FGML-DG: Feynman-Inspired Cognitive Science Paradigm for Cross-Domain Medical Image Segmentation

Yucheng Song; Chenxi Li; Haokang Ding; Zhining Liao; Zhifang Liao

cs.CVApr 12, 2026

FGML-DG: Feynman-Inspired Cognitive Science Paradigm for Cross-Domain Medical Image Segmentation

Authors: Yucheng Song, Chenxi Li, Haokang Ding, Zhining Liao, Zhifang Liao

Organizations: School of Computer Science and Engineering, Central South University, Changsha, China · School of Health & Wellbeing, University of Glasgow, UK

Abstract

In medical image segmentation across multiple modalities (e.g., MRI, CT, etc.) and heterogeneous data sources (e.g., different hospitals and devices), Domain Generalization (DG) remains a critical challenge in AI-driven healthcare. This challenge primarily arises from domain shifts, imaging variations, and patient diversity, which often lead to degraded model performance in unseen domains. To address these limitations, we identify key issues in existing methods, including insufficient simplification of complex style features, inadequate reuse of domain knowledge, and a lack of feedback-driven optimization. To tackle these problems, inspired by Feynman's learning techniques in educational psychology, this paper introduces a cognitive science-inspired meta-learning paradigm for medical image domain generalization segmentation. We propose, for the first time, a cognitive-inspired Feynman-Guided Meta-Learning framework for medical image domain generalization segmentation (FGML-DG), which mimics human cognitive learning processes to enhance model learning and knowledge transfer. Specifically, we first leverage the 'concept understanding' principle from Feynman's learning method to simplify complex features across domains into style information statistics, achieving precise style feature alignment. Second, we design a meta-style memory and recall method (MetaStyle) to emulate the human memory system's utilization of past knowledge. Finally, we incorporate a Feedback-Driven Re-Training strategy (FDRT), which mimics Feynman's emphasis on targeted relearning, enabling the model to dynamically adjust learning focus based on prediction errors. Experimental results demonstrate that our method outperforms other existing domain generalization approaches on two challenging medical image domain generalization tasks.

Figures & tables

Figure 1: Illustration of the motivation for FGML-DG, inspired by human cognitive mechanisms through the Feynman learning technique, including: (1) understanding and simplifying style concepts in medical images, (2) reusing and memorizing existing style knowledge, and (3) conducting targeted feedback learning on errors during the learning process

Figure 2: Overview of the FGML-DG framework. (a) In the meta-learning stage, we employ meta-style knowledge alignment methods and meta-style memory and review methods, inspired by the Feynman learning technique, to train the model. (b) Drawing from the feedback-targeted learning aspect of Feynman learning, we designed a feedback-driven retraining strategy that allows the model to dynamically adjust its learning focus based on prediction errors.

Figure 3: By employing Bezier transformations, style divergence can be facilitated from the source domain to different style domains.

	T2								T1CE
	Dice (%)				HD (mm)				Dice (%)				HD (mm)
Method	Flair	T1	T1CE	Average	Flair	T1	T1CE	Average	Flair	T1	T2	Average	Flair	T1	T2	Average
Fed-DG [ Liu et al.(2021a)Liu, Chen, Qin, Dou, and Heng ]	75.77	5.82	9.51	30.37	14.45	54.03	51.06	39.85	33.03	58.30	4.09	31.72	32.07	22.35	56.08	36.83
MixStyle [ Zhou et al.(2021)Zhou, Yang, Qiao, and Xiang ]	77.03	45.68	40.23	54.31	12.97	23.10	24.36	20.14	37.55	63.12	68.31	56.32	28.75	18.74	14.91	20.80
CSDG [ Ouyang et al.(2022)Ouyang, Chen, Li, Li, Qin, Bai, and Rueckert ]	61.37	47.53	43.84	50.91	16.74	22.77	21.58	20.36	42.11	62.77	65.79	56.89	22.15	19.75	16.73	19.54
SADN [ Zhou et al.(2022b)Zhou, Qi, Yang, Ni, and Shi ]	75.87	49.36	38.09	54.44	13.44	20.15	23.56	19.05	47.31	63.64	63.00	57.98	21.03	18.06	17.56	18.88

Table 1: Comparison of different methods on the BraTS dataset.

