Design And Implementation Of Denoising And Enhancing System For Heart Ultrasound Images

Ultrasound images of human fetal hearts are important for medical applications such as the diagnosis of congenital heart disease and the evaluation of early fetal heart function.However,due to the particularity of embryo ultrasound imaging,ultrasound images have high noise and low contrast.Therefore,it is very important to denoise and enhance it.This paper designs a special method framework that includes denoising and enhancement.In the denoising part of this framework,we first designed a 4D-NLM(non-local mean)denoising method for 4D fetal heart ultrasound image sequences.It makes full use of similar information between adjacent images to reduce the noise of the target image.Afterwards,we used a denoising method of deep convolutional neural network for the long running time of 4DNLM and no 4D sequence images,and the method of expanding the data set using deep convolutional generation adversarial network.In terms of enhancement,we designed an enhancement method called "Adaptive Clipping of Each Histogram Pillar"(ACEHP)to enhance myocardial space to distinguish it from blood space.Denoising and enhancement experiments show that the 4D-NLM algorithm has better denoising effects than several classic and latest technologies such as NLM and WNNM.DnCNN using 4D-NLM dataset or DCGAN dataset can also achieve better denoising effect.Similarly,compared with several classic and latest algorithms such as CLAHE and SVDDWT,the ACEHP method can obtain better enhancement effects while ensuring a low noise level of the image.In addition,in the volume rendering of the image processed by the "4D-NLM + ACEHP" combination,the heart cavity borders are neat and clear.At the same time,we completed the design and implementation of human fetal heart ultrasound image denoising and enhancement system based on Qt Creator.This system can realize the denoising and enhancement of human fetal heart ultrasound image,display the image processing results in time,and also realize the image batch processing.It provides convenience for researchers and medical staff.