99 results about "Chest ct" patented technology

The invention discloses a pulmonary nodule automatic detection method and system based on a chest CT image; the method comprises the following steps: receiving a chest CT image, using a preset 3D convolution nerve network to carry out pulmonary nodule focus detection for the chest CT image, and outputting a detection result; receiving the CT image and the detection result, and loading the detection result with the corresponding chest CT image to form a fused file; obtaining the fused file and carrying out visualization treatment, providing browse auxiliary operations for users on the fused file with the visualization treatment, and providing a calculation and measuring function. The method and system can isolate lung areas, extract candidate pulmonary nodules and remove false positives, thus improving focus detection efficiency, and the false positive efficiency can reach 97.7%. the method can use visualization three dimensional forms to directly assist a doctor to determine the pulmonary nodule forms.

The invention belongs to the application field of computer analysis technology of medical image, and relates to a pulmonary nodule three-dimensional segmentation and feature extraction method and a system thereof based on chest CT. By adopting a level set method based on a three-dimensional space, pulmonary nodule is divided within a three-dimensional interest area, and the feature of the nodule can be extracted by combining the characteristics of a bounding curved surface and the interest area of the pulmonary nodule. The invention comprises a lung area segmentation module, a three-dimensional interest area extracting module, a pulmonary nodule three-dimensional segmentation module, a nodule feature extraction and quantification module and an output module. The invention fully utilizes the three-dimensional space information of HRCT image, thus reducing the interference of normal tissues around the nodule for segmentation and feature extraction, and improving the accuracy of feature extraction.

The invention discloses an end-to-end chest CT image segmentation method based on a fully convolutional neural network. The method comprises the steps of firstly performing clinic scanning for obtaining k sets of chest CT images, and dividing each set of CT images according to each slice as a training sample; then performing manual cutting on each training sample and a testing sample by professional medical care personnel, dividing the image to four parts, namely a lung part, a tracheae art, a skin part and a background part; then constructing an end-to-end fully convolutional neural network for training marked chest CT training data, and obtaining a trained parameter model; separating the slices of the scanned CT image, and inputting the slices into the trained model one by one, and obtaining segmented output models; and finally combining the output models, thereby obtaining a segmented chest CT image model. The convolutional neural network model which utilizes an image neighborhood content for performing characteristic extraction can perform dense predication on the chest CT image and furthermore simplifies an image segmentation process.

The invention discloses a CT lung nodule detection system based on 3D-Unet. The system includes a CT image input and preprocessing module, a candidate nodule detection module and a false-reporting elimination module which are sequentially connected. The CT image input and preprocessing module is used for reading a chest CT graph, obtaining spacing and origin information of the CT graph, and carrying out lung volume segmentation on the CT graph. The candidate nodule detection module is used for inputting the image, of which preprocesing is completed, into a Unet network, and obtaining a location of a candidate lung nodule. The Unet network includes sixteen residual blocks, two path units, four maximum-pooling units, one preparation unit, one probabilistic neuron failure unit and one outputunit. According to the system, automatic detection of the lung nodule is realized, problems of high doctor workloads and high misdiagnosis rates in lung nodule diagnosis are solved, an effect of detecting the lung nodule by the 3D-Unet is realized, more contextual semantic information in the CT image is utilized, and detection accuracy is higher.

A method of automatically detecting pulmonary nodules is provided, including the operations of acquiring a chest computed tomography (CT) image, extracting a lung region from the chest CT image, extracting a group of nodule candidates from the lung region using a gray-level thresholding technique and a three-dimensional (3-D) region growing technique, and performing 3-D feature recursive analysis on all of the nodule candidates.

The invention belongs to the technical field of medical image-based image processing, and particularly relates to a method for automatically extracting a tracheal tree from a chest CT image. The method comprises the steps of obtaining main tracheae and main bronchi; according to a 3D region growing segmentation mode and information of the obtained main tracheae and main bronchi, building an adaptive threshold 3D region growing segmentation model and an adaptive threshold leakage model; by utilizing the adaptive threshold 3D region growing segmentation model and the adaptive threshold leakage model, extracting second tracheal branches of the chest CT image; according to intermediate information of the extracted second tracheal branches, adjusting parameters of the adaptive threshold 3D region growing segmentation model and the adaptive threshold leakage model, and then extracting third tracheal branches of the chest CT image; and based on an obtained tracheal tree topology structure, extracting terminal tracheal branches, and obtaining the tracheal tree of the chest CT image. According to the method provided by the invention, the tracheal segmentation precision of extracting the tracheal tree from the CT image is improved and the extraction time is shortened.

