85 results about "Tissue segmentation" patented technology

The present invention discloses a method and a system for melanoma image tissue segmentation based on a deep neural network. A deep neural network technology is employed to solve the tissue segmentation problem of dermatoscope images of melanoma, and the method and the system for melanoma image tissue segmentation based on a deep neural network are mainly a medical image analysis processing technology. An improved deep neural network structure to perform modeling, dermatoscope images having segmentation tags are employed to train the model, and the trained model has an ability of suspicious tissues to new dermatoscope images. The method and the system are to locate suspicious areas in the melanoma dermatoscope images and to perform pixel segmentation. The method and the system for melanomaimage tissue segmentation based on the deep neural network employ a new deep learning technology, fully develop the capacity of collection of various hierarchical features of the image data, can be applied to the modeling process, can perform location and segmentation of the suspicious skin tissues and can provide good reference for further analysis by dermatologists.

The invention discloses a deep-network-based tissue segmentation method of a panoramic digital colorectum pathology image. The method comprises the following steps: step one, acquiring a panoramic digital colorectum pathology image; step two, segmenting the panoramic digital colorectum pathology image; step three, establishing a training sample image; step four, extracting deep features of tissuesof different types; step five, determining types of tissues in the segmented images by using a classifier based on the extracted deep tissue features; step six, splicing image classification resultsand determining a tissue class of an overall picture; and step seven, splicing the images according to block coordinates. According to the invention, the panoramic digital colorectum pathology image is segmented and tissue type marking is carried out on all segmented images by using a sliding window and a trained model; and tissue type determination is carried out by using the classifier based onthe extracted deep tissue features to obtain an image classification result. The classification becomes accurate and the classification speed is increased.

The invention discloses a ling tissue segmentation method, comprising: obtaining a first medical image of the target area of an examinee wherein the first medical image consists of a plurality of layered images with each layered image containing a plurality if pixels; linearly enhancing the sagittal plane of the first medical image; adding the gray values of the pixels corresponding to the first medical image and the linearly enhanced first image to generate a second medical image; determining the sagittal skin line of the second medical image and determining the cross-sectional skin line of the second medical image on the sagittal skin line determined second medical image; obtaining the connected domain enclosed by the sagittal skin line and the cross-sectional skin line; and obtaining the lung tissue segmentation result in the connected domain. The lung tissue segmentation method of the invention can accurately obtain the boundary area of a lung tissue. At the same time, the invention also proposes lung tissue segmentation apparatus and a medical image processing system using the lung tissue segmentation apparatus.

A method to segment images that contain multiple objects in a nested structure including acquiring an image; defining the multiple objects by layers, each layer corresponding to one region, where a region contains an innermost object and all the objects nested within the innermost object; stacking the layers in an order of the nested structure of the multiple objects, the stack of layers having at least a top layer and a bottom layer; extending each layer with padded nodes; connecting the top layer to a sink and the bottom layer to a source, wherein each intermediate layer between the top layer and the bottom layer are connected only to the adjacent layer by undirected links; and measuring a boundary length for each layer.

The invention provides a bone tissue geometrical morphology parameter automatic measuring device and a method based on the image processing technology. A bone tissue X-ray film is shot through a medical x-ray machine. The bone tissue X-ray film is subjected to pre-treatment, bone tissue segmentation and bone tissue parameter measurement by a computer, and a bone tissue parameter measurement result report is formed, transmitted and printed. A user only needs to select the name of a bone tissue and to-be-measured parameters before treatment, and segmentation and measurement processes are completely and automatically completed. Meanwhile, the initial contour selecting or labeling is not required for a doctor. The bone tissue geometrical morphology parameter automatic measuring device comprises an X-ray film data input interface unit, an image processing unit, a measurement parameter storage and output unit, a network interface unit and a printer interface. According to the invention, bone tissues in the X-ray film can be rapidly and automatically segmented and measured. By adopting the existing advanced GPU (graphics processing unit) and other hardware equipment, the bone tissues in the X-ray film can be automatically segmented and measure based on the image processing technology. Therefore, the film-reading automation level and the film-reading intelligence level of the doctor are improved.

