Few-shot medical image segmentation (FSMIS) has become one of the potential solutions for limited annotated medical image analysis. However, the realistic multiple domains of medical data demands the FSMIS models generalizing across domains. Thus, the cross-domain few-shot medical image segmentation (CD-FSMIS) is introduced, and we propose the Adversarial Prototypical Perturbation (APP) model which employs the adversarial learning strategy in the prototype learning process for gaining the domain robust prototypes. Specifically, the method consists of two components: adversarial signal formulation (ASF) and interactive prototypical attack (IPA). The ASF module collects perturbations from the gained gradients from both of the intra-class variation measurement loss and the inter-class variation measurement loss, and the IPA module aims to impose gained perturbations on the prototypical representation construction process with the two stages of interactive attacking manner. Additionally, a local-imbalance aware whitening loss is designed to resist the shift-sensitive local components for further enforcing to learn the domain robust prototypical representation in the IPA module. Extensive experiments are conducted on three cross-domain medical imaging datasets, and the results demonstrate that our model outperforms the state-of-the-art few-shot medical image segmentation methods. The code is available at https://github.com/YazhouZhu19/APP .

Existing few-shot medical image segmentation (FSMIS) models fail to address a practical issue in medical imaging: the domain shift caused by different imaging techniques, which limits the applicability to current FSMIS tasks. To overcome this limitation, we focus on the cross-domain few-shot medical image segmentation (CD-FSMIS) task, aiming to develop a generalized model capable of adapting to a broader range of medical image segmentation scenarios with limited labeled data from the novel target domain. Inspired by the characteristics of frequency domain similarity across different domains, we propose a Frequency-aware Matching Network (FAMNet), which includes two key components: a Frequency-aware Matching (FAM) module and a Multi-Spectral Fusion (MSF) module. The FAM module tackles two problems during the meta-learning phase: 1) intra-domain variance caused by the inherent support-query bias, due to the different appearances of organs and lesions, and 2) inter-domain variance caused by different medical imaging techniques. Additionally, we design an MSF module to integrate the different frequency features decoupled by the FAM module, and further mitigate the impact of inter-domain variance on the model's segmentation performance. Combining these two modules, our FAMNet surpasses existing FSMIS models and Cross-domain Few-shot Semantic Segmentation models on three cross-domain datasets, achieving state-of-the-art performance in the CD-FSMIS task.

Cross-domain few-shot medical image segmentation (CDFSMIS) presents the fundamental challenge of segmenting novel anatomical or tissue structures on unfamiliar medical imaging domains with limited annotated data. In this paper, we conduct an in-depth investigation of CDFSMIS and identify two critical observations: 1) the conventional matching mechanisms from existing few-shot models are particularly vulnerable to discrepancies in local characteristics between different domains and 2) the semantic representations learned from source domains often lack robustness when generalizing to unfamiliar target domains. Motivated by these insights, we propose a novel Dynamic Semantic Matching (DSM) framework that addresses these challenges through a three-component approach. First, we design a support-query feature re-weighting (SFR) mechanism that leverages multilevel hidden features to suppress domain-specific contents. Second, we introduce a dynamic semantic information selection (DSIS) strategy that adaptively identifies and combines domain-robust channels to construct generalizable representations. Third, we develop a dual-perspective semantic center calculation method to address the inherent texture imbalance in medical images. Extensive experiments on four unfamiliar target domains (MS-CMR, PI-PMR, Chest-X-Ray and ISIC2018) demonstrate that our approach significantly outperforms state-of-the-art few-shot segmentation and cross-domain few-shot segmentation models, validating the effectiveness of DSM in simultaneously addressing domain generalization and semantic matching challenges in medical image segmentation. The source code is available at https://github.com/YazhouZhu19/DSM.

Magnetic resonance imaging (MRI) is a widely used diagnostic technique in the initial evaluation of patients with primary brain tumor. Automatic segmentation algorithms for brain tumor plays an important role in diagnosis of tumor subregions and treatments for patients. In this paper, we propose a novel 3D convolutional neural network for segmentation of brain tumor, which is based on the traditional encoder-decoder architecture and inspired by multi-scale strategy. The proposed network employs multi-scale strategy with the designed dualencoder structure which can be considered as two scales streams to extract features from two scales inputs respectively. To integrate the output features from two scales encoders, a multiscale attention is model designed to weight and integrated the multi-scale features. Moreover, with the purpose of modeling long-range feature dependencies more efficiently, hidden features from different positions of stream are also fused and then utilized by attention mechanism in each scale encoder. To validate the performance of the proposed network, we conduct experiments on the brain tumor segmentation dataset BraTS2019 against several state-of-art methods. The results show that proposed method has the effective and competitive performance in brain tumor segmentation.

Magnetic resonance imaging (MRI) is an important medical diagnosis technique in clinical diagnosis, while the quality of MR images is always damaged by the noise which is caused in the image acquisition process. In the classic image denoising methods, how to design an excellent regularizer with the prior knowledge of image is the key to solve the denoising problem. In this work, we introduce the deep neural regularizer for the MRI denoising tasks, the deep neural regularizer is made up of neural network structure and objective function, similar to the classic regularizer, both of these two parts are designed with the prior knowledge of image. The proposed neural network has three main parts: encoder network, decoder network and skip connections, the encoder network which consists of five down-sampling blocks is enforced to deeply extract low-resolution or highly-abstract MR image features, similar to the encoder network architecture, the decoder network is made up of five up-sampling blocks and is enforced to restore high-resolution MR image features. To generate more finer image features, we also use skip connections to transmit the abstract information from encoder to decoder directly. The objective function consists of data fidelity term and image quality penalty term, specifically, to enforce the capability of data fidelity term, we add the self-designed image structural consistency calculation to data fidelity term besides only calculating the image consistency over image pixels with mean squared error. Meanwhile, to guide the network generate more clearer image and reduce noise information, with the prior knowledge of image sharpness, an image quality penalty term which calculates the MR image sharpness is also added to the objective function. Experimental results over the simulated MRI data and real clinical data demonstrate the proposed network can achieve superior performance compared with other methods in terms of peak signal to noise ratio, structure similarity index, image average gradient and image information entropy.