MRI-Based Multi-Class Relevance Vector Machine Classification of Neurodegenerative Diseases
Poster No:
1128
Submission Type:
Abstract Submission
Authors:
Yann Cobigo1, Kyan Younes2, John Kornak3, Katherine Rankin3, Adam Staffaroni3, Faizal Beg4, Lei Wang5, Howard Rosen6
Institutions:
1University of California, San Francisco, San Francisco, CA, 2Stanford University School of Medicine, Stanford, CA, 3University of California San Francisco, San Francisco, CA, 4Simon Fraser University, Burnaby, BC, 5Ohio State University, Colombus, OH, 6Weill Institute for Neurosciences, University of California San Francisco, San Francisco, CA
First Author:
Co-Author(s):
Howard Rosen, MD
Weill Institute for Neurosciences, University of California San Francisco
San Francisco, CA
Weill Institute for Neurosciences, University of California San Francisco
San Francisco, CA
Introduction:
The increasing need for dementia experts and the development of MRI-based machine learning analyses have led to a growing interest in using MRI scans to classify neurodegenerative diseases. However, misclassification remains a significant challenge due to disease heterogeneity and overlapping symptoms (McCarthy, 2018) and (Rathore, 2017). We aim to investigate the efficacy of Relevance Vector Machine (RVM)-based (Tipping, 1999) multi-class classification models combined with logistic regression using MRI data to improve differential diagnostic classification accuracy across seven diagnostic categories, including healthy controls (CO), Alzheimer's disease (AD), behavioral variant FTD (bvFTD), primary progressive aphasia (PPA) variants, corticobasal syndrome (CBS), and progressive supranuclear palsy syndrome (PSP).
Methods:
We analyzed structural MPRAGE T1-weighted MRI data scans from 468 participants, including 44 CO, 84 AD, 108 bvFTD, 30 semantic variant PPA (svPPA), 30 non-fluent PPA (nfvPPA), 30 CBS, and 42 PSPS. We performed dimensionality reduction techniques and multi-class extension by pairwise coupling (Hastie, 1997) of a nested 3-fold cross-validation RVM classification to predict clinical diagnoses. A cross-validated logistic regression was applied to optimize probability distributions. Diagnostic accuracy was validated against clinical and genetic data, and cases that had pathological diagnosis (n=72) to identify misclassification patterns and assess algorithm robustness.
Results:
Average (SD) diagnostic accuracy across the outer validation loop for CO and svPPA was 0.90(0.06) and 0.84(0.18) respectively, and performance was moderate for AD and bvFTD, 0.70(0.08) and 0.71(0.09), respectively. Challenges emerged in classifying rarer conditions like nfvPPA, CBS and PSPS with 0.30(0.15), 0.43(0.27) and 0.60(0.10) respectively. Across all disorders, misclassified cases often exhibited minimal brain atrophy or had diagnosis of early-onset AD or genetic FTD.
·Figure-1: Lowest mean W-score region of interest of the brain for each subject in the entire cohort by RVM performance.
·Table-1: Confusion matrices comparing the efficiency of probability interpretation schemes. Pair-wise coupling, multiclass classification confusion matrix corrected using the logistic regression.
Conclusions:
We demonstrated that RVM-based classification algorithm was effective in classifying neurodegenerative diseases based on MRI scans. Misclassification stemmed from disease heterogeneity and limitations of structural MRI in capturing the full spectrum of biological changes, especially in rarer diseases. Larger and more inclusive datasets are needed to train better machine learning techniques, and a margin of error should be accepted.
Disorders of the Nervous System:
Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) 2
Genetics:
Genetics Other
Modeling and Analysis Methods:
Classification and Predictive Modeling 1
Keywords:
Degenerative Disease
Machine Learning
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Are you Internal Review Board (IRB) certified? Please note: Failure to have IRB, if applicable will lead to automatic rejection of abstract.
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
For human MRI, what field strength scanner do you use?
Which processing packages did you use for your study?
Provide references using APA citation style.
McCarthy J., Collins D. L., Ducharme S. (2018), Morphometric MRI as a diagnostic biomarker of frontotemporal dementia: A systematic review to determine clinical applicability, Neuroimage: Clinical.
Rathore S., Habes M., Iftikhar M. A., Shacklett A., Davatzikos C. (2017), A review on neuroimaging-based classification studies and associated feature extraction methods for Alzheimer's disease and its prodromal stages, Neuroimage.
Tipping M. E. (1999), The Relevance Vector Machine, proceeding