Multi-View Large Kernel Attention Network for Multi-Contrast MRI Volumetric Super-Resolution

DCLA-UNet: dynamic cross-layer compressive attention and large axial separable convolution for multi-modal MRI brain tumor segmentation

Accurate brain tumor segmentation from multi-modal Magnetic Resonance Imaging (MRI) is critical for clinical prognosis, yet existing 3D architectures face a persistent dilemma. Traditional Convolutional Neural Network (CNN) suffer from restricted receptive fields, while emerging Vision Transformers incur prohibitive quadratic computational complexity and excessive parameterization, severely hindering their deployment on resource-constrained clinical devices. To address this fundamental trade-off between global contextual modeling and computational efficiency, this study proposes DCLA-UNet, a highly parameter-efficient 3D segmentation framework driven by two core architectural innovations. First, we design a lightweight encoder-decoder structure integrating a Slim Large-Kernel Module (SLKM) that utilizes depthwise axial convolutions to enlarge the effective receptive field, coupled with a Multi-Scale Fusion Module (MSFM) employing parallel dilated branches for robust semantic reconstruction. Second, we introduce a Dynamic Cross-Layer Compressive Attention (DCLA) mechanism that leverages extreme single-channel compression and multi-kernel spatial alignment to adaptively modulate bottleneck features, effectively bridging the semantic gap and suppressing background noise. Extensive evaluations across the BraTS (2019, 2020, 2021) and MSD BrainTumour bench- marks systematically validate the efficacy and generalization capability of DCLA-UNet. The proposed architecture requires only 0.53M trainable parameters—a 94.76% reduction compared to the 3D U-Net baseline—and consumes just 39.23 Giga Floating Point Operations (GFLOPs). Despite this extreme compression, DCLA-UNet achieves a competitive mean Dice Similarity Coefficient (DSC) of 87.5 ± 0.6% on BraTS2021, outperforming emerging lightweight architectures such as SegFormer3D and Mamba3D. While it matches the DSC of heavier models like MogaNet using only 9.2% of its parameters, a trade-off is observed in the 95th percentile Hausdorff Distance (HD_95) (7.6 ± 1.8), objectively reflecting the inherent limits of boundary precision under strict parameter constraints. DCLA-UNet successfully breaks the conventional bottleneck of 3D global context modeling by achieving a competitive balance between parameter efficiency and segmentation accuracy. It provides a robust, parameter-efficient solution for 3D medical image segmentation in storage-constrained clinical scenarios. (The source code is available at https://github.com/Helium-327/DCLA-UNet-3D ).

A cross-domain deep learning framework with graph CNN extrapolation for MRI reconstruction

A novel cross-domain DL framework for MRI reconstruction that leverages graph-based convolutional neural networks to model the autoregressive nature of Fourier features and integrates spatial domain networks and FDNs through a cross-lattice structure, enhancing feature extraction by promoting dense representation through alternating layers.

SVRCL-SR: a high spatial resolution imaging method for large-size plate-shaped components

Comprehensive evaluations on multiple datasets and SR scales indicate that the SVRCL-SR achieves superior performance in artifact suppression and high-frequency detail restoration, along with strong robustness.

Structure-Aware Attention Prototype Network for Cross-Modality Few-Shot Brain Tumor Segmentation.

Multi-modal magnetic resonance imaging (MRI) plays a crucial role in brain tumor diagnosis. However, the substantial physiological sensitivity discrepancies across imaging modalities create a severe domain gap that challenges current cross-modality segmentation methods. Many unsupervised do main adaptation (UDA) approaches reduce this gap through image style translation or distribution alignment, which have shown promising results but may struggle to preserve modality specific anatomical and pathological information. In contrast, few-shot segmentation (FSS) provides a promising alternative paradigm with strong cross-domain generalization capability, avoiding explicit domain alignment. Therefore, in this work, we propose a novel Structure-aware Attention Prototype Network (SAPNet) for cross-modality few-shot brain tumor segmentation, which fully exploits task-agnostic, multi-scale features from the large vision model. Specifically, our SAPNet consists of three main components. First, we introduce a masked support feature reconstruction (MSFR) module to encourage the network to understand anatomically meaningful structures. Second, a patch-level attention-to-prototype alignment (APA) module is proposed to adaptively balance the aggressive and conservative segmentation tendencies by cross-attention (CA) and prototype based learning. Third, a lightweight multi-scale decoder with the contrastive embedding space is employed to enhance fine grained pixel-level prediction. Extensive experiments on three public benchmarks, BraTS 2020, VS-SEG, and BraTS 2023 PEDdatasets, demonstrate that SAPNet consistently outperforms state-of-the-art UDA and FSS methods, exhibiting strong gener alization and robustness.

Dynamic multi-scale fusion single image super-resolution: a method based on Swin Transformer

Single image super-resolution (SISR) aims to reconstruct high-resolution images from low-resolution inputs, essential for medical imaging, surveillance, and remote sensing. Traditional methods falter on complex textures and edges due to limited feature extraction and inefficient cross-scale fusion. This study introduces a dynamic multi-scale fusion mechanism using the Swin Transformer for SISR. It begins with a pre-trained Swin Transformer backbone to extract hierarchical features from the low-resolution image, spanning local to global scales. A dynamic fusion module then adaptively weights features across resolutions via scale-attention learned from input texture complexity through a multi-layer perceptron (MLP), reducing losses from fixed upsampling. Multi-head cross-attention with residuals aggregates these features for enhanced representation. A pixel reconstruction head yields the high-resolution output. Training combines L1 and VGG-based perceptual losses for better visual and structural quality. Evaluations on Set5, Set14, BSD100, Urban100, and Manga109 show superior results, with average PSNR gains of 1.5-2.5 dB over EDSR, RCAN, and SwinIR. Ablations validate component efficacy, while sensitivity tests confirm hyperparameter robustness. This advances SISR via adaptive Transformer fusion, providing a lightweight solution for practical use.

[Application of hybrid attention guided dual-path residual learning in magnetic resonance imaging diagnosis].