	CT-MRI					MRI -CT
Method	Liver	R-Kidney	L-Kidney	Spleen	Average	Liver	R-Kidney	L-Kidney	Spleen	Average
Cutout [ DeVries and Taylor(2017) ]	79.80	82.32	82.14	76.24	80.12	86.99	63.66	73.74	57.60	70.50
MixStyle [ Zhou et al.(2021)Zhou, Yang, Qiao, and Xiang ]	77.63	78.41	78.03	77.12	77.80	86.66	48.26	65.20	55.68	63.95
AdvBias [ Cheng et al.(2023)Cheng, Gokhale, and Yang ]	78.54	81.70	80.69	79.73	80.17	87.63	52.48	68.28	50.95	64.84
RandConv [ Xu et al.(2020)Xu, Liu, Yang, Raffel, and Niethammer ]	73.63	79.69	85.89	83.43	80.66	84.14	76.81	77.99	67.32	76.56
CSDG [ Ouyang et al.(2022)Ouyang, Chen, Li, Li, Qin, Bai, and Rueckert ]	86.62	87.48	86.88	84.27	86.31	85.62	80.02	80.42	75.56	80.40

Table 2: Comparison of different methods on the Abdominal Multi-Organ dataset.

Figure 4: Visual Comparison Results on the BraTS Dataset.

Figure 5: Visual Comparison Results on the Abdominal Multi-Organ Dataset.

	MKA	MetaStyle	FDRT	Avg.Dice	Avg.HD
Meta - Base	$\times$	$\times$	$\times$	25.33	45.23
Varant 1	$\times$	$\checkmark$	$\times$	26.04	44.37
Varant 2	$\times$	$\times$	$\checkmark$	28.67	41.98
Varant 3	$\times$	$\checkmark$	$\checkmark$	29.97	40.03
Varant 4	$\checkmark$	$\times$	$\times$	50.36	24.74
Varant 5	$\checkmark$	$\checkmark$	$\times$	60.21	16.67

Table 3: Ablation Experiment Results for the Key Components of the FGML-DG Framework.

	$L_{align}$	$L_{cons}$	Avg.Dice	Avg.HD
FGML-DG w/o $L_{cons}$ + $L_{align}$	$\times$	$\times$	29.97	40.03
FGML-DG w/o $L_{align}$	$\times$	$\checkmark$	40.21	32.57
FGML-DG w/o $L_{cons}$	$\checkmark$	$\times$	60.33	19.38
FGML-DG	$\checkmark$	$\checkmark$	69.63	13.37

Table 4: Ablation Experiment Results for the Loss Functions in the Meta-Style Knowledge Alignment Method.

Domain Generalization

Domain generalization develops models that perform reliably in new environments without training on target-domain data.

See all 375 papers

Oct 4, 2026

CoDG-Net: Structure-Guided Style Diffusion and Collaborative Learning to Mitigate Catastrophic Forgetting in Medical Image Domain Generalization

Domain Generalization (DG) for medical image segmentation is both highly challenging and critically important. However, existing medical DG methods largely overlook the issue of Catastrophic Forgetting (CF): Models often sacrifice their ability to retain source-domain knowledge while pursuing cross-domain robustness. This can directly threaten diagnostic safety in already-deployed clinical scenarios. To address this, we investigate data augmentation strategies and catastrophic forgetting for medical image DG segmentation. First, we propose a structure-guided style diffusion augmentation method. Constrained by anatomical structure consistency in the frequency domain, this method performs cross-domain diffusion on the amplitude spectrum, generating samples with more diverse and broader style coverage to better support domain generalization. Then, we design a collaborative learning network with a dual-branch interactive architecture (CoDG-Net), together with a novel learning bias-guided strategy that adaptively regulates knowledge transfer at both the layer level and the task level, thereby effectively mitigating catastrophic forgetting on the source domain. Experiments and ablation studies on single-source and multi-source medical DG benchmark datasets demonstrate that CoDG-Net not only outperforms existing state-of-the-art methods in target-domain segmentation performance, but also achieves a lower forgetting rate on the source-domain data. The code is available at: https://github.com/wangprocess/CoDG-Net.