A method and apparatus for processing medical images from one of a plurality of digital acquisition modalities or manufacturers with different imaging condition is proposed for creating consistent appearance of the images. The method comprises (1) tissue segmentation to isolate the region of interest; (2) dynamic extraction of the optimal parameters for image transformation from the segmented region; (3) generation of a transformation function from the individual image optimized parameters; and (4) use of the transformation function to produce images that have consistent image characteristics. This method also applies to multiple images from a single study or multiple studies. The transformed images can be used for computer-aided lesion detection, review and diagnosis.

The invention discloses a medical image segmentation method, device and equipment and a readable storage medium. The method comprises the steps: obtaining a to-be-segmented medical image, and inputting the medical image into a deep learning image segmentation network; carrying out tissue segmentation on the medical image by utilizing the target segmentation parameter which focuses on the specifiedtissue edge to obtain an image segmentation result; the training process of the deep learning image segmentation network comprises the following steps: determining an edge enhancement region of a medical image training sample by utilizing a segmentation label corresponding to the medical image training sample; inputting the medical image training sample into a deep learning image segmentation network for tissue segmentation to obtain a training segmentation result; when a loss function is used to calculate a loss value, increasing a loss weight for a pixel corresponding to an edge enhancementregion in the training segmentation result; and adjusting the segmentation parameter of the deep learning image segmentation network by using the loss value to obtain a target segmentation parameter.The method can improve the medical image segmentation precision.

The invention relates to an individualized brain covariant network construction method based on three-dimensional textural features, which comprises the following steps of: 1) segmenting a brain structure image into brain tissue component concentration graphs by using tissue segmentation, and registering the brain tissue component concentration graphs to a standard space template to obtain a standardized brain structure image; 2) extracting three-dimensional texture features corresponding to the standardized brain structure image at a voxel level through at least two gray feature extraction modes, and obtaining a spatial distribution diagram of each texture feature; and 3) defining a brain region map as a network node, extracting the texture feature of each brain region of the individual subject from the grayscale matrix texture feature data set, calculating the Pearson's correlation of the texture feature vectors of any two brain regions, and constructing a covariant matrix of the texture features between the brain regions. According to the method, brain image data of an individual subject can be utilized, the similarity of brain region texture feature vectors serves as measurement of a brain network edge, and then a brain covariant network of the individual subject is constructed.

The invention provides a target classification system. The target classification system can specifically be a knowledge data and artificial intelligence driven multi-disease identification system. The system can be used for: acquiring medical images and non-image medical information; using a medical expert knowledge base for matching information data, and obtaining an ophthalmic disease weight result through a medical knowledge reasoning algorithm; obtaining an ophthalmic disease weight result by using an organ tissue segmentation algorithm and a disease focus recognition algorithm in medical image diagnosis; obtaining weight results by combining knowledge reasoning and medical image recognition, and obtaining a final disease diagnosis result through weighted calculation. According to a medical clinical diagnosis thought, structured processing is carried out on medical data through big data, and automatic diagnosis of diseases is realized through the detection capability of an artificial intelligence model on disease focuses. The system improves the existing doctor seeing mode, realizes artificial intelligence preliminary diagnosis and screening of diseases, effectively relieves the current situation of shortage of medical resources, and has a wide application prospect.

The invention discloses a UNET-based cervical pathological tissue segmentation method. The method comprises five steps of scanning splicing, noise reduction, region division, UNET model training and auxiliary diagnosis. According to the method, the UNET (a convolutional neural network model adopting a deep automatic coding and decoding structure) is applied to effectively solve the common problemsthat in the field of cervical tissue pathology digital image intelligent analysis, the number of samples is small, the standardization of the samples is poor, and the real-time performance of sampleprocessing is poor, so that clinical application is hindered. The method can also help doctors to quickly and accurately identify diseased tissues and give auxiliary diagnosis suggestions, so that notonly is the workload of pathologists greatly reduced, but also the detection cost is remarkably reduced, and the method has huge economic benefits and social benefits.

