57 results about "Ultrasound image segmentation" patented technology

The present invention generally relates to intravascular ultrasound (IVUS) image segmentation methods, and is more specifically concerned with an intravascular ultrasound image segmentation method for characterizing blood vessel vascular layers. The proposed image segmentation method for estimating boundaries of layers in a multi-layered vessel provides image data which represent a plurality of image elements of the multi-layered vessel. The method also determines a plurality of initial interfaces corresponding to regions of the image data to segment and further concurrently propagates the initial interfaces corresponding to the regions to segment. The method thereby allows to estimate the boundaries of the layers of the multi-layered vessel by propagating the initial interfaces using a fast marching model based on a probability function which describes at least one characteristic of the image elements.

The present invention generally relates to intravascular ultrasound (IVUS) image segmentation methods, and is more specifically concerned with an intravascular ultrasound image segmentation method for characterizing blood vessel vascular layers. The proposed image segmentation method for estimating boundaries of layers in a multi-layered vessel provides image data which represent a plurality of image elements of the multi-layered vessel. The method also determines a plurality of initial interfaces corresponding to regions of the image data to segment and further concurrently propagates the initial interfaces corresponding to the regions to segment. The method thereby allows to estimate the boundaries of the layers of the multi-layered vessel by propagating the initial interfaces using a fast marching model based on a probability function which describes at least one characteristic of the image elements.

A method for enhancing an ultrasound image is provided, wherein the ultrasound image is segmented into a feature region and a non-feature region, while sufficiently utilizing features contained in the ultrasound image, in particular including some inconspicuous features. The enhanced image according to present invention is not susceptive of the image segmentation and avoid dependence of the enhancement effect on the segmentation template, so as not to produce an evident artificial boundary between the feature region and the non-feature region but to highlight some special information in the image and to remove or mitigate invalid information. Thus the enhanced ultrasound image is particularly suitable for the visual system of the human beings.

A method for determining a candidate lesion region in a digital ultrasound medical image of anatomical tissue. The method includes the steps of: accessing the digital ultrasound medical image of anatomical tissue; applying an anisotropic diffusion filter to the ultrasound image to generate a filtered ultrasound image; performing a normalized cut operation on the filtered ultrasound image to partition the filtered ultrasound image into a plurality of regions; and selecting, from the plurality of regions, at least one region as a candidate lesion region.

This invention relates to one two-dimension fuzzy poly spot noise filter and brightness compensation B type of hypersonic cutting method, which comprises the following steps: extending the image brightness information to the update two-dimension poly fuzzy type near the pixel zone based on image brightness information; leading each heter diffusion spot noise filter to the fuzzy poly aim function to provide near information and to strengthening spot robust property; leading two-dimension fuzzy poly aim function with brightness compensation factor based on the noisy on uneven hypothesis and using image aim and background even to strengthen uneven noise robust.

The invention discloses a region growth ultrasound image automation segmentation method based on energy constraint. The method comprises the following steps: (1) preprocessing is performed on an original ultrasound image, (2) automatic selection of a seed point is performed on the ultrasound image after the preprocessing, and (3) the seed point obtained in the step (2) is used to be taken as a start point of region growth, constrained growth is performed under an energy function, the growth stops when a constraint condition is reached, and then a final segmentation result is obtained. The obtained segmentation result is accurate, the segmentation process needs no artificial participation, fully automatic ultrasound image segmentation is really achieved, and subsequent processing such as focal region characteristic extraction is facilitated.

The invention provides a method and a device for processing ultrasonic images and breast cancer diagnosis equipment. The method for processing the ultrasonic images comprises the following steps of: partitioning an ultrasonic image into a plurality of sub regions of the same type; determining the sub regions are object sub regions or background sub regions; and combining the determined object sub regions to form an object region. According to the method for processing the ultrasonic images, relatively clear object images can be obtained.

