cs.CVMar 11, 2026

Med-DualLoRA: Local Adaptation of Foundation Models for 3D Cardiac MRI

Authors: Joan Perramon-Llussà, Amelia Jiménez-Sánchez, Grzegorz Skorupko, Fotis Avgoustidis, Carlos Martín-Isla, Karim Lekadir, Polyxeni Gkontra

Organizations: Artificial Intelligence in Medicine Lab (BCN-AIM), Departament de Matemàtiques i Informàtica, Universitat de Barcelona, Spain · Institució Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Spain

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Foundation models (FMs) show great promise for robust downstream performance across medical imaging tasks and modalities, including cardiac magnetic resonance (CMR), following task-specific adaptation. However, adaptation using single-site data may lead to suboptimal performance and increased model bias, while centralized fine-tuning on clinical data is often infeasible due to privacy constraints. Federated fine-tuning offers a privacy-preserving alternative; yet conventional approaches struggle under heterogeneous, non-IID multi-center data and incur substantial communication overhead when adapting large models. In this work, we study federated FM fine-tuning for 3D CMR disease detection and propose Med-DualLoRA, a client-aware parameter-efficient fine-tuning (PEFT) federated framework that disentangles globally shared and local low-rank adaptations (LoRA) through additive decomposition. Global and local LoRA modules are trained locally, but only the global component is shared and aggregated across sites, keeping local adapters private. This design improves personalization while significantly reducing communication cost, and experiments show that adapting only two transformer blocks preserves performance while further improving efficiency. We evaluate our method on a multi-center state-of-the-art cine 3D CMR FM fine-tuned for disease detection using ACDC and combined M&Ms datasets, treating each vendor as a federated client. Med-DualLoRA achieves statistically significant improved performance (balanced accuracy 0.768, specificity 0.612) compared to other federated PEFT baselines, while maintaining communication efficiency. Our approach provides a scalable solution for local federated adaptation of medical FMs under realistic clinical constraints.

Figures & tables

Figure 1: (a) Fine-tuning baselines. (b) Our Med-DualLoRA method is a dual low-rank adaptation framework for collaborative learning of transferable global representations with reduced communication, while preserving client-specific differences under heterogeneous data. CLS: classifier head.
 CanonGEPhilipsSiemensACDCTotal
 DISNORDISNORDISNORDISNORDISNORDISNOR
Train2875320782615116721838287
Val34141157323827217
Test53152194343401011322
Total36148223112372172212030567126
Table 1: Distribution of diseased (DIS) and non-diseased (NOR) samples across clients and splits, including overall totals.
ExperimentBalanced Acc.SensitivitySpecificityF1
Centralized
Full-ft0.745 ± 0.049    {}^{~~~}0.953 ± 0.013   {}^{~~}0.538 ± 0.111   {}^{~~}0.927 ± 0.006
Head-only0.810 ± 0.124    {}^{~~~}0.949 ± 0.006   {}^{~~}0.672 ± 0.250   {}^{~~}0.937 ± 0.022
LoRA0.780 ± 0.042    {}^{~~~}0.943 ± 0.017   {}^{~~}0.616 ± 0.092   {}^{~~}0.929 ± 0.009
Federated
Head-only0.514 ± 0.024 ∗∗0.997 ± 0.005 ∗0.030 ± 0.053 ∗∗0.909 ± 0.000
Table 2: Performance metrics (mean ± std) for centralized and federated baselines, and Med-DualLoRA. Asterisks indicate statistically significant differences compared to Med-DualLoRA using paired t-test: ∗ p<0.05, ∗∗ p<0.01.
Figure 2: (a) Client-wise performance under centralized, client-specific, and federated training. Model performance is shown per client and averaged across clients (size-normalized); error bars indicate mean ± standard deviation across runs. (b) Pareto frontier of communication cost versus balanced accuracy. Stars denote frontier configurations, illustrating the trade-off between performance and communication per round.
More in this topic

Federated Fine-Tuning

Federated fine-tuning adapts pretrained models across decentralized data holders without pooling their private data.