Methods and systems for automated target and tissue segmentation using multi-modal imaging and ensemble machine learning models are provided herein. In some embodiments, a method comprises: receiving a plurality of medical images, wherein each of the plurality of medical images includes a target and normal tissue; combining the plurality of medical images to align the target and normal tissue across the plurality of medical images; inputting the combined medical images into each of a plurality of machine learning models; receiving, in response to the input, an output from each of the plurality of machine learning models; combining the results of the plurality of machine learning models; generating a final segmentation image based on the combined results of the plurality of machine learning models; assigning a score to each segmented target and normal tissue; and sorting the segmented targets and normal tissues based on the scores.

The present invention relates to a method and system for enhancing resolution of a magnetic resonance image of an object. The method combines information from a plurality of low-resolution images with a Field Of View shifted by a distance less than a pixel width to create a synthesized image having substantially improved image quality. Information from the low-resolution images is merged and application of an aperture function enhances the SNR of the synthesized image resulting in synthesized images having a substantially higher spatial resolution as well as a substantially increased SNR. The method and system for enhancing resolution of a magnetic resonance image is highly beneficial for a MRI practitioner by substantially improving image quality, thus facilitating diagnostic methods such as texture analysis and disease specific tissue segmentation.

The invention relates to a spine CT sequence image segmentation method and system, the method comprises a training stage and a test stage. The training stage comprises the following steps: (A1) carrying out manual annotation, (A2) preprocessing a data set, (A3) constructing a global semantic segmentation network and a local semantic segmentation network, and (A4) training the global semantic segmentation network and the local semantic segmentation network. The test stage comprises the following steps: (B1) acquiring a CT sequence image to be segmented, (B2) carrying out image preprocessing, (B3) performing global semantic segmentation on a bony structure and a non-bony tissue in the CT sequence image, (B4) carrying out local semantic segmentation on various non-bony tissues in the spine core section, and (B5) synthesizing and obtaining a segmentation result. Compared with the prior art, the method can achieve the quick and automatic segmentation of the bone structure and various typesof non-bone tissues in the spine CT sequence image, and is accurate and reliable in segmentation result.

The invention belongs to the technical field of medical image processing, and particularly relates to a multi-channel head magnetic resonance imaging tissue segmentation method. However, the existingdeep learning method does not utilize anatomical structure information of relatively fixed brain. The invention provides a multi-channel head magnetic resonance imaging tissue segmentation method, which comprises the following steps of: 1, matching a pre-segmentation label most similar to each image to form four channels; And 2, inputting the obtained four channel data into a convolutional neuralnetwork, training the network through a real label of the input data to obtain a training model, and enabling the test data to pass through the trained model to obtain a segmentation result. The priortexture information of brain tissue is fully utilized, a new channel is added, accurate segmentation of the network is promoted, and the network segmentation precision is improved. The method provided by the invention is simple and high in robustness, and can be added to any segmented network for segmentation without changing the original network structure.

The invention discloses an Alzheimer's disease classification method and system based on an anatomical landmark and a residual network, and the method comprises the steps: carrying out the tissue segmentation modulation of a training image set, and obtaining a gray matter image; comparing voxels of the Alzheimer's disease image and the normal subject image in the training image set to identify ananatomical landmark; taking the obtained anatomical landmark as a center, and extracting grey matter blocks of the gray image; and connecting the grey matter blocks, inputting the obtained synthetic blocks into a residual error network model for feature extraction, taking features extracted by the residual error network model as input of a classifier, and classifying Alzheimer's disease patients and normal subjects in the test image set. The anatomical landmark is taken as the center, feature blocks of gray matters in three tissues of the brain are taken as the input of the residual network, the feature blocks are taken as the feature expression of each MR image, the residual network model is adopted to enhance the feature learning capability of the network, and the classification accuracyis enhanced.