The embodiment of the invention discloses an image similarity comparison method. Through two elements of the fusion division image mass center Euclidean distance and the division image area, the automatic comparison on pictures of ultrasound image sequence groups can be realized; the pictures with poorer quality can be selected and removed. The embodiment of the invention also discloses an image similarity comparison device and a computer readable storage medium. By using the invention, the direct embedment into the ultrasound image division algorithm can be realized; the division image robustness is improved; in addition, the labor resources can be greatly saved.

The invention discloses a method for acquiring an initial contour in ultrasonic image segmentation based on an active contour model. The method comprises the following steps that the textural features of an ultrasonic image tumor area are trained, and a normal vector is formed by the textural features, a standard elliptical experience value and the tumor prior size in an ultrasonic image to be detected; the ultrasonic image to be detected is preprocessed; dynamic threshold segmentation is carried out on the image; all closed contours generated in the dynamic threshold segmentation result are extracted to form corresponding sub-images; 24 textural features of the sub-images, an ellipse fitting result parameter and the number of pixels in the closed contours of the sub-images are calculated, and a vector is formed by 26 data; the distance between the obtained vector and the normal vector obtained from the tumor area through training is calculated, and the closed contour, corresponding to the vector with the smallest distance, in the sub-images is the contour of a tumor in the segmented ultrasonic image. The method solves the problem that the ultrasonic image has much noise and is fuzzy in boundary, and the initial contour accuracy is high.

The invention is suitable for the technical field of image processing, and particularly relates to an ultrasound image segmentation method and system. The method includes the following steps that coarse registration is conducted on a statistical shape model and collected three-dimensional ultrasound data of a specified organ to obtain initialized coordinate transformation parameters; according to the initialized coordinate transformation parameters, an image segmentation algorithm based on particle filter is used for conducting iteration segmentation on the three-dimensional ultrasound data, wherein the statistical shape model is a combination of an average value obtained by training manual segmentation results of a plurality of high-definition three-dimensional data and a group of feature vectors of a representation change mode. Therefore, the problem that more artificial participation is needed in manual segmentation and semi-automatic segmentation is solved. Compared with an existing full-automatic segmentation method, the ultrasound image segmentation method and system solve the problems of low image resolution and segmentation accuracy under the image fuzzy condition.

The invention discloses a method for constructing a similarity matrix in the ultrasound image Ncut segmentation process. The method includes the following steps of firstly, preprocessing an ultrasound image; secondly, conducting over-segmentation on the ultrasound image through a simple linear iterative clustering method to generate even-medium sub-areas with irregular borders; thirdly, enabling average gray level information of all the sub-areas to serve as one characteristic for constructing the Ncut similarity matrix, and calculating the texture characteristics of all the sub-areas through a gray level co-occurrence matrix at the same time, wherein the texture characteristics totally include 24 characteristics with six characteristics in each of the four directions, and the 24 data are combined to form a texture characteristic vector, namely, another characteristic for constructing the Ncut similarity matrix; fourthly, calculating the distance between the characteristics of every two adjacent sub-areas, and combining the sub-areas according to a certain proportion to form a new Ncut similarity matrix calculation formula. When the method is used for segmenting the ultrasound image based on Ncut clustering, a good segmentation result is obtained, the problems that the ultrasound image is high in noise and low in contrast ratio are solved, and the tumor areas and the background areas can be effectively separated.

An automatic vertebral recognition method based on ultrasonic coronal plane image of a spine comprises the following steps: 1) segmenting a vertebral body one by one in target spine segment ultrasonicimage by using the ultrasonic image segmentation technology; 2) performing identifying according to the characteristic anatomical structure, the characteristic vertebral body and the characteristic structure of the vertebral body, judging and identifying the characteristic vertebral body according to the characteristic anatomical structure, and calculating other vertebral body from the characteristic vertebral body; 3) performing verifying and calibrating by characteristic anatomy and vertebral body characteristics. The invention gives consideration to the accurate and efficient identification of ultrasonic bone image features, By using the original ultrasonic image segmentation method of spine, the characteristic anatomical structure of different vertebral segments of the spine is recognized, and the characteristic vertebral body is judged by the characteristic vertebral body. Then the other vertebral bodies are deduced from the count of the characteristic vertebral body according tothe long axis direction of the spine, until the target operative segment. Finally, the characteristic anatomical structure and the characteristic vertebral body are deduced by the bidirectional count, and the identification verification and calibration are completed.

