57 results about "Lesion Identification" patented technology

A method for using computer-aided diagnosis (CAD) for digital tomosynthesis mammograms (DTM) including retrieving a DTM image file having a plurality of DTM image slices; applying a three-dimensional gradient field analysis to the DTM image file to detect lesion candidates; identifying a volume of interest and locating its center at a location of high gradient convergence; segmenting the volume of interest by a three dimensional region growing method; extracting one or more three dimensional object characteristics from the object corresponding to the volume of interest, the three dimensional object characteristics being one of a morphological feature, a gray level feature, or a texture feature; and invoking a classifier to determine if the object corresponding to the volume of interest is a breast cancer lesion or normal breast tissue.

The invention discloses an intelligent film reading system for tumor image data. One-key film reading report writing is realized, steps such as image data preprocessing, lesion identification, quantitative parameter measurement, diagnostic scoring and visualization are completed automatically by a computer, the operation procedure and the diagnostic process are simplified, the advantages of a doctor and the computer are executed to the maximum extent, the current image film reading working efficiency is improved significantly, the film reading result is objective and accurate, and the repeatability is high.

The invention discloses a wireless capsule endoscope small intestine lesion computer auxiliary detection system and method. At present, the technology of the capsule endoscope in the aspect of intelligent lesion identification and accurate positioning is very limited. The data input module is used for acquiring the video data of the wireless capsule endoscope of the patient, and an image of the capsule endoscope is obtained by extracting a video frame technology; the image preprocessing module is used for preprocessing the image of the wireless capsule endoscope, the small intestine image recognition module is used for recognizing and extracting a small intestine image and an image sequence in the preprocessed capsule endoscope image; the small intestine lesion analysis and positioning module is used for identifying and classifying the small intestine lesion and extracting the specific position of the focus; the user interaction module forms an auxiliary diagnosis result according to the small intestine lesion information and the analysis result of the positioning module, and the doctor can confirm, modify or input the medical advice so as to form a diagnosis report. According to the invention, the practical value of the capsule endoscope in clinic is promoted, so that a more efficient and more standard diagnosis mode is formed.

The invention discloses a coronary artery virtual stent implantation method and system based on haemodynamics analysis. The coronary artery virtual stent implantation method comprises the steps that CTA image data of a patient are read, plaque is identified, a vascular three-dimensional image model is constructed, the plaque is removed, and the position of the plaque is recorded; the physiological parameters of the patient is received, a CFD boundary condition is set, and CFD calculating is conducted; according to a three-dimensional vascular form, a plaque characteristic and FFR values throughout blood vessels, a lesion is identified; according to the identified lesion, a stent specification database and a virtual stent implantation strategy, an implantation scheme is generated, and a new vascular three-dimensional geometric model is generated; under the new three-dimensional geometric model, fluid calculation is conducted again, according to a preset selecting standard, an optimal stent implantation scheme is output. According to the coronary artery virtual stent implantation method and system, blood vessel stenosis degree calculating, stent implantation strategy generating and quantitative evaluating of the virtual stent implantation effect are completed automatically, the operation scheme is planned accurately, the doctor decision efficiency is improved effectively, and the risk of dependence on the artificial judgment is reduced.

A lesion identification system for surgical operations has a light marker adapted for placement in the vicinity of a lesion inside an organ of an organism, and a location-identifying device detecting a light emitted from the light marker at an outside of the organ for identifying a location of the lesion. The marker includes a light emitter and an engagement member associated with the emitter to be engageable with a wall of the organ. The engagement member includes a clip. The light emitter emits light with a wavelength in a near-infrared range. The device includes an endoscope comprising an inserter section having an image pickup element picking up a reflected light, involving the light from the light marker, at an outside of the organ, and an image pickup unit allowing the reflected light, picked by the image pickup element, to be generated as an image for display on a monitor.

The invention relates to the field of medical image processing, in particular to a method for automatically identifying a liver tumor type in an ultrasonic image. The method can provide auxiliary diagnosis when a doctor diagnoses a liver lesion according to a CEUS medical image, wherein the auxiliary diagnosis comprises the process of giving out a lesion identification result, the time when the lesion is relatively remarkable and a position where the lesion is relatively remarkable. The method specifically comprises: representing a CEUS image with a plurality of regions of interest (ROI); distinguishing different lesions by expressions and changes of the ROI in time and space; and representing a space-time relationship among the ROI by establishing models in time and space at the same time, wherein the models can determine relatively proper ROI and related parameters of the models according to existing CEUS lesion samples with an iterative learning method. After a sample is given out, the proper ROI can be determined by removing part of improper ROI in advance and a fast search method, and reference diagnosis is given out for the lesion.