The invention provides a slice tissue identification method and device, a cell identification model and a tissue segmentation model. The method comprises the following steps: acquiring a to-be-identified slice image; inputting the to-be-identified slice image into a cell identification model, and obtaining the identified cell type and the identified cell position information; inputting the to-be-identified slice image into the tissue segmentation model to generate a first image; wherein the first image comprises an identified tissue area; combining the identified cell type and the identified cell position information with the first image to generate a second image; wherein the second image comprises the identified tissue area and the cell type in the identified tissue area; and classifying the tissue areas identified in the second image according to the cell types in the tissue areas identified in the second image. Through the above mode, the workload of a pathologist is reduced, and the identification efficiency and the identification accuracy of the slice tissue are improved.

he invention provides a method and apparatus for segmenting a tissue in a CTA image. The method comprises: pretreatment is carried out on a CTA image; a watershed operation is carried out on an original image after pretreatment to determine basins, and combination information between the basins and statistic information of the basins are recorded; boundary distance values of specific pixel points are obtained, tissue identification is carried out on the basins based on the boundary distance values and the statistic information of the basins, the parts of basins are marked as bone basins or blood vessel basins; according to the recorded combination information between the basins, unmarked basins are combined with the marked basins; and according to the combination result, a morphological operation is carried out and a tissue segmentation result of the CTA image is displayed. With the method, the work load of the user can be reduced and the user operation becomes convenient.

The invention discloses a medical image feature recognition prediction model which comprises the following steps: constructing a multi-tissue segmentation model based on a deep convolutional neural network, obtaining data from a sample library, extracting image features, obtaining an image segmentation result based on pyramid up-sampling, and segmenting different tissue regions; constructing a cell detection model based on the deep convolutional neural network, firstly magnifying times to take blocks, and then standardizing images of the taken small blocks; sending the cells into a regression detection module to detect the image in each small block, and then cascading the detected cells with a deep classification network to obtain an interested target; and constructing a visual sub-vision module, and selecting the features with the most predictive capability from the features in all different tissue areas and interested objects by using a method of combining a feature selection method and cross validation. According to the invention, multi-tissue segmentation is carried out on the pathological image, cells are accurately identified, and a doctor is assisted in reading through sub-visual features.

The invention discloses a semi-supervised learning image segmentation method and system, and a terminal. The method comprises the following steps: training an image segmentation network by using a first data set of an image with a label; inputting the data set of the non-label image into the trained image segmentation network to obtain a pseudo-label data set corresponding to the non-label image; combining the pseudo-label data set and the initial data set with the tag image to obtain a second data set; and training the image segmentation network again by adopting the second data set, inputting the next batch of label-free image data set after training is completed, and predicting to generate false labels. After a pseudo tag is generated and expanded to a training set every time, due to the fact that new training data is added, the network continuously learns new features until a trained segmentation network is finally obtained. According to the method, a small amount of marked data and a large amount of unmarked data are jointly used to effectively train the segmentation network, so that the purpose of improving the tissue segmentation precision is achieved, and the dependence of a deep learning image segmentation method on label data is reduced.

The embodiment of the invention provides a rib display method and device, and the method comprises the steps of carrying out the tissue segmentation processing of a chest three-dimensional scanning image, obtaining a three-dimensional rib image in the chest three-dimensional scanning image, wherein the three-dimensional rib image comprises at least one pair of ribs, and the at least one pair of ribs comprise a left rib and a right rib which are matched with each other; based on the three-dimensional rib image, extracting rib center lines of the ribs, wherein each center point on the rib center lines is located in the center of the rib section; for each central point, sampling processing along the target direction of the central point, determining rib pixel points in the three-dimensional rib image, wherein the target direction is the radial direction of a rib section corresponding to the central point; and generating a two-dimensional rib display image based on the rib pixel points. The ribs can be automatically displayed on a two-dimensional plane under the condition that the basic bending deformation of the ribs is reserved, the diagnosis time can be saved, and the workload of doctors is reduced.