The invention discloses a breast ultrasound image segmentation method based on FCN and iterative acoustic shadow correction. The method constructs a new deep neural network in series. The network usesthe initial segmentation result of the first FCN as the initial segmentation of the acoustic shadow correction, which can effectively initialize the determined cost function to acquire the acoustic shadow field of the image and thus complete the acoustic shadow correction of the ultrasonic image, eliminate the problems of blurring the edge of anatomical layer and the poor image quality of some anatomical layers, and improve the input image quality of the second fully convolutional connected network. Then the network inputs the equalized gray-level image after correction into the second FCN toachieve final anatomical layer segmentation and a more accurate anatomical layer segmentation result. The research on the breast ultrasound image segmentation method is helpful to solve some scientific problems of ultrasound image segmentation, offers great help for the subsequent computer-aided detection and diagnosis of lesions, and has important scientific significance.

The invention discloses a support vector machine-based ultrasonic image segmentation method. The method comprises the following steps: respectively extracting sample points from the background of an image and a target of interest, estimating the variance features and the significance features of the pixel values of the ultrasonic image based on the gray values of the above sample points, and generating a sample training set; selecting a first-order linear polynomial as a kernel function, training samples by using the SVM train function in the MATLAB, and establishing a segmentation model basedon a support vector machine; segmenting a whole image according to a segmentation model for each sample on the whole image by utilizing the SVM predict function in the MATLAB, and extracting the target of interest. According to the method, the method is good in segmentation effect for low-contrast and low-resolution ultrasonic images. By adopting the method, details which cannot be distinguishedby naked eyes can be recognized and segmented. Therefore, the method is good in practicability for the diagnosis and the quantitative evaluation of medical ultrasonic images and industrial ultrasonicimages.

The invention belongs to the technical field of medical image processing, and particularly relates to an ultrasonic image segmentation method. Based on U-Net Baseline,. multiple technologies such as amulti-scale framework, a dense convolution network, an attention mechanism and small sample enhancement are fused, multi-dimensional feature extraction, irrelevant region response inhibition and small ROI performance improvement are facilitated, problems of few ultrasonic image samples, low pixels and fuzzy boundaries are solved, and the optimal segmentation effect is acquired.

A system and method for facilitating interaction by an ultrasound operator with an artificial intelligence segmentation module configured to identify and track biological and / or artificial structures in ultrasound images is provided. The method includes acquiring an ultrasound image and segmenting the ultrasound image with artificial intelligence to identify structure(s) in the ultrasound image. The method includes labeling the structure(s) in the ultrasound image to create a labeled ultrasound image. The method includes presenting the labeled ultrasound image at a display system. The method includes receiving a user selection of at least one target corresponding with at least one labeled structure. The method includes tracking the selected target(s) by identifying the selected target(s) in subsequently acquired ultrasound images.

The embodiment of the invention discloses an ultrasonic image classification device, and the device comprises a conversion unit which carries out the format conversion of an obtained ultrasonic image,and obtains a target ultrasonic image; and a residual error processing unit used for carrying out residual error processing on the target ultrasonic image to obtain a plurality of sub-images. According to the invention, residual error processing is carried out on the image, and the ultrasonic image can be segmented into a plurality of sub-images, so that the ultrasonic image can be divided into smaller image units for processing, and the accuracy of image analysis is effectively improved. Compared with the traditional convolution analysis, the residual error processing mode can reduce the data processing amount, so that the image classification processing efficiency is improved. And the determination unit analyzes the plurality of sub-images by using the trained capsule network to determine the image category to which the ultrasonic image belongs. The capsule network can obtain the correlation between the ultrasonic image and each image category by performing clustering analysis on the sub-images, so that the image category to which the ultrasonic image belongs can be evaluated more accurately.