Yucheng Song, Jincan Wang, Haokang Ding +3

School of Computer Science and Engineering, Central South University, Changsha, China · School of Software Engineering, Xi’an Jiaotong University, Xi’an, Shaanxi, China · College of Computer Science and Artificial Intelligence, Fudan University, Shanghai, China

Data Augmentation Cross-Domain Generalization Medical Image Segmentation Diffusion-Based Data Augmentation Catastrophic Forgetting

Ask PDF

May 11, 2026

Frequency Adapter with SAM for Generalized Medical Image Segmentation

Medical image segmentation is a critical task in computer-aided diagnosis and treatment planning. However, deep learning models often struggle to generalize across datasets due to domain shifts arising from variations in imaging protocols, scanner types, and patient populations. Traditional domain generalization (DG) methods utilize causal feature learning, adversarial consistency, and style augmentation to improve segmentation robustness. While effective, these approaches rely on explicit feature alignment, adversarial objectives, or handcrafted augmentations, which may not fully exploit the capabilities of foundation models. Recently, the Segment Anything Model (SAM) has demonstrated strong generalization capabilities in segmentation tasks. SAM-based DG methods attempt to improve medical image segmentation. However, these approaches primarily operate in the spatial domain and overlook frequency-based discrepancies that significantly affect model robustness. In this work, we propose Frequency-based Domain Generalization with SAM (FSAM), a novel framework that integrates Low-Rank Adaptation (LoRA) for efficient fine-tuning and a frequency adapter to incorporate frequency-domain representations for single-source domain generalization. FSAM enhances SAM's segmentation robustness by extracting domain-invariant high-frequency features, mitigating frequency-related domain shifts. Experimental results on fundus and prostate datasets demonstrate that FSAM outperforms existing traditional DG and SAM-based DG approaches in domain generalization. Codes and pre-trained models will be made available on GitHub.

Phuoc-Nguyen Bui, Van-Nguyen Pham, Duc-Tai Le +2

Sungkyunkwan University, Korea

Frequency-Domain Feature Learning Domain Generalization Vision Foundation Model Adaptation Medical Image Segmentation

Ask PDF

Apr 19, 2026

From Adaptation to Generalization: Adaptive Visual Prompting for Medical Image Segmentation

Visual prompting has emerged as a powerful method for adapting pre-trained models to new domains without updating model parameters. However, existing prompting methods typically optimize a single prompt per domain and apply it uniformly to all inputs, limiting their ability to generalize under intra and inter-domain variability, which is especially critical in the medical field. To address this, we propose APEX, an Adaptive Prompt EXtraction framework that retrieves input-specific prompts from a learnable prompt memory. The memory stores diverse, domain-discriminative prompt representations and is queried via domain features extracted from the Fourier spectrum. To learn robust and discriminative domain features, we introduce a novel Low-Frequency Feature Contrastive (LFC) learning framework that clusters representations from the same domain while separating those from different domains. Extensive experiments on two medical segmentation tasks demonstrate that APEX significantly improves generalization across both seen and unseen domains. Furthermore, it complements any existing backbones and consistently enhances performance, confirming its effectiveness as a plug-and-play prompting solution in medical fields. The code is available at https://github.com/cetinkayaevren/apex/

Evren Çetinkaya, Sangmin Lee, Jung Uk Kim +2

Technical University of Munich · Korea University · Kyung Hee University +1

Domain Generalization Medical Image Segmentation Visual Prompt Learning

Ask PDF

Abstract

Domain Generalization

Explore similar work

CoDG-Net: Structure-Guided Style Diffusion and Collaborative Learning to Mitigate Catastrophic Forgetting in Medical Image Domain Generalization

Frequency Adapter with SAM for Generalized Medical Image Segmentation

From Adaptation to Generalization: Adaptive Visual Prompting for Medical Image Segmentation