The invention relates to an automatic heart locating method based on a chest locating film, belonging to the field of CT (computed tomography) image intelligent aid application. The automatic heart locating method can be used for automatically locating a heart area of a chest CT locating film, so that the manual operation can be omitted, and the working efficiency can be improved; three-dimensional location is automatically carried out for the heart, so that the location of the heart in a local low-dose scanning image can be realized, the radiating area of the local X-ray can be determined, the preparation can be made for the local fine scanning, and the purpose of reducing the CT radiation dose can be realized.

A method for tracing, rib posteriors includes providing an incomplete rib tracing comprising a digitized 3-dimensional representation of a rib and a digital image from which said rib tracing was extracted, initializing, a tracing direction for a remainder of said rib, detecting a plurality of ridge points in a sub-volume of said digital image about said initial tracing direction, and deforming a closed curve in a plane perpendicular to said tracing direction until a rib-edge contour is obtained, using said ridge points as constraints.

The invention relates to a classification method for chest CT (Computed Tomography) images based on deep learning. In the method, characteristics of the chest CT images are extracted and analyzed by use of a relatively mature deep learning method at present, and a model structure is trained by use of an existing image data set to implement quantitative classification of multiple indexes affectingdisease judgment. Therefore, a quantified physiological index classification result is provided for a diagnosis doctor, and the work load and the analytical judgment difficulty of an occupational physician are reduced.

The invention provides a chronic obstructive pulmonary disease detection system based on a deep neural network. The system comprises a preprocessing module, which is used for carrying out gray-scale treatment on an obtained chest CT image of a first patient and extracting a plurality of pulmonary lobule area images of the chest CT image obtained after gray-scale treatment; and a detection module,which is used for inputting obtained body mass index BMI of the first patient and the plurality of pulmonary lobule area images into a trained deep neural network model, and obtaining probability of the first patient in getting the chronic obstructive pulmonary disease. The chronic obstructive pulmonary disease detection system based on the deep neural network, through combination of the deep neural network and medical images, and with clinical experience knowledge obtained by diagnosing the COPD manually being as prior knowledge, carries out detection on early-stage pulmonary lobule small lesions, and carries out highly-reliable prediction on the cases, thereby improving COPD detection accuracy.

The invention discloses a manufacturing method for an individual breast prosthesis based on a breast digital image fusion technique. The manufacturing method comprises the following steps: acquiring image data; performing breast three-dimensional laser scanning and chest CT scanning; respectively performing three-dimensional modeling for three-dimensional laser scanning data and chest CT data; fusing two digital images; acquiring a fused breast three-dimensional digital model; performing 3D printing. According to the manufacturing method for the individual breast prosthesis based on the breast digital image fusion technique, a breast three-dimensional digital model with an internal structure on a standing position can be acquired through the compound image fusion of breast three-dimensional scanning and CT; a 3D printing technique is adopted for printing the individual breast prosthesis and has a significance in achieving better operation result; the problem that the structure with clear breast interior on the standing position cannot be acquired according to the traditional imaging method and only can be made up through software can be solved.

The invention relates to a medical information cross-modal hash coding learning method based on a generative adversarial network, and belongs to the technical field of medical information processing and information retrieval. According to the method, a generative adversarial network is adopted to learn hash codes of chest CT images and texts, and the learned hash codes are constrained through a semantic similarity matrix. Finally, accurate hash codes are learned, and semantic association between the two modes is successfully achieved. According to the method, on the basis of single-layer fine-grained pulmonary nodule features, more complete feature information of the three-dimensional pulmonary nodule is extracted, and a hash code generation model is obtained by adopting a supervised training mode in the method, so that relatively high accuracy is realized in cross-modal retrieval.

The invention discloses a pulmonary nodule recognition and segmentation method based on deep learning, and the method comprises the steps: preprocessing a DICOM file so as to generate a chest CT image, segmenting a lung mask from the chest CT image, and repairing the lung mask; generating a three-channel lung image according to the chest CT image and the repaired lung mask, and inputting the three-channel lung image into a two-dimensional YOLO v3 neural network to detect a suspicious area of pulmonary nodules; standardizing the chest CT image according to the suspicious area to generate a standardized matrix, inputting the standardized matrix into a 3D Dense Net neural network and a C3D neural network for prediction, and generating a target prediction box according to a prediction result;and normalizing the chest CT image according to the target prediction box to generate a normalization matrix, inputting the normalization matrix into the 3D UNet neural network for segmentation, and optimizing a segmentation result. The invention further discloses a pulmonary nodule recognition and segmentation system based on deep learning. According to the invention, the accuracy and speed of identification and segmentation can be effectively improved.

