Dual-path mixture of experts for Alzheimer’s diagnosis using volume and entropy MRI features
Journal of Neuroscience Methods · 2026-07 · 431 · 110728
DOI: 10.1016/j.jneumeth.2026.110728 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Neuroscience Methods, 2026-07. Verfasst von Sangyoon Park und 6 weiteren Autor:innen. Beteiligte Einrichtung: Korea University, Department of Biomedical Engineering, 145 Anam-ro, Seongbuk-gu, Seoul, 02841, Republic of Korea. Electronic address: yoonni99@gmail…. Hauptthemen laut MEDLINE: Alzheimer Disease, Magnetic Resonance Imaging, Brain, Image Interpretation, Computer-Assisted. Die Arbeit nennt 38 Literaturangaben. Gefördert von: National Research Foundation of Korea, Korea Medical Device Development Fund.
Themen: Demenz
Abstract (Originalsprache)
BACKGROUND Alzheimer's disease (AD) is a progressive neurodegenerative disorder marked by brain atrophy, observable via MRI. Volumetric measures like hippocampal volume are widely used but often miss early microstructural changes. Texture features such as intensity entropy capture tissue heterogeneity and may improve early diagnosis. Combining these orthogonal modalities could yield better biomarkers. NEW METHOD We propose a Dual-Path Mixture of Experts (DP-MoE) model that integrates volumetric and entropy-based MRI features through two expert subnetworks and a self-attention gating mechanism. A residual fusion branch preserves raw input distributions, enhancing flexibility and expressiveness. The code is available at https://github.com/syoonni/DP-MOE. RESULTS The model was evaluated on three binary classification tasks (CN vs. AD, CN vs. MCI, MCI vs. AD) using T1-weighted and entropy-derived MRI. For CN vs. AD, it achieved an accuracy of 93.02% (0.9302 ± 0.0109) and an AUC of 0.975 ± 0.004, outperforming both the baseline MoE and logistic fusion models. Ablation studies confirmed the importance of both attention and residual components. COMPARISON WITH EXISTING METHODS Prior methods used concatenation or fixed weighting, neglecting modality-specific characteristics. Volume and entropy provide orthogonal, complementary information. While Mixture of Experts is common in other fields, it remains underused in AD imaging. Our adaptive fusion approach improved both interpretability and accuracy, especially in early-stage discrimination. CONCLUSIONS DP-MoE offers an effective, interpretable framework for fusing heterogeneous MRI biomarkers and holds promise for deployment in early Alzheimer's diagnosis workflows.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Korea University, Department of Biomedical Engineering, 145 Anam-ro, Seongbuk-gu, Seoul, 02841, Republic of Korea. Electronic address: yoonni99@gmail.com.
- Korea University, Department of Biomedical Engineering, 145 Anam-ro, Seongbuk-gu, Seoul, 02841, Republic of Korea; FieldCure Co., Ltd., 82-3 Bomun-ro, Seongbuk-gu, Seoul, 02852, Republic of Korea.
- Korea University, Department of Biomedical Engineering, 145 Anam-ro, Seongbuk-gu, Seoul, 02841, Republic of Korea.
- Korea University, Department of Neurology, College of Medicine, 73 Goryeodae-ro, Seongbuk-gu, Seoul, 02841, Republic of Korea.
- FieldCure Co., Ltd., 82-3 Bomun-ro, Seongbuk-gu, Seoul, 02852, Republic of Korea.
- Korea University, Department of Biomedical Engineering, 145 Anam-ro, Seongbuk-gu, Seoul, 02841, Republic of Korea; FieldCure Co., Ltd., 82-3 Bomun-ro, Seongbuk-gu, Seoul, 02852, Republic of Korea. Electronic address: ra…
Förderung
- National Research Foundation of Korea
- Korea Medical Device Development Fund
Zitieren
APA
Park, S., Gi, Y., Lim, H., Kim, B. J., Baek, S. H., Kim, J. H., & Yoon, M. (2026). Dual-path mixture of experts for Alzheimer’s diagnosis using volume and entropy MRI features. Journal of Neuroscience Methods, 431, 110728. https://doi.org/10.1016/j.jneumeth.2026.110728
Vancouver
Park S, Gi Y, Lim H, Kim BJ, Baek SH, Kim JH, et al. Dual-path mixture of experts for Alzheimer’s diagnosis using volume and entropy MRI features. Journal of Neuroscience Methods. 2026;431:110728. doi:10.1016/j.jneumeth.2026.110728
BibTeX
@article{Park2026,
author = {Park, Sangyoon and Gi, Yongha and Lim, Hyeongjin and Kim, Byung-Jo and Baek, Seol-Hee and Kim, Jong Hyun and Yoon, Myonggeun},
title = {Dual-path mixture of experts for Alzheimer’s diagnosis using volume and entropy MRI features},
journal = {Journal of Neuroscience Methods},
year = {2026},
volume = {431},
pages = {110728},
doi = {10.1016/j.jneumeth.2026.110728}
}
EUVIMED