Unsupervised multimodal semantic discovery aims to learn discriminative representations from multimodal data. However, existing methods suffer from two key limitations. First, they only align instances across modalities without modeling semantic-level consistency, which fails to mitigate semantic bias caused by the gaps among feature distributions of multiple modalities. Second, they inevitably generate incorrect negative pairs during contrastive learning, pushing semantically similar samples apart. To address these challenges, we propose GLAD (Global and Local semantic Alignment for unsupervised multimodal semantic Discovery), which aligns multimodal data at both global and local semantic levels. At the global level, GSA integrates multi-modal features into a shared space and employs joint clustering via optimal transport to capture common semantic patterns while mitigating cross-modality semantic bias. At the local level, LSA adaptively weights samples within each cluster based on their semantic importance, alleviating the effect of incorrect negative pairs. Through the joint optimization of GSA and LSA, GLAD effectively captures both the global semantic structure and the local semantic nuances of multimodal data. Extensive experiments on three benchmark datasets demonstrate GLAD significantly outperforms state-of-the-art methods, with an average improvement of 3.22%.

State Space Models (SSMs), particularly Mamba, have demonstrated significant potential in medical image segmentation due to their capability to model long-range dependencies with linear computational complexity. However, achieving accurate medical image segmentation necessitates the effective learning of both multi-scale detailed feature representations and global contextual dependencies. Although existing research has attempted to address this challenge by integrating CNNs and SSMs to leverage their respective strengths, they have not developed specialized modules to effectively capture multi-scale feature representations, nor have they sufficiently addressed the directional sensitivity issue when applying Mamba to 2D image data. To overcome these limitations, we propose a Multi-Scale Vision Mamba UNet model for medical image segmentation, termed MSVM-UNet. Specifically, by introducing multi-scale convolutions in the VSS blocks, we can more effectively capture and aggregate multi-scale feature representations from the hierarchical features of the VMamba encoder and better handle 2D visual data. Additionally, the large kernel patch expanding (LKPE) layers achieve more efficient upsampling of feature maps by simultaneously integrating spatial and channel information. Extensive experiments on the Synapse and ACDC datasets demonstrate that our approach is more effective than some state-of-the-art methods in capturing and aggregating multi-scale feature representations and modeling long-range dependencies between pixels. Our implementation is available at https://github.com/gndlwch2w/msvm-unet.

In this paper, we propose a novel method, namely boundary feature pyramid network (BFP-Net), which can effectively segment targets with blurred boundaries. Specifically, we first propose a global feature fusion module (GFFM) at the top of BFP-Net to fuse feature maps within different scales. It can learn more global region localization of the target and help the learning of boundaries more effectively. Then, we propose a series of boundary enhancement modules (BEMs) at the decoder to effectively extract and integrate boundary information during the upsampling process, thereby enhancing the ability to capture fine details (such as the boundaries). Furthermore, we introduce a boundary-enhanced composite loss function to effectively segment both the regions and their boundaries within different scales. Finally, extensive experiments on two widely-used datasets demonstrate that BFP-Net is more effective in fusing contextual information and guiding feature map boundaries compared to previous competitive methods.