analiza aiqbench
This paper addresses a core machine learning challenge in medical imaging by replacing raw intensity-based similarity with complementary semantic representations derived from a 3D foundation model. The approach combines fine-grained per-voxel features from "anatomix" with coarse segmentation labels to solve whole-body PET/CT registration while preserving non-biological tumor volumes. Winning the PSMAReg Challenge without task-specific training demonstrates the robustness of representation learning for longitudinal medical data analysis.
Registration is a Representation Problem: Solving Whole-Body PET/CT Registration
arXiv:2610.18804v1 Announce Type: cross Abstract: Raw image intensities are a poor surrogate for similarity across images in correspondence problems. Further, while auxiliary segmentation labels can provide semantic identification and improve matching, they can assign only coarse identities to fine-grained anatomy and often cover only a subset of structures, particularly in whole-body settings. We present an approach that instead establishes correspondence using complementary representations of anatomy and demonstrate its utility for longitudinal whole-body PET/CT registration in the Learn2Reg 2026 PSMAReg Challenge. Our 3D foundation model, anatomix, is used to derive fine-grained, per-voxel semantic representations that are consistent across volumes, and we further employ off-the-shelf segmentation models to obtain coarse anatomical identities. Together, these representations drive a GPU-accelerated iterative registration solver (FireANTs) in a hybrid multistage framework: the first stage estimates alignment from dense multi-channel features, after which a second refines the deformation using both anatomix- and segmentation-based losses. Lastly, because registration should not introduce non-biological tumor volume changes across time, we develop lesion-specific component-wise volume preservation schemes. Without any task-specific training, our system won first place in the PSMAReg Challenge leaderboard. Code and interactive visualizations are available at http://www.neeldey.com/reps-for-reg.
Abstract:Raw image intensities are a poor surrogate for similarity across images in correspondence problems. Further, while auxiliary segmentation labels can provide semantic identification and improve matching, they can assign only coarse identities to fine-grained anatomy and often cover only a subset of structures, particularly in whole-body settings. We present an approach that instead establishes correspondence using complementary representations of anatomy and demonstrate its utility for longitudinal whole-body PET/CT registration in the Learn2Reg 2026 PSMAReg Challenge. Our 3D foundation model, anatomix, is used to derive fine-grained, per-voxel semantic representations that are consistent across volumes, and we further employ off-the-shelf segmentation models to obtain coarse anatomical identities. Together, these representations drive a GPU-accelerated iterative registration solver (FireANTs) in a hybrid multistage framework: the first stage estimates alignment from dense multi-channel features, after which a second refines the deformation using both anatomix- and segmentation-based losses. Lastly, because registration should not introduce non-biological tumor volume changes across time, we develop lesion-specific component-wise volume preservation schemes. Without any task-specific training, our system won first place in the PSMAReg Challenge leaderboard. Code and interactive visualizations are available at this http URL.
Technical report. First place winning solution for the MICCAI Learn2Reg 2026 PSMAReg challenge. Code and volumetric visualizations are available at this http URL . Keywords: representation learning, medical image registration
Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
arXiv:2610.18804 [cs.CV]
(or arXiv:2610.18804v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2610.18804 arXiv-issued DOI via DataCite (pending registration)
Text integral sub licența deschisă a publicației · 239 cuvinte
Subiecte conexe
HumanoidArena: Benchmarking Egocentric Hierarchical Whole-body Learning
arXiv · 17 iun. 2026
Dyna-2: General Purpose Physical Agents with Hour-Long Autonomy — Jason Ma, Dyna Robotics
Latent Space · 13 oct. 2026
The Entangled Dimensions of Decision Theory
LessWrong · 30 iul. 2026
If This Were a Test, How Much Would It Cost?
LessWrong · 16 iun. 2026