ProMUSE: Progressive Multi-modal Uncertainty-guided Staged Evidential Alzheimer Disease Classification

📅 2026-06-11
📈 Citations: 0
Influential: 0
📄 PDF
🤖 AI Summary
This study addresses the high cost and limited accessibility of multimodal neuroimaging (e.g., MRI/PET) in early Alzheimer’s disease diagnosis by proposing a progressive, uncertainty-guided, staged evidence network. The method first performs an initial assessment using low-cost clinical data and adaptively incorporates imaging modalities only when model uncertainty is high. Innovatively integrating subjective logic with Dempster–Shafer evidence theory, the framework models uncertainty via Dirichlet distributions to enable on-demand modality fusion and staged decision-making. Evaluated on the ADNI, AIBL, and OASIS datasets, the approach achieves diagnostic accuracy comparable to or better than full-modality models while reducing imaging usage by 50%–90%, substantially lowering diagnostic costs.
📝 Abstract
Alzheimer's disease (AD) is a fatal disorder that destroys memory and cognitive skills in the elderly population. Most treatments for AD are effective in the early stage, leading to an increasing demand for early AD diagnosis. AD diagnosis increasingly relies on multimodal data such as clinical assessments, structural Magnetic Resonance Imaging (MRI), and Positron Emission Tomography (PET) imaging. However, MRI and PET acquisition remain costly and not universally accessible, making full-modality inference impractical in real-world clinical workflows. We propose ProMUSE, a Progressive Multi-modal Uncertainty Guided Staged Evidential Network that adaptively determines when additional modalities are necessary, helping reduce the overall cost of data acquisition while maintaining accuracy. ProMUSE first performs evidential classification using low-cost clinical data and quantifies uncertainty via a Dirichlet-based subjective logic model. When uncertainty exceeds a learned threshold, ProMUSE progressively incorporates MRI or PET features, fusing modality-wise belief and uncertainty through Dempster-Shafer theory to obtain a calibrated multimodal prediction. This staged acquisition strategy enables accurate diagnosis while minimizing reliance on expensive imaging. Experiments on ADNI, AIBL, and OASIS across CN-AD, CN-MCI, and MCI-AD tasks demonstrate that ProMUSE achieves competitive or superior accuracy compared to full-modality baselines while reducing MRI/PET usage by 50-90%, yielding substantial cost savings. These results highlight ProMUSE as a practical, uncertainty-aware, and resource-efficient solution for real-world AD screening.
Problem

Research questions and friction points this paper is trying to address.

Alzheimer's disease
early diagnosis
multimodal data
cost-effective
clinical workflow
Innovation

Methods, ideas, or system contributions that make the work stand out.

Progressive multimodal learning
Uncertainty-guided diagnosis
Evidential deep learning
Dempster-Shafer theory
Cost-efficient AD screening
Long Doan
Long Doan
George Mason University
LLMAutomated Reasoning
B
Branden Chen
Department of Computer Science, Kennesaw State University
E
Ethan Litton
Department of Computer Science, Kennesaw State University
H
Huan Huang
Department of Computer Science, Kennesaw State University
J
Jiajing Huang
Department of Data Science and Analytics, Kennesaw State University
Y
Yixin Xie
Department of Information Technology, Kennesaw State University
Weihua Zhou
Weihua Zhou
Michigan Technological University
Medical Imaging and Informatics
N
Nandakumar Narayanan
Department of Neurology, University of Iowa
Chen Zhao
Chen Zhao
Assistant Professor of Computer Science, Kennesaw State University
deep learningmedical image processing