The invention provides a lung detection method and device based on PET / CT image features. The method comprises the steps that the local image of a region corresponding to a lesion is acquired from a PET / CT image library as a training image; the PET / CT image library comprises the lung lesion images of multiple types of lesions; the training image is transformed and filtered to acquire a processed training image; the feature information of the processed training image is extracted, wherein the feature information comprises one or combination of texture feature information and color feature information; a lesion detection model is established according to the feature information; and a lung image to be detected is detected based on the lesion detection model to determine the type of the lesion in the lung image to be detected. An image processing technology is used to carry out lesion analysis and identification on a PET / CT image, which improves the accuracy and sensitivity of lesion identification.

The invention discloses a rice lesion detection method and system based on deep learning, belonging to the image processing field, the method comprising: providing a photo sample set and a manual labeling sample set, and cutting the photo sample set and the manual labeling sample set according to a proportion to form a second photo sample set and a second manual labeling sample set; inputting thesecond photo sample set and the second label sample set into the Linknet network model, and obtaining the optimal model by training the Linknet network model based on the Pytorch deep learning framework; using the optimal model to identify the rice lesion images needed to be detected at present, and calculating the proportion of rice lesion area and classifying the disease status. Through the Linknet network model of Pytorch deep learning framework, the generalization ability and field practicability of rice leaf lesion identification can be improved, and the utilization rate of information can be improved, which is conducive to the subsequent quantitative application of pesticides and reduce environmental pollution.

The invention discloses a method and a system for recognizing a focus of infection by perfusion imaging. The method comprises the following steps: analyzing features of a plurality of original perfusion images obtained by repeated scanning of a recognizing system on the same symmetric part layer of a patient to obtain a perfusion parameter diagram of the symmetric part layer; analyzing perfusion parameters in the perfusion parameter diagram to obtain a region of the focus of infection positioned in the symmetric part layer; determining a reference region which is symmetric with the region of the focus of infection by symmetry of the symmetric part layer; comparing symmetric perfusion parameters in the region of the focus of infection and the reference region to determine different lesion degrees in the region of the focus of infection. By lesion recognition of a normal region and the region of the focus of infection of the patient, the problem that the focus of infection is recognized inaccurately due to individual difference of patients is solved effectively.

The invention provides a lesion identification method including the steps that a first magnetic resonance image and a second magnetic resonance image obtained by magnetic resonance scanning of the preset part of the human body with different magnetic resonance scanning sequences are acquired; image registration is performed on the first magnetic resonance image and the second magnetic resonance image; using the trained regression convolutional neural network model to detect the first preset region in the first magnetic resonance image after registration and the second preset region in the second magnetic resonance image after registration; the first convolutional neural sub-network of the trained dual-trained convolutional neural network model is appled to predict the first lesion probability of the first preset region and the second lesion probability of the second preset region; and whether the preset part is the lesion or not is judged according to the first lesion probability and the second lesion probability. The invention also provides a lesion identification device, a computer device and a computer readable storage medium. The invention can realize high-accuracy lesion identification.

The invention discloses a fundus illumination multi-disease detection system based on a regional feature set neural network. The system comprises a computer memory, a computer processor and a computerprogram which is stored in the computer memory and can be executed on the computer processor, wherein a trained multi-fundus disease detection network model is stored in the computer memory, and themulti-fundus disease detection network model comprises a feature extraction network, a semantic segmentation sub-network and a plurality of classifiers; when the computer processor executes the computer program, the following steps are realized: obtaining a to-be-detected original fundus picture, inputting the to-be-detected original fundus picture into the multi-fundus-disease detection network model, obtaining the probability of each type of lesion, and realizing fundus lesion identification. According to the invention, a single model can be used for detecting various fundus diseases, and the detection precision is high.

Method and system for grading a medical image. For example, a system for grading a medical image comprising a grading network configured to provide a grading result corresponding to the medical image based on at least the medical image and / or a list of lesion candidates generated by a lesion identification network.

