Research On Medical Ultrasound Image Denoising Based On 2D-VMD And Fast Nonlocal Mean

With the development of science and technology,medical imaging technology has become an indispensable and important means for clinical medical diagnosis.Doctors can obtain the characteristics of the patient’s organs and lesions based on the medical images,so as to make targeted judgments on the condition and provide diagnosis and treatment opinions.Medical ultrasound imaging technology is widely used in abdomen,urinary,obstetrics and gynecology examinations because it can complete real-time imaging,is inexpensive and non-invasive to patients.During the ultrasound imaging process,due to its own mechanism,the image resolution is low,the contrast is low,and the edges are blurred,which greatly increases the difficulty of the doctor’s diagnosis.This problem is caused by the fact that a lot of noise is generated when the equipment is running.The thermal noise can be suppressed by cooling the equipment,but the speckle noise has to rely on computer technology for postdenoising processing.This article focuses on removing speckle noise in medical ultrasound images to complete the following research work:We have studied whether some classic algorithms and hot spot algorithms in recent years can effectively suppress speckle noise in ultrasound.Twelve denoising algorithms are listed in the article.First,select two images as test images,add speckle noise with a noise variance of 0.02-0.8 to the image,and then use these algorithms to denoise the image.Make a subjective evaluation based on the visual perception of the denoised image,and then further evaluate it through two objective evaluation indicators,PSNR and SSIM.Choose 2D-VMD algorithm and FNLM algorithm with better denoising performance as the filtering algorithm of this paper.Two-dimensional variational modal decomposition can adaptively decompose the image into a series of sub-modes with different center frequencies.Most of the main information and details of the image remain in the low-frequency mode,while the noise part is mainly concentrated in the high-frequency mode..For the test image,the noisy image is decomposed by 2D-VMD,the low-frequency modal is retained,and then the low-frequency modal image is filtered by FNLM.The experimental results show that objective evaluation indicators can fully illustrate the superiority of the algorithm in this paper,and it is found that the algorithm in this paper has more advantages in the background of high noise.When verifying the real ultrasound image,because the ultrasound image itself contains noise,there is no need to add noise,and the non-parameter evaluation index NIQE is used to evaluate the image.In order to verify the general applicability of the algorithm in this paper,four ultrasound images of the pelvis,gallbladder,abdomen,and kidney were selected for experiments.The experimental results show that the NIQE value of different parts is better than other algorithms for denoising by the algorithm in this paper.In order to facilitate the follow-up research,the algorithm of this paper is added to the medical image processing platform created by the research group using the GUI module in MATLAB to increase the diversity of the platform’s denoising function.Through the research in this article,we understand the effect of speckle noise on medical ultrasound images,and propose a new denoising algorithm,which provides a certain reference value for the advancement of medical image processing technology.