machines classified PC coefficients, with cross-validation determining accuracy.
Main Results:
- Automated analysis achieved an overall mean accuracy of 92.2% in discriminating between PPA variants and controls.
- Models incorporating both MRI imaging and linguistic features demonstrated superior performance compared to imaging-only or linguistic-only models for PNFA and SD differentiation.
Conclusions:
- Automated structural MRI analysis shows significant potential for assisting in the differential diagnosis of PPA variants.
- Integrating neuroimaging with linguistic data may enhance diagnostic precision, facilitating etiology-specific treatment strategies.