The invention provides an artery plaque ultrasound image self-supervision segmentation method based on image restoration, which comprises the following steps: (1) preprocessing an artery ultrasound image training data set, (2) training a self-supervision auxiliary task network based on image restoration, and (3) migrating an auxiliary task model obtained in the step (2) to an artery plaque ultrasound image segmentation task, (4) training an artery plaque ultrasonic image segmentation convolutional neural network, and (5) segmenting an artery plaque ultrasonic test image by using the model obtained in the step (4), and outputting a result. The invention discloses an artery plaque ultrasonic image self-supervised segmentation method based on image restoration for the first time, artery plaque ultrasonic image segmentation under the condition of a small number of label samples is realized, and the accuracy of automatic measurement of artery plaques is improved. The method can be applied to an arterial ultrasound image auxiliary diagnosis system to monitor the growth and fading conditions of plaques, and is of great significance to early warning of heart and cerebral vessels.

The invention relates to an abdomen ultrasonic image segmentation method. The method comprises the following steps: S1, acquiring an abdomen ultrasonic image sample; S2, performing layered labeling onthe abdomen ultrasonic image sample; S3, constructing a U-NET neural network, and training the U-NET neural network by using the abdomen ultrasonic image sample subjected to layered labeling to obtain an ultrasonic image segmentation network model; and S4, based on the ultrasonic image segmentation network model, directly segmenting the abdominal ultrasonic image, and outputting a layered image of the abdominal ultrasonic image at one time. Compared with the prior art, the scale quantity and range of training samples are increased through data expansion, the step of cutting small images is avoided, meanwhile, computer storage resources are saved, end-to-end learning is conducted through the full convolutional neural network, the segmentation precision of all layers of the abdomen is higher, the segmentation results of all layers in the abdomen ultrasonic image can be output at a time, and the segmentation speed is effectively increased.

The invention is applied to the technical field of image processing, and provides an ultrasound image segmentation method and system. The method comprises steps as follows: an ultrasound image is preprocessed, and the preprocessing step comprises removal of speckle noise of the ultrasound image by establishing a multi-scale image sequence through anisotropic diffusion filters; the preprocessed ultrasound image is segmented based on a shape-constrained geodesic line active contour model. With adoption of the method and the system, the problem that a contour line is easily absorbed by noise in the evolution process and cannot reach an object boundary due to the fact that speckle noise belonging to high-frequency information is difficult to distinguish from boundary information during high-noise ultrasound image processing in the prior art, and the problem that boundary leakage is easily caused due to the fact that the boundary area with lower image contrast doesn't have enough absorption force to stop evolution of the contour line in the prior art are solved.

The invention provides a medical ultrasonic image segmentation method and device, and the method comprises the steps: obtaining a human body ultrasonic image through ultrasonic scanning equipment; training the semantic segmentation model to obtain a semantic segmentation prediction model; segmenting the to-be-detected image according to the semantic segmentation prediction model; and performing optimization processing on the segmented to-be-detected image to obtain a final target area. The medical ultrasonic image segmentation method and device have the advantages that a deep learning method and a traditional image processing method are combined, the problem that segmentation is inaccurate when the signal-to-noise ratio of the ultrasonic image is low can be effectively solved, the segmentation precision of the bladder ultrasonic image can be improved, and the anti-interference performance of the algorithm is improved.