The invention relates to an artificial intelligence-based anatomical variation recognition prompting method and system, and the method comprises the steps: collecting an endoscope image in real time, and obtaining real-time organ segmentation data and instrument key point data according to an endoscope organ segmentation model and an instrument key point detection model; acquiring a position relationship between the variation structure and the surrounding organs by analyzing the preoperative iconography image, extracting the tissue segmentation data of the surrounding organs of the variation structure of the iconography image for comparison, and judging the position of the variation structure; detecting the position information of the key points of the instrument and the variation structure in real time, prompting the variation anatomical structure area when the instrument operates in the variation anatomical structure area, and operating according to the prompt. According to the method, by establishing the corresponding relation between the imaging examination and the endoscopic surgery visual field, the vein pipeline structure of the key operation in the surgery is accurately and effectively positioned, and important conditions are provided for more accurate operation of the surgical operation.

The invention relates to a method for extracting the three-dimensional tooth root morphology of an intravital tooth based on periodontal membrane imaging anatomical characteristics. The method comprises the steps of performing imaging scanning on a patient, and storing the scanned data to be a DICOM-format body layer data; importing the DICOM-format body layer data of the patient into Mimics19.0 software, and extracting a discontinuously distributed periodontal membrane mask from the image data; enabling the discontinuous distribution to be expanded into continuous distribution; taking a seedmask as a seed region to expand outwards to the root bone boundary, wherein the maximum expansion step does not exceed an artificially set maximum step parameter; and rebuilding a three-dimensional model of the corresponding tooth, and storing and outputting the three-dimensional model in an STL format to complete segmentation and extraction for the intravital tooth. The method is simple and easyto use, and can quickly realize tissue segmentation for the intravital tooth root.

A regional brain age prediction method comprises the following steps: receiving a first T1 weighted image of the brain of a to-be-tested person; performing a tissue segmentation procedure on the firstT1 weighted image to obtain a first gray image; registering the first gray matter image to a normal subject image template established by the superposition program; capturing a first gray matter structure characteristic of the to-be-detected brain area; predicting the predicted age of the to-be-tested brain area through the optimized model parameter value corresponding to the to-be-tested brain area and a machine learning algorithm or a deep learning algorithm of the optimized model parameter value according to the first gray matter structure characteristic of the to-be-tested brain area; andobtaining an age difference value corresponding to the to-be-tested brain area according to the predicted age of the to-be-tested brain area and the actual foot age of the to-be-tested person, and outputting an evaluation report. Therefore, an objective biometric indicator can be provided to detect possible changes in different brain regions.

The invention relates to an embryonic tissue segmentation method based on a generative adversarial network, and belongs to the technical field of medical image processing. The method comprises the steps of step 1, performing tissue segmentation mask mapping on an embryonic tissue slice image through a U-net network; step 2, making a segmentation network training set; step 3, configuring parametersrequired by network training to obtain a set network; step 4, training the set organization quality identification network by using the manufactured segmentation network training set; step 5, fixingparameters of the organization quality identification network, and training the set U-net network by using the manufactured segmentation network training set in combination with the organization quality identification network; and step 6, taking the embryonic tissue slice image without the marked segmentation result as input, and generating a corresponding mask image. The network relied on by thesegmentation method uses a classification model to supplement loss during training and segmentation, fully utilizes the information of the cell growth state, and improves the accuracy of the segmentation network in the field of embryo tissue segmentation.

