Meeting Banner
Abstract #0221

HomoNet: Enhancing Multi-Center Alzheimer's Disease Classification via Disentangling Homogeneous Features in Structural MRI

Linlin Gao1, Chenyang Lin1, Yihong Dong1, Jialu Zhang2, and Jianhua Wang3
1Ningbo University, Ningbo, China, 2GE Healthcare, MR Research, Beijing, China, 3The First Affiliated Hospital of Xiamen University, Xiamen, China

Synopsis

Keywords: Analysis/Processing, Alzheimer's Disease, Multi-center structural MRI; Data heterogeneity; Disentangling learning; Homogeneous features

Motivation: Variations in imaging conditions across centers challenge accurate diagnosis of Alzheimer’s Disease (AD) and Mild Cognitive Impairment (MCI) using structural MRI (sMRI).

Goal(s): Develop a robust and scalable model that accurately classifies AD, MCI, and normal controls (NC) across multi-center sMRI datasets without re-training on new data.

Approach: This study presents "HomoNet", a model employing weakly-supervised disentangling learning to separate homogeneous (center-invariant) and heterogeneous (center-specific) features.

Results: HomoNet outperforms current methods, achieving 74.37% accuracy in distinguish NC, MCI, and AD on an external dataset, enhancing diagnostic robustness across diverse multi-center MRI data.

Impact: HomoNet enhances multi-center sMRI image classification, improving diagnostic accuracy for AD and MCI. It effectively addresses data heterogeneity, increases model generalizability, and provides a scalable, re-training-free solution, making it highly applicable for real-world clinical imaging.

How to access this content:

For one year after publication, abstracts and videos are only open to registrants of this annual meeting. Registrants should use their existing login information. Non-registrant access can be purchased via the ISMRM E-Library.

After one year, current ISMRM & ISMRT members get free access to both the abstracts and videos. Non-members and non-registrants must purchase access via the ISMRM E-Library.

After two years, the meeting proceedings (abstracts) are opened to the public and require no login information. Videos remain behind password for access by members, registrants and E-Library customers.

Click here for more information on becoming a member.

Keywords