The invention provides an ultrasonic image segmentation system and method based on a side window attention mechanism. The system comprises an ultrasonic data acquisition module, a first data transmission module, a server module, a second data transmission module and a visualization module. The first data transmission module is respectively connected with the ultrasonic data acquisition module, the server module and the visualization module; the server module comprises an image preprocessing unit and an ultrasonic image segmentation model; the ultrasonic image segmentation model comprises a convolutional neural network, a cavity convolution module, a side window attention module and a classifier; the convolutional neural network is respectively connected with the image preprocessing unit, the cavity convolution module, the side window attention module and the classifier; the second data transmission module is connected with the visualization module; according to the method, the convolutional neural network based on the side window attention mechanism is adopted, side window convolution is fused into the convolution process, the segmentation effect is more accurate in combination with the attention mechanism, and the calculation amount is greatly reduced.

The invention discloses an ultrasonic image segmentation device and an ultrasonic image segmentation method, which generate and display image data of an examination part based on volume data generatedby ultrasonic scanning of an examined object, and comprise an image enhancement module, an FRRN module and an output module which are cascaded. The technical scheme of the invention is applied to anultrasonic image segmentation task, so that the network automatically learns image features, and the segmentation precision and robustness are improved.

The invention discloses an end-to-end mammary gland ultrasonic image segmentation method based on Distance-AttU-Net. The method comprises the following steps: firstly, an original mammary gland ultrasonic image is directly used as input of a network, a trained loss value is calculated by using a self-defined loss function, and a network model is trained by optimizing the loss value and updating model parameters through back propagation until the model converges. Through the training of the network, the network itself has the capability of distinguishing the lesion from the similar regions around the lesion, and finally the lesion region in the mammary gland ultrasonic image is segmented. Through testing on the public data set B, the network model has high segmentation capability, it is proved that the mammary gland lesion segmentation model based on the ultrasonic image with excellent performance can be trained under the condition that training samples are few, and the mammary gland lesion segmentation model based on the ultrasonic image has the advantages of end-to-end segmentation, multi-scale lesion processing, high discrimination and the like.

The invention discloses a method for segmenting ultrasonic images on the basis of shape correlation active contour models. The method includes steps of 1, mining information of correlation between shape change of lesion regions in the continuous ultrasonic images and building low-rank shape correlation models; 2, building the active contour models on the basis of low-rank constraint; 3, creating augmented-Lagrange-based optimization algorithms for quickly computing segmentation results. The method has the advantage that the problem of inaccurate segmentation results of the traditional segmenting method on the basis of supervised statistical learning under the conditions of insufficient training sets, edge blur of lesion regions in existing ultrasonic images and deformation of the lesion regions in the existing ultrasonic images can be solved.

The invention belongs to the technical field of deep learning computer vision and medical information processing, and particularly relates to a medical ultrasonic image segmentation method. Accordingto the method, on the basis of a general image segmentation neural network model, multiple novel technologies such as a multiple-input-multiple-output technology, a hole convolution technology and small sample medical data enhancement are fused, the problems of difficult pain points such as small sample learning, low ultrasonic image contrast and nodule edge blurring are mainly solved, and an optimal segmentation strategy of the method is obtained.

The invention discloses an ultrasonic image segmentation method. Implementation steps are as follows: implementation steps are as follows: first of all, ultrasound images are read, secondly, manuallyinitializing an edge contour curve; An edge detection function of an image and a balloon force factor of an evolution model are constructed by using a multi-scale filtering method; according to the ultrasonic image edge contour segmentation method, the multi-scale filtering method and the improved balloon force factor are used, the edge leakage phenomenon caused by the weak edge problem in the ultrasonic image is avoided, and the segmentation precision of the ultrasonic image is improved.

The invention discloses a prostate ultrasonic image segmentation and classification method. The method comprises the following steps: establishing an image segmentation model; establishing an image classification model; combining the image segmentation model with the image classification model to form an image segmentation classification model; and inputting the acquired prostate ultrasonic imagesinto an image segmentation and classification model for classification. Compared with the prior art, the method can improve the classification precision of the image, and improves the later diagnosisefficiency and accuracy.