A cancerous lesion identifying method via hyper-spectral imaging technique comprises steps of: acquiring a plurality of first pathology images via an endoscopy, wherein the first pathology images are cancerous lesion images respectively; importing the first pathology images into an image processing module to acquire a plurality of first simulating spectra of the first pathology images so as to generate a principle component score diagram in accordance with the first simulating spectra; defining a plurality of triangle areas in the principle component score diagram in accordance with the first simulating spectra; determining whether a principle component score of a second simulating spectrum of a second pathology image is within any one of the triangle areas; and confirming the second pathology image belongs to one of the cancerous lesion images when the principle component score of the second simulating spectrum is within any one of the triangle areas.

The invention belongs to the field of artificial intelligence auxiliary diagnosis of medical pathology, and discloses an urine erythrocyte lesion identification and statistics method and a system based on improved Faster R-cnn, and the method comprises the following steps: S1, collecting a plurality of sample urine erythrocyte images of a nephrotic patient; S2, labeling the types of the urine redcells in the sample urine red cell image, generating a sample by using an adversarial generative network, and performing data enhancement preprocessing after the sample is generated to serve as training data; S3, inputting the training data into an improved Faster R-cnn neural network for training; wherein a 1 * 1 convolution layer is added into a basic network of the improved Faster R-cnn neuralnetwork, and a BN layer in the network is replaced by a GN layer; and S4, inputting the to-be-identified urine erythrocyte image into the network to obtain the classification of all urine erythrocytesin the image, and carrying out statistics. According to the method, a quicker and more accurate erythrocyte classification malformation rate statistical result can be obtained.

Described herein is the use of gamma cameras in the fringe field of the MRI system. Specifically, an MR image of the breast with lesion identification is first produced. Then, a gamma camera is attached to the existing breast immobilization system for generating one or more gamma images of the breast. The gamma camera is then removed from the breast immobilization system, and a breast biopsy is performed. The gamma camera can then be used to image the biopsy cores that have been removed from the patient in order to verify that the biopsy cores are radioactive, that the biopsy cores extend from one end of the lesion to the other and have a radioactive profile in which the tip is not as radioactive as the middle, and in which the ratio of amount of radioactivity in the middle of the core to the amount of radioactivity that is present in the tip of the core can be expressed as a ratio.

The invention relates to the technical field of image recognition, and discloses a multi-scene digestive tract endoscopic image recognition method and system based on artificial intelligence. The method comprises the steps of obtaining a data sample of a digestive tract endoscope image, marking a shooting part of the image for the data sample, checking a scene, and marking a segmentation region ofa lesion; training a corresponding shooting part recognition model / inspection scene recognition module based on the data sample set of the same shooting part / inspection scene; aiming at the combination of each shooting part and each examination scene, training to obtain a lesion identification model of the corresponding shooting part in the corresponding examination scene; for a to-be-identifiedendoscopic image, identifying a shooting part and an examination scene through a corresponding identification model, selecting a lesion identification model of the corresponding shooting part in the corresponding examination scene, and identifying a lesion part is identified. According to the scheme, the accuracy of identification and the generalization ability of the system can be improved.

The invention belongs to the technical field of image processing, and discloses an image processing system in orthopedic anesthesia based on optimized sedation management and regional blocking. The image processing system comprises an ultrasonic image obtaining module, an operation state detection module, a main control module, an image processing module, a three-dimensional navigation module, a storage module and a display module. Through the image processing module, a doctor is prevented from manually identifying lesion tissue, the image identifying efficiency is improved, the doctor does not need to have a high lesion identification ability, and the labor cost is reduced. Through the three-dimensional navigation module, a virtual bone image of a surgical spot of a user and a three-dimensional image of the specific location, where a surgical instrument enters, in the user body are overlaid on a real vision of the doctor, in the surgical process, real-time navigation is conducted, thesurgical operation efficiency of the doctor is greatly improved, and a surgery is more safely, accurately and efficiently implemented.

The invention is suitable for the technical field of computers, and provides an endoscope system, an endoscope image recognition method and equipment, and a storable medium, and the endoscope system comprises an image receiving unit which is used for receiving a to-be-recognized endoscope image, an image processing unit which is used for extracting biological characteristic information from the endoscopic image to be identified and generating a lesion part recognition result according to the biological feature information and a biological feature-lesion image recognition model generated by training a plurality of pre-collected endoscope image samples carrying lesion part information through a convolutional neural network, and a display unit which is used for identifying and displaying thelesion part in the to-be-identified endoscope image according to the lesion part identification result. Compared with an existing system which directly carries out lesion information analysis according to image apparent characteristics, the system is higher in recognition precision and efficiency, can be suitable for lesion recognition of endoscope images with different illumination and shooting angles, and greatly reduces the probability of missed diagnosis or misdiagnosis.

