Academic Journal
Medical hierarchical image classification via dual-geometry image-text learning.
| Τίτλος: | Medical hierarchical image classification via dual-geometry image-text learning. |
|---|---|
| Συγγραφείς: | Fan L; Centre for Healthy Brain Ageing, Discipline of Psychiatry and Mental Health, School of Clinical Medicine, Faculty of Medicine and Health, UNSW Sydney, Australia; School of Computer Science and Engineering, UNSW Sydney, Australia. Electronic address: lei.fan1@unsw.edu.au., Sowmya A; School of Computer Science and Engineering, UNSW Sydney, Australia., Meijering E; School of Computer Science and Engineering, UNSW Sydney, Australia., Yu Z; Department of Data Science & AI, Monash University, Australia., Ge Z; Department of Data Science & AI, Monash University, Australia., Song Y; School of Computer Science and Engineering, UNSW Sydney, Australia. |
| Πηγή: | Medical image analysis [Med Image Anal] 2026 Jul; Vol. 112, pp. 104120. Date of Electronic Publication: 2026 Apr 30. |
| Τύπος έκδοσης: | Journal Article |
| Γλώσσα: | English |
| Στοιχεία περιοδικού: | Publisher: Elsevier Country of Publication: Netherlands NLM ID: 9713490 Publication Model: Print-Electronic Cited Medium: Internet ISSN: 1361-8423 (Electronic) Linking ISSN: 13618415 NLM ISO Abbreviation: Med Image Anal Subsets: MEDLINE |
| Imprint Name(s): | Publication: Amsterdam : Elsevier Original Publication: London : Oxford University Press, [1996- |
| Ιατρικοί όροι (MeSH): | Image Interpretation, Computer-Assisted*/methods , Pattern Recognition, Automated*/methods , Classification Algorithms* , Machine Learning*, Skin Diseases/diagnostic imaging ; Uterine Cervical Neoplasms/diagnostic imaging ; Female ; Humans |
| Περίληψη: | Hierarchical image classification is a fundamental challenge in medical image analysis, as tree-structured taxonomies inherently reflect biological and clinical relationships, spanning the general categorisation of disease entities and fine-grained cellular distinctions. Existing approaches primarily rely on multi-task learning and fine-grained detection, often requiring intricate model design and complex training strategies. In this paper, we aim to exploit the negative curvature property of hyperbolic space, which allows efficient representation of hierarchical structures. We propose a dual-geometry image-text framework, termed H2CL. Specifically, we introduce a lightweight classifier head on top of image backbones to extract both Euclidean and hyperbolic features, which are then combined to simultaneously preserve taxonomic consistency from an etiological perspective and enhance instance discrimination from a morphological perspective. Furthermore, a text branch is incorporated to integrate label semantics, where an entailment loss is employed to jointly model image-text alignment and inter-sample relationships. Extensive experiments on cervical cell, skin lesion, and gallbladder disease datasets demonstrate that our framework consistently outperforms advanced methods. Compared to the standard Swin Transformer, H2CL achieves an average accuracy improvement of 7% across all three datasets at the fine-grained level, with similarly consistent gains observed when integrated with other backbone models. The source code is publicly available at https://github.com/MCPathology/H2CL. (Copyright © 2026 The Authors. Published by Elsevier B.V. All rights reserved.) |
| Competing Interests: | Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. |
| Contributed Indexing: | Keywords: Contrastive learning; Hierarchical classification; Hyperbolic space |
| Entry Date(s): | Date Created: 20260505 Date Completed: 20260613 Latest Revision: 20260629 |
| Update Code: | 20260629 |
| DOI: | 10.1016/j.media.2026.104120 |
| PMID: | 42085920 |
| Βάση Δεδομένων: | MEDLINE |
| ISSN: | 1361-8423 |
|---|---|
| DOI: | 10.1016/j.media.2026.104120 |