NeuroBridge - Bridging Multi-Task MRI Knowledge for Neurodegenerative Disease Diagnosis

Sep 21, 2026

Accurate MRI-based identification of Alzheimer’s disease, mild cognitive impairment, and related dementias remains difficult due to subtle, heterogeneous structural changes. NeuroBridge is a clinically guided multi-task MRI framework that combines large-scale self-supervised pretraining with hippocampal segmentation, hippocampal atrophy classification, and reconstruction objectives, followed by gated fusion fine-tuning. Evaluation on ADNI and OASIS demonstrates top classification performance, strong cross-cohort generalization, and probability-based opportunistic screening potential.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages48

Reading Time2 h 1 min

AccessFree to download

Abstract

Introduction

Methods

Results

Discussion

Background

The need for early, reliable neurodegenerative diagnosis

Representation learning and foundation-style MRI models

Clinical interpretation as multi-cue reasoning

NeuroBridge: Bridging Multi-Task MRI Knowledge for Neurodegenerative Disease Diagnosis Mengyu Li1,2,3, Guoyao Shen1,2,3, Chad W. Farris3,4,5*, Xin Zhang1,2,3,6,7,8* 1 Department of Mechanical Engineering, Boston University, Boston, MA 02215, United States 2The Photonics Center, Boston University, Boston, MA 02215, United States 3 Rafik B. Hariri Institute for Computing and Computational Science & Engineering, Boston University, Boston, MA, 02215, United States 4 Department of Radiology, Boston University Chobanian & Avedisian School of Medicine, Boston, MA 02118, United States 5 Department of Radiology, Boston Medical Center, Boston, MA 02118, United States 6 Department of Electrical and Computer Engineering, Boston University, Boston, MA 02215, United States 7 Department of Biomedical Engineering, Boston University, Boston, MA 02215 , United States 8 Division of Materials Science and Engineering, Boston University, Boston, MA 02215 , United States *Corresponding authors: Prof. Xin Zhang, Department of Mechanical Engineering, Boston University, Boston, MA 02215, United States Email: [xinz@bu.edu](xinz@bu.edu) Dr. Chad Farris, Department of Radiology, Boston University Chobanian & Avedisian School of Medicine, Boston, MA 02118, United States [Email:](Email: cfarris@bu.edu)[ cfarris@bu.edu](Email: cfarris@bu.edu) Present/permanent address: College of Engineering, 8 St Mary's St, Boston, MA 02215 Declarations of interest: none. Abstract INTRODUCTION: Accurate MRI-based identification of Alzheimer's disease (AD), mild cognitive impairment (MCI), and related dementias remains challenging because disease-related structural changes are often subtle and heterogeneous. We developed NeuroBridge, a clinically guided multi-task MRI framework for neurodegenerative disease diagnosis. METHODS: NeuroBridge integrates large-scale self-supervised MRI pretraining with hippocampal segmentation, hippocampal atrophy classification, and reconstruction objectives, followed by gated fusion fine-tuning. Performance was evaluated across ADNI and OASIS cohorts, including cross-cohort transfer, probability-based analysis, and opportunistic screening. RESULTS: NeuroBridge achieved the highest performance across evaluated classification tasks, reaching 88. 17% accuracy for AD versus cognitively normal controls in ADNI and 82.78% in OASIS. The largest gains occurred in MCI-related and mixed-diagnosis settings. The framework demonstrated strong cross-cohort generalization, systematic associations between predicted-class probability and accuracy, and the feasibility of probability-based opportunistic screening. DISCUSSION: Clinically guided multi-task representation learning improves neurodegenerative MRI diagnosis beyond conventional single-task approaches. NeuroBridge provides a robust and scalable framework for dementia assessment and MRI-based opportunistic screening. Keywords: Alzheimer's disease; Mild cognitive impairment; Neurodegenerative disease; Brain MRI; Magnetic resonance imaging; Deep learning; Transformer; Masked autoencoder; Foundation model; Multi-task learning; Representation learning; Cross-cohort generalization; Prediction probability analysis; Opportunistic screening; Dementia diagnosis. 1. BACKGROUND Neurodegenerative diseases, particularly Alzheimer’s disease (AD), represent a growing global health and economic burden as populations age [1] . AD is characterized by a prolonged preclinical and prodromal phase during which subtle structural brain changes precede overt cognitive decline [2,3] . Mild cognitive impairment (MCI) is an intermediate stage between cognitively normal status (CN) and dementia and is associated with a substantially increased risk of progression to dementia [4 ,5] . Timely and accurate identification of individuals in early stages of neurodegenerative diseases, as well as reliable differentiation among MCI, AD, and other dementia subtypes, is critical for early interven

FAQs

What problem does NeuroBridge address in MRI-based diagnosis?

NeuroBridge addresses the challenge of accurately identifying Alzheimer’s disease, mild cognitive impairment, and related dementias from MRI when structural changes are subtle and heterogeneous.

How is NeuroBridge built and trained?

NeuroBridge integrates large-scale self-supervised MRI pretraining with hippocampal segmentation, hippocampal atrophy classification, and reconstruction objectives, then applies gated fusion fine-tuning.

What performance results are reported for NeuroBridge?

The paper reports the highest performance across evaluated classification tasks, including 88.17% accuracy for AD versus cognitively normal controls in ADNI and 82.78% in OASIS, with the largest gains in MCI-related and mixed-diagnosis settings.

Sep 21, 2026

Uploaded by Olivia Brown