Research on spinal ultrasound image denoising based on an improved BM3D algorithm

Ultrasound imaging is widely utilized in clinical diagnosis due to its non-invasiveness, real-time capability, and cost-effectiveness, especially in the assessment of the spine, muscles, and joints. However, ultrasound images are often corrupted by noise—particularly speckle noise—caused by factors such as device settings, tissue structure, and operator variability. This significantly degrades image quality and diagnostic accuracy. To improve the clarity of spinal ultrasound images, this paper proposes an improved image denoising method based on the Block Matching and 3D Filtering (BM3D) algorithm. The method utilizes adaptive similar block searching and sparse processing of flat image regions to enhance denoising efficiency. Experimental results show that the proposed method can effectively and rapidly remove noise from ultrasound images while preserving structural edges and detailed information.