Medical image denoising is particularly important in brain image processing. Noise in acquisition and transmission degrades image quality and affects the reliability of diagnosis and research. Due to the complexity of the brain's structure and minor density differences, noise can increase diagnosis difficulty, so high-quality images are essential for disease detection, prognosis assessment, and treatment plan development. This paper proposes a multi-convolutional neural network based on feature distillation learning and dense residual attention to enhance the quality of brain images and improve denoising performance. The overall network structure contains four parts: a global sparse network (GSN), a dense residual attention network (DRAN), a feature distiller network (FDN), and a feature processing block (FPB). Before feeding the brain images into the denoising network model, they are preprocessed using a modified watershed algorithm based on a combination of a morphological gradient, Sobel's operator, and Canny's operator. The GSN is used to extract global features and increase the sensory field, and the DRAN efficiently extracts key features by combining improved channel attention and spatial attention mechanisms. The FDN extracts useful features through two feature distillation blocks, suppresses redundant information, and reduces computational complexity. The FPB performs feature fusion. Experimental results on brain image datasets and ground-based open datasets show that the proposed model outperforms existing methods in several metrics, and helps to improve the accuracy of brain disease diagnosis and treatment.

Brain images are often disturbed by noise during acquisition, compromising subsequent medical analysis and reducing diagnostic accuracy. Effective denoising that preserves the structural details is therefore essential. This essay proposes a multi-branch convolutional neural network for brain image denoising, integrating a U-shaped network, dynamic convolution, and multi-scale feature extraction. The model includes an attention-enhanced encoder–decoder to improve feature representation, dynamic convolution with a sparse mechanism for adaptive global modeling, multi-scale dense residual blocks with depth-separable convolution to capture local details efficiently, and a multi-branch fusion strategy to process features at different scales in parallel and refine them. Experiments on brain image datasets demonstrate that the proposed method achieves superior denoising performance, effectively suppressing noise while retaining fine structural details. In conclusion, the network significantly enhances the quality of brain images and shows potential for improving the accuracy of subsequent medical image analysis and diagnosis.

The fusion of infrared (IR) and visible (VIS) images aims to synthesize fused images with salient targets and enriched details. However, existing fusion methods face challenges in integrating modality-specific features. Accordingly, we proposed an image fusion method based on a dual-channel fusion strategy, termed DCGAN-Fuse. First, we created a dual-channel fusion strategy and constructed a dual-channel fusion module (DCFM) to integrate shared and complementary information across both modalities. Second, during the feature enhancement phase, we designed an attention-enhanced gradient retention module (AEGRM) to enhance edge feature extraction and enforce spatial consistency. Moreover, we used the multi-scale module (MSM) to capture fused features and avoid information loss from the two source images. We have improved the loss function by introducing the maximum intensity loss function for our proposed method. Experiments on public datasets demonstrated that our method generates fused images with highlighted infrared targets and enriched textures. Both subjective and objective assessments indicated that our DCGAN-Fuse is better than the other thirteen advanced algorithms.