The invention discloses an orbital bone tissue segmentation method based on body registration. The method comprises: acquiring a CT image needing orbital bone tissue segmentation; processing the CT image and the tetrahedral mesh model standard data of the orbital bone tissue; performing initial alignment on the processed data; transforming the tetrahedral mesh data after initial alignment; whereinthe transformed tetrahedral mesh data is final orbit bone tissue data obtained by segmentation from the CT image. According to the orbital bone tissue segmentation method based on body registration provided by the invention, innovative down-sampling operation is adopted in a data processing stage, so that the influence of noise data on registration and segmentation results is reduced while the sampling efficiency is improved. Meanwhile, shape deformation is completed in an error driving mode in the transformation process, the segmentation effect of orbital bone tissue is improved, and the method is high in segmentation precision, simple and rapid.

The invention relates to an atlas-free brain tissue segmentation method. An MRI sequence configured for acquiring two image volumes of the part at different inversion times within a single acquisition is combined to a fat-water separation method for acquiring a fat-water separated image. For each echo time two image volumes are acquired, respectively a first image volume and a second image volume at the first echo time, and a first image volume and a second image volume at the second echo time, and combined to a uniform image. The acquired images are combined to form a final uniform image, a final fat-water separated image, and a final second image volume that are fed into a multichannel image segmentation algorithm using a Markov random field model for segmenting the part into multiple classes of cranial tissues, in order to obtain a segmented image of said part.

The invention relates to the technical field of medical image processing, and discloses a lumbar vertebra bone analysis method and device based on deep learning, equipment and a storage medium. The method comprises the following steps: acquiring training sample data, training a lumbar vertebra segmentation model through the training data, and identifying a target CT sequence image through the trained lumbar vertebra segmentation model; then obtaining a segmentation result of each vertebral block in the abdomen area and a category of the corresponding vertebral block; and carrying out tissue segmentation to obtain a tissue segmentation result, and based on the tissue segmentation result, carrying out calculation processing on the cross section area and the tissue structure volume to obtain a target result. The invention further relates to a block chain technology, and the target CT sequence image is stored in the block chain. According to the method, accurate positioning and tissue segmentation are carried out on the lumbar vertebra region of the target CT sequence image through the trained lumbar vertebra segmentation model, so that the accuracy of a quantification result of a tissue corresponding to the lumbar vertebra region is improved.

The invention discloses an image three-dimensional tissue segmentation and determination method based on a deep neural network, and the method comprises the steps: collecting a CT image of a living pig, dividing the CT image into a training set and a test set, and marking the training set; constructing a CT bed segmentation network and a viscera segmentation network, and performing training by utilizing the marked training set to obtain a CT bed segmentation model and a viscera segmentation model; utilizing the CT bed segmentation model and the viscera segmentation model to predict a mask mapand remove the CT bed and viscera; and extracting fat, muscle and skeleton parts of the pig body by combining the CT image of the living pig, and calculating the total mass of the pig body and the proportion of each tissue. The method can automatically, quickly and accurately segment fat, muscles, bones and other tissues of the breeding pigs, and is suitable for breeding pigs of any shape and anysize.

The present invention relates to a method for visualizing ST data based on principal component analysis. ST data indicative of a plurality of local S spectra, each local S spectrum corresponding to an image point of an image of an object are received. In a first step principal component axes of each local S spectrum are determined. This step is followed by the determination of a collapsed local S spectrum by projecting a magnitude of the local S spectrum onto at least one of its principal component axes, thus reducing the dimensionality of the S spectrum. After determining a weight function capable of distinguishing frequency components within a frequency band a texture map for display is generated by calculating a scalar value from each principal component of the collapsed S spectrum using the weight function and assigning the scalar value to a corresponding position with respect to the image. The visualization method according to the invention is a highly beneficial tool for image analysis substantially retaining local frequency information but not requiring prior knowledge of frequency content of an image. Employment of the visualization method according to the invention is highly beneficial, for example, for motion artifact suppression in MRI image data, texture analysis and disease specific tissue segmentation.