The invention discloses a lossless compression method used for a chest computed tomography radiological image. The method comprises the steps of 1, automatically extracting chest of a CT image; 2, performing pixel conversion and normalization; and 3, performing pixel and mean coding. According to the method, a relatively high compression ratio can be provided in chest CT data compression; and compressed data meets the usage requirements of radiological diagnosis.

The invention relates to the technical field of medical treatment, in particular to a deep learning-based rib fracture auxiliary detection method and an image recognition method, and the method comprises the steps: selecting a certain number of chest CT images as a training set, and marking rib fracture regions and rib numbers in images; performing data normalization on the images; performing model training by taking the processed image as input and the marked rib fracture area and rib number in the image as output, wherein the training model comprises a rib detection model, a rib fracture segmentation model and a rib number and segmentation model; and processing a chest CT image to be detected, inputting the processed chest CT image to the trained rib fracture detection model, and outputting a detection result. According to the rib fracture auxiliary detection method based on the deep learning algorithm provided by the embodiment of the invention, false positive and false negative of rib fracture detection are effectively reduced, and the detection result provides position information of suspected rib fracture, so that a doctor can be assisted in diagnosis.

The invention discloses a static artery separation method and device based on a CT image, and the method and device improve the efficiency and precision of static artery separation compared with a conventional method and manual labeling of a doctor, achieve the full-automatic static artery separation, and do not need the manual intervention. The method mainly comprises the following steps: carrying out lung region segmentation on a chest CT image by using a preset three-dimensional lung segmentation model to obtain a three-dimensional mask of a lung region; carrying out convex hull operation on a lung mask, taking a lung region according to the lung mask subjected to convex hull operation, setting the pixel value of the extrapulmonary region to be 0, and obtaining a maximum extrapulmonarybounding box according to the lung mask; and performing static artery separation on the segmented CT image of the lung in the lung external bounding box by using a preset downsampling-free three-dimensional cavity convolutional neural network to obtain a static artery mask. As the accurate annotation data is used for training, and the three-dimensional cavity convolutional neural network is used for learning, the loss of information amount is reduced, and the accurate separation of the static artery is realized.

The invention belongs to the technical field of image processing, and discloses a liver lesion image processing method and system, a storage medium, a program and a terminal. An information processingserver acquires basic information and a CT image of an examiner who has undergone chest CT from a security port; through a gRPC, the information processing server transmits the image to the algorithmserver for calculation, and an algorithm server transmits a result back to the information processing server; the information processing server writes the inspector information and the calculation result into a MySQL and MongoDB database, and a RESTful standard interface is established to expose the data; and a front-end webpage displays liver and possible focus information. On the basis of the existing classic neural network unet architecture, on the basis of continuously feeding back an increased liver lesion data set, through comprehensive improvement of the model, the liver recognition accuracy can be improved to 97%, and the liver occupation recognition accuracy is improved to 98%.

The invention provides a pulmonary emphysema image processing method and system based on low data demands. The method comprises the following steps: M1, preparing a pulmonary CT film marked with the yin and yang of a pulmonary emphysema focus, and forming a group of medical digital imaging and communication files; m2, preprocessing the prepared lung CT film, and obtaining a three-dimensional arraythrough a group of medical digital imaging and communication files; m3, constructing a deep convolutional neural network architecture, training a deep convolutional neural network through the three-dimensional data, and judging a pulmonary emphysema image through the deep convolutional neural network; required characteristics can be automatically learned from chest CT with emphysema yin-yang marks, and image processing yin-yang judgment is carried out. Compared with a common CT deep neural network image processing auxiliary diagnosis technology, the technology avoids the problems that a 3D model occupies a large amount of memory and is poor in performance on a CT with a thick layer, also avoids the limitation that a 2D model cannot comprehensively utilize three-dimensional space information, and fully utilizes the spatial relationship between layers.

