Applications of Hypergraph Learning for Brain Disorder Diagnosis with Neuroimaging: A Survey

TY - JOUR

T1 - Applications of Hypergraph Learning for Brain Disorder Diagnosis with Neuroimaging

T2 - A Survey

AU - He, Meng Shen

AU - Tian, Xu

AU - Li, Jun Jian

AU - Yue, Hai Lin

AU - Li, Xin Yu

AU - Kuang, Hu Lin

AU - Lin, Hanhe

AU - Qiu, Zhen

AU - Liu, Jin

N1 - Publisher Copyright: © Institute of Computing Technology, Chinese Academy of Sciences 2026.

PY - 2026/5/6

Y1 - 2026/5/6

N2 - The human brain, as the most complex organ, comprises billions of neurons forming intricate and dynamic networks. This complexity results in brain disorders exhibiting multifaceted manifestations, both in their pathological mechanisms and clinical symptoms, thereby posing significant challenges for accurate diagnosis and effective treatment. In the light of this, the development of advanced diagnostic techniques and analytical methodologies has become increasingly crucial. While graph learning has promising performance in modeling neuroimaging data, its reliance on pairwise (binary) relationships limits its capacity to capture higher-order interactions among brain regions. Hypergraph learning frameworks address this shortcoming by modeling complex, multi-way relationships, offering a richer and more expressive mathematical foundation for understanding brain network structure and function. This survey provides a comprehensive overview of recent advancements in hypergraph learning for neuroimaging-based brain disorder diagnosis, discussing current methodological challenges and outlining promising directions for future research in this rapidly evolving field.

AB - The human brain, as the most complex organ, comprises billions of neurons forming intricate and dynamic networks. This complexity results in brain disorders exhibiting multifaceted manifestations, both in their pathological mechanisms and clinical symptoms, thereby posing significant challenges for accurate diagnosis and effective treatment. In the light of this, the development of advanced diagnostic techniques and analytical methodologies has become increasingly crucial. While graph learning has promising performance in modeling neuroimaging data, its reliance on pairwise (binary) relationships limits its capacity to capture higher-order interactions among brain regions. Hypergraph learning frameworks address this shortcoming by modeling complex, multi-way relationships, offering a richer and more expressive mathematical foundation for understanding brain network structure and function. This survey provides a comprehensive overview of recent advancements in hypergraph learning for neuroimaging-based brain disorder diagnosis, discussing current methodological challenges and outlining promising directions for future research in this rapidly evolving field.

KW - brain disorder

KW - computer aided diagnosis

KW - functional connectivity

KW - hypergraph learning

KW - neuroimaging

UR - https://www.scopus.com/pages/publications/105038101381

U2 - 10.1007/s11390-025-5199-0

DO - 10.1007/s11390-025-5199-0

M3 - Article

AN - SCOPUS:105038101381

SN - 1000-9000

JO - Journal of Computer Science and Technology

JF - Journal of Computer Science and Technology

ER -