The invention relates to the field of medical science and technology, and provides a hypertension risk prediction method, device and equipment and a medium. The method comprises the following steps: obtaining the individual information and a fundus image of a to-be-detected target; respectively inputting the fundus image into a preset arteriovenous segmentation model, an optic disk segmentation model and a lesion recognition model, and respectively extracting arteriovenous blood vessels, optic disk regions and fundus lesion features in the fundus image; quantifying the diameters of arteriovenous blood vessels in the optic disc area to obtain blood vessel diameter quantized values corresponding to the arteriovenous blood vessels respectively; using a machine learning regression method to construct a risk prediction model for predicting the hypertension of the target to be detected based on the blood vessel diameter quantized value, the fundus lesion features and the individual information; and inputting the fundus image corresponding to the to-be-detected target into the risk prediction model for detection to obtain a hypertension prediction result of the to-be-detected target. The risk prediction model is trained by adopting a plurality of variable factors, so that the accuracy of hypertension risk prediction is greatly improved.

The invention provides a fundus lesion recognition method and device. The method comprises the following steps: acquiring an OCT fundus image of a specific patient; determining boundary information of a retina tissue of the OCT eye fundus image by adopting a first neural network model, and segmenting the OCT eye fundus image into a plurality of image blocks according to the boundary information of the retina tissue; the plurality of image blocks are provided for a second neural network model to obtain the fundus lesion type of the specific patient, and the second neural network model calculates the feature vector of each image block and the attention weight of each image block for the set fundus lesion type; and performing weighted calculation to obtain a fused feature vector so as to obtain an identification result aiming at a set fundus lesion type. Wherein the second neural network model reinforces important information in the image block group in combination with the spatial domain significance and the scanning position significance, and filters unimportant information, so that low-cost model training and application are realized.

The invention relates to the technical field of medical image processing, in particular to an artificial intelligence system for identifying high myopia retinopathy, comprising: an image acquisition module for acquiring an optical coherence tomography image and a fundus reference of a patient to be identified; the first recognition module is used for inputting the optical coherence tomography image into a high myopia retinopathy recognition model, judging whether the optical coherence tomography image has lesion or not and obtaining a lesion type result; the second recognition module is used for inputting the fundus color photograph into a high myopia retinal lesion staging model, judging the retinal lesion staging of the fundus color photograph and obtaining a lesion staging judgment result; and the report generation module is used for generating a diagnosis and treatment suggestion report according to the lesion type result and the lesion staging judgment result. According to the method, the optical coherence tomography image and the fundus color photograph of the patient can be quickly and effectively recognized to recognize common retinopathy of high myopia.

The invention provides a retinopathy recognition model generation method, a recognition device and equipment. The method comprises the steps: training a preset network model by employing fundus imagesin a training set, obtaining a trained retinopathy recognition model, and enabling the fundus images in the training set to carry classification mark information corresponding to the fundus images, wherein the classification mark information is retina feature classification information corresponding to the eye fundus picture, and the preset network model learns the classification mark informationcontained in the eye fundus picture in the training process. Thus, the trained retinal lesion recognition model can be used for recognizing the retina image contained in the detection picture, and anaccurate classification result of the retinal features is given. According to the method, the device and the equipment provided by the embodiment of the invention, the network model for retinopathy recognition is trained in a deep learning mode, so that the accuracy and the detection efficiency of fundus image recognition are improved.

The invention discloses a Tibetan highland barley leaf lesion identification method, and belongs to the field of image processing. The method comprises the following steps of converting a preprocessedimage into an HSV color space; performing threshold processing on the image in an HSV color gamut; wherein a color is light brown and dark brown, and the color in the middle is dark brown; remainingcolor setting to white, merging with an original image, setting H and S classification grades and change ranges; and counting H and S component data in the image according to a setting result, creating a histogram, normalizing the histogram height of the image, training a KNN classifier by taking the normalized histogram height and the corresponding abscissa as image features and labels, performing the same processing on an image sample for testing, and inputting the image sample into the KNN classifier to obtain a prediction result. According to the highland barley lesion identification method, images of normal highland barley leaves and lesion highland barley leaves can be well classified by filtering and denoising the images, extracting and normalizing color space features and classifying the images by a support vector machine, and the highland barley lesion identification efficiency and accuracy are improved.

