HypCBC: Domain-Invariant Hyperbolic Cross-Branch Consistency for Generalizable Medical Image Analysis

Francesco Di Salvo [email protected], Sebastian Doerrich [email protected], Jonas Alle [email protected], Christian Ledig [email protected] · 2026

Read HypCBC: Domain-Invariant Hyperbolic Cross-Branch Consistency for Generalizable Medical Image Analysis by Francesco Di Salvo [email protected], Sebastian Doerrich [email protected], Jonas Alle [email protected], Christian Ledig [email protected] (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-03264.txt.

Description

Hyperbolic geometry, where space curves away ever faster, helps medical image models generalize in research by Francesco Di Salvo with 3 co-authors. Clinical data has layered form that flat Euclidean geometry captures poorly. Shifts from other scanners, plans, and patient populations further hinder reliable results and slow clinical adoption of deep networks trained at one site and deployed at another hospital or device.

Their method adds an unsupervised consistency constraint between branches in hyperbolic space, promoting domain-invariant traits. It is presented as the first comprehensive review of hyperbolic representation learning for medical image analysis. Gains were clear across many standard datasets and three vision transformer models. On domain generalization benchmarks spanning skin, histopathology, and retinal imaging, the method beat Euclidean methods on average.

Catalog record

CREATORFrancesco Di Salvo [email protected], Sebastian Doerrich [email protected], Jonas Alle [email protected], Christian Ledig [email protected]
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE50 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-03264

Browse the full library · About the Tower of Babel