See all 115 papers

Explore similar work

CardsList
Jul 15, 2026

FM2^2: Unified Federated Foundation Models for Heterogeneous Multimodal Medical Imaging

Building foundation models for medical imaging requires pooling data across institutions, yet privacy regulations prohibit centralized aggregation. Existing Federated Foundation Models either fine-tune natural-image models with poor medical-domain transfer, or train from scratch within a single modality, lacking the flexibility to unify tasks. We identify an under-explored challenge, Imaging Modality Heterogeneity, where clients operate under two structural regimes: Overlapped (shared modalities with heterogeneous label distributions) and Non-overlapped (fully disjoint modalities per client). We propose FM2^2, a unified framework that trains the core backbone from scratch to preserve medical domain fidelity while optionally incorporating biomedical pretrained encoders for vision-language alignment. FM2^2 equips each client with dual Mixture-of-Experts modules (a Class-wise MoE for personalized category knowledge and a Domain-wise MoE for shared cross-modality representations), coupled with a Heterogeneous Modality Alignment (HMA) regularizer that explicitly aligns modality-specific expert parameters, admitting provable O(1/T)O(1/\sqrt{T}) convergence and generalization guarantees. FM2^2 further incorporates Caption-Enhanced Learning (CEL), where locally retained GPT-4o-generated captions serve as a textual semantic bridge enabling representation transfer across clients with disjoint modalities, and demonstrates extensibility to Federated Medical VQA. Experiments on our MIMH benchmark (classification and CEL) and real-world medical VQA datasets confirm consistent superiority over state-of-the-art federated baselines and strong out-of-modality generalization across all three tasks.

Shengchao Chen, Ting Shu

School of Artificial Intelligence, Shenzhen University Shenzhen, China · Australian AI Institute, University of Technology Sydney Sydney, Australia

Multimodal Federated LearningCross-Modal AlignmentMedical VQAMedical Image ClassificationMultimodal Foundation ModelsFederated Learning+1
AskPDF
Aug 17, 2026

CoM3^3eT: A foundation model for medical image analysis through federated, multidimensional context integration

Medical foundation models improve generalization when training AI models with limited labeled data, but remain confined to a single specialty, such as pathology or radiology, and to either sparse or dense outputs, such as classification or segmentation. Here, we present CoM3^3eT (Co-representation Multidimensional Multitask Medical Transformer), a medical vision foundation model that unifies pathology and radiology, sparse and dense predictions, and two- and higher-dimensional inputs by modeling multidimensional context with attention. CoM3^3eT outperformed other medical foundation models in an open competition spanning five tomographic, four whole-specimen, and three two-dimensional datasets, covering sparse and dense prediction tasks as well as report generation. When adapted across diverse clinical applications, training fewer than 2.5% of parameters achieved performance comparable to full fine-tuning, enabling research without access to high-performance GPU clusters. Applied to federated learning across hospitals, this approach achieved performance comparable to pooled-data training over internet connections and with consumer-grade hardware.

J. Raphael Schäfer, Kai Geissler, Till Nicke +27

Fraunhofer Institute for Digital Medicine MEVIS · Institute for Experimental Molecular Imaging, RWTH Aachen University, Aachen, Germany · Institute of Diagnostic and Interventional Radiology, Medizinische Hochschule Hannover, Hannover, Germany +11

Medical Imaging Foundation ModelsMulti-Task Learning
AskPDF
Oct 19, 2026

PFedQFT: Personalized Federated Quantum-Inspired Fine-Tuning for Medical Large Language Models

Medical data remain fragmented across institutional silos because of privacy, governance, and ownership constraints. Federated learning enables medical institutions to adapt large language models (LLMs) collaboratively without transferring raw records. However, independently parameterized low-rank adapters incur repeated parameter and communication costs across Transformer projections, while uniform aggregation may limit adaptation to heterogeneous institutional data. We propose PFedQFT, a quantum-inspired framework for personalized federated fine-tuning with a compact shared adapter. Specifically, PFedQFT constructs reusable orthonormal adaptation bases from compact skew-symmetric generators using a truncated matrix-exponential approximation followed by QR orthonormalization. Compatible projections reuse these bases, while a globally shared full core preserves the flexibility of each projection. Each client further maintains a lightweight private residual operator, whose directions are learned globally and whose coefficients remain local. Experiments on seven public medical tasks and multiple LLM backbones show substantial reductions in trainable parameters and per-round adapter upload volume, with competitive predictive performance on most tasks.

Qihan Huang, Lingxia Zhang, Guye Zhang +2

Institute of Fundamental and Frontier Sciences, University of Electronic Science and Technology of China, Chengdu, Sichuan, China · Key Laboratory of Quantum Physics and Photonic Quantum Information, Ministry of Education, University of Electronic Science and Technology of China, Chengdu, Sichuan, China · Sichuan Provincial Key Laboratory for Human Disease Gene Study and the Center for Medical Genetics, Department of Laboratory Medicine, Sichuan Academy of Medical Sciences and Sichuan Provincial People’s Hospital, University of Electronic Science and Technology of China, Chengdu, Sichuan, China +1

Federated Fine-TuningFederated LearningParameter-Efficient Fine-TuningLow-Rank AdaptationMedical LLMsQuantum-Inspired ML
AskPDF