Method and system for grading a medical image. For example, a system for grading a medical image comprising a grading network configured to provide a grading result corresponding to the medical image based on at least the medical image and / or a list of lesion candidates generated by a lesion identification network.

The invention discloses a crop lesion detection method and device. The detection method comprises a lesion classification step and a lesion recognition step. In the lesion classification step, training of a decision tree classifier is completed according to lesion type labels and image features of all sample images; and a plurality of feature template images are generated according to the decisiontree classifier; the lesion identification step comprises the step of calculating the matching degree between the to-be-detected image and each feature template image by using a semi-naive Bayes classifier; and judging possibility of lesion of to-be-detected crop according to the matching degree between the to-be-detected image and each feature template image. According to the lesion detection method, in the lesion classification step, a decision tree classifier is used for determining crop characteristics corresponding to various different lesion types, so that a template image is generated;in the lesion identification step, the matching degree of the to-be-detected image of the to-be-detected crop and each template image is calculated by using the semi-naive Bayes classifier, so that the lesion type of the to-be-detected crop and the possibility data of the lesion are obtained, and the reliability is high.

The invention discloses a blood vessel tree generation and lesion recognition method, device and equipment and a readable storage medium. The method comprises the following steps: acquiring a medicalimage, wherein the medical image comprises a plurality of voxels; segmenting a plurality of blood vessel regions from the medical image, wherein the blood vessel regions comprise blood vessels and voxels having similar characteristics with the blood vessels; establishing an incidence matrix of the plurality of blood vessel areas according to the proximity relation among the plurality of blood vessel areas; respectively acquiring a blood vessel section and a blood vessel bifurcation point of each blood vessel area; according to the incidence matrix, the blood vessel sections, the blood vessel bifurcation points and the connection relations between the blood vessel sections and the blood vessel bifurcation points contained in the blood vessel areas are constructed into blood vessel graphs ofthe blood vessel areas; and establishing a plurality of corresponding blood vessel trees based on the plurality of connected sub-graphs in the blood vessel graph. According to the method, the form, the structure and the characteristics of the blood vessel image and the connection relationship between the blood vessels can be analyzed, and an accurate blood vessel tree is constructed.

The CT image auxiliary diagnosis system based on cascade connection comprises a server side, the server side comprises a server, input equipment and output equipment, the input equipment and the output equipment are connected with the server, and an image preprocessing module and a cascade connection detection model are stored in the server; the cascade detection model comprises a convolutional neural network, a region suggestion network, an ROI Pooling layer and a plurality of cascade detectors. According to the invention, medical imaging and computer image processing are combined to completeautomatic detection of lesions. Aiming at the problem that CT focus features are complex and difficult to extract, a deformable convolution kernel is introduced into the cascade detection model to replace a standard convolution kernel, that is, 2D offset is increased in the standard convolution kernel, so that the focus features can be better extracted. Experimental results show that the lesion detection quality can be effectively improved, doctors can be helped to finish case classification and lesion identification more quickly and accurately, and the workload of the doctors is reduced.

The invention discloses an intelligent film reading system for tumor image data. One-key film reading report writing is realized, steps such as image data preprocessing, lesion identification, quantitative parameter measurement, diagnostic scoring and visualization are completed automatically by a computer, the operation procedure and the diagnostic process are simplified, the advantages of a doctor and the computer are executed to the maximum extent, the current image film reading working efficiency is improved significantly, the film reading result is objective and accurate, and the repeatability is high.

The present disclosure describes a method, an apparatus, and storage medium for recognizing medical image. The method includes obtaining, by a device, a medical image. The device includes a memory storing instructions and a processor in communication with the memory. The method further includes determining, by the device, the medical image through a first recognition model to generate a lesion recognition result used for indicating whether the medical image comprises a lesion; and in response to the lesion recognition result indicating that the medical image comprises a lesion, recognizing, by the device, the medical image through a second recognition model to generate a lesion degree recognition result of the medical image used for indicating a degree of the lesion. Manual analysis and customization of a feature extraction solution are not required, so that the efficiency and accuracy of medical image recognition are improved.