Research On Key Technologies Of Ultrasound Image Segmentation Based On Threshold Selection

Medical ultrasound image is the main image in medical imaging.Due to the low radiation,safety and low price,they are widely used in clinical practice.Ultrasound image segmentation is an important part of ultrasound image processing.It is a key step in qualitative and quantitative analysis of ultrasound images organization information,which plays an important role in clinical imaging diagnosis.At present,in ultrasound image segmentation,target region extraction analysis mostly relies on manual and semi-manual operations.With the development of medical intelligence,computer-aided segmentation,and feature analysis of targets will become the direction of ultrasound image processing research.However,due to the influence of the imaging mechanism,the ultrasound image contains large speckle noise and the edges are blurred,which makes the ultrasound image segmentation a more complicated and challenging task.At present,methods for segmentation of ultrasound images include threshold,region growing,active contour,K-means,machine learning and neural networks.The threshold method is a simple and easy method among them,which has been widely used in ultrasound image segmentation.After analyzing the characteristics of ultrasound images and related segmentation methods,this paper proposes a method of ultrasound image segmentation with multilevel threshold based on differential search algorithm.The method adopts multilevel threshold between-class variance and differential search algorithm to obtain appropriate thresholds for ultrasound image segmentation and achieve fast segmentation.At the same time,the ultrasonic image is filtered by the bilateral filter before the image segmentation,to improve the segmentation effect.From the experimental results and data,multilevel threshold based on differential search algorithm can segment the ultrasound image better,extract special tissues,and has higher accuracy and segmentation efficiency than other methods.In order to achieve more accurate segmentation of the ultrasound image lesion area,this paper continues to introduce the CV active contour model,and combined with multi-threshold segmentation,proposes an ultrasound image lesion segmentation method based on the combination of threshold and active contour.In this method,the ultrasound image is segmented by threshold,morphological processing,to locating and generating initial contour of the lesion to the target lesion area automatically,and the initial contour is used for segmentation of the CV active contour model.At the same time,in order to improve the accuracy of the contour of the lesion area,this paper improves the CV active contour model,performs boundary processing in the CV model segmentation and introduces the local CV model segmentation method to overcome the gray level unevenness of the ultrasound image.The experimental results show that the ultrasonic image segmentation combined with the threshold and the active contour can select target,segment the region and extract edge contour automatically.Compared with region growth,k-means and CV model segmentation,the proposed method has obvious improvement in segmentation accuracy rate.