Computing Univariate Neurodegenerative Biomarkers with Volumetric Optimal Transportation: A Pilot Study

Area of Science:

• Neuroimaging and computational anatomy
• Biomarker development for neurodegenerative diseases

Background:

• Cognitive decline in Alzheimer's disease (AD) correlates with brain structure changes.
• Need for robust, univariate neuroimaging biomarkers for clinical diagnosis and prognosis.
• Existing biomarkers are limited in noise robustness and clinical applicability.

Purpose of the Study:

• Introduce a variational framework for optimal transportation (OT) on brain MRI.
• Develop a novel univariate neuroimaging index, the Wasserstein Index (WI), to quantify neurodegeneration.
• Assess WI's potential as a biomarker for Alzheimer's disease.

Main Methods:

• Computed optimal transportation (OT) from individual MRI volumes to a template.
• Quantified neurodegeneration using the Wasserstein distance (WD), termed Wasserstein Index (WI).
• Employed Newton's method for efficient WI computation in large datasets.
• Validated on 314 subjects (140 AD, 174 controls) from the Alzheimer's Disease Neuroimaging Initiative (ADNI).

Main Results:

• The proposed Wasserstein Index (WI) is robust to image noise and informative.
• WI demonstrated significant correlation with the Mini-Mental State Examination (MMSE) cognitive score.
• WI effectively identified group differences and achieved good classification accuracy for AD.
• WI outperformed hippocampal volume and entorhinal cortex thickness as a univariate biomarker.

Conclusions:

• The Wasserstein Index (WI) shows potential as a powerful univariate biomarker for neurodegenerative alterations.
• WI's noise robustness and computational efficiency support its clinical applicability.
• Further research is warranted to establish WI as a clinical tool for Alzheimer's disease.