The invention relates to the technical field of medical image segmentation, and provides a new coronal pneumonia focus segmentation method based on feature fusion deep supervision U-Net, and the method comprises: 1, obtaining an initial sample set; 2, preprocessing the initial sample set; 3, obtaining a training sample set and a verification sample set; 4, performing data augmentation on the training sample set; 5, building a COVID-19 lesion area automatic segmentation model in the chest CT image based on feature fusion deep supervision U-Net, wherein the automatic segmentation model comprises an encoder, a decoder, jump connections, feature fusion blocks and deep supervision branches, adding the feature fusion blocks between the jump connections of different levels, and adding the deep supervision branches to the decoder part; 6, training a segmentation model; and 7, segmenting the chest three-dimensional CT scanning image to be segmented. According to the method, automatic segmentation of the COVID-19 lesion area in the chest CT can be realized, and the accuracy and rapidity of segmentation are improved.

The invention provides a pulmonary fibrosis detection and severity evaluation method based on deep learning, and the method comprises the steps: S1, preprocessing chest CT sequence images of a plurality of pulmonary fibrosis patients, and obtaining a plurality of first CT images; S2, extracting and labeling a plurality of first CT images to generate a training set and a verification set; S3, pre-training a first deep convolutional neural network model and a second deep convolutional neural network model through the training set and the verification set; S4, inputting the CT image sequence of the patient to be detected into the trained first and second deep convolutional neural network models, and identifying a lung region and a pulmonary fibrosis focus region contained in each CT image inthe CT image sequence of the patient; calculating to obtain the proportion of the pulmonary fibrosis focus of the patient in the lung; S5, marking pulmonary fibrosis staging based on the proportion; and S6, grading the pulmonary fibrosis severity of the patient based on the detection result of the physiological parameters. The invention further comprises a pulmonary fibrosis detection and severityevaluation system based on deep learning.

The invention discloses a lung tissue segmentation method based on a capsule network, and the method comprises the steps: inputting a to-be-segmented chest CT image into a convolution layer of a segmentation network, and obtaining a first feature map; inputting the first feature map into a capsule layer of the segmentation network to obtain a second feature map; and finally, performing binarization processing on the second feature map according to the modulus value of the multi-dimensional vector corresponding to each pixel point of the second feature map, so that a lung tissue segmentation image can be obtained. Therefore, the bottom-layer features of the chest CT image are extracted through the convolution layer of the segmentation network, the bottom-layer features are subjected to vectorization operation through the capsule layer of the segmentation network, features with higher dimension and a hierarchical structure can be learned in the training and learning process, and therefore when lung tissue in the chest CT image is segmented, the difference between targets can be described more accurately, and accurate lung tissue segmentation of the chest CT image is achieved.

The invention provides a method for performing aesthetic three-dimensional reconstruction in combination with a chest CT film and pre-modeling data. According to the method, an accurate and attractivethree-dimensional chest organ reconstruction output can be automatically generated from the chest picture input by the user in combination with a pre-prepared three-dimensional artistic model. Compared with a common CT deep learning segmentation reconstruction technology, due to the fact that the information of the three-dimensional artistic model is fully utilized, the problem that the final reconstruction effect is poor due to inevitable segmentation errors of the deep learning network can be relieved to a great extent, and the situation that the three-dimensional reconstruction effect is not ideal due to the fact that the layer thickness resolution ratio is poor on thick-layer CT input can also be relieved.

The invention discloses a semi-supervised chest CT image lesion segmentation model training method and system, and the method comprises the steps of obtaining a sample CT image, and calculating the credibility of the sample CT image, the sample CT image comprising a labeled image and an unlabeled image; according to the credibility of the sample CT image, pairing of the CT image is carried out, an augmented CT image is generated, and the CT image is an image only containing a focus; training a deep learning network according to the augmented CT image to obtain a focus segmentation model; and optimizing the segmentation precision of the focus segmentation model based on a self-learning strategy of a teacher-student model. According to the semi-supervised chest CT image lesion segmentation model training method, lesion information is reserved, the advantages of a full-supervised loss function and a common semi-supervised loss function are integrated, a large amount of labor cost is saved, and the requirement for massive annotation data is avoided.

The invention relates to the technical field of medical treatment, in particular to a CT rib fracture auxiliary diagnosis system based on a deep learning algorithm, and the system comprises a data processing module which is used for carrying out data normalization of a to-be-detected chest CT image, and obtaining a normalized chest CT image; and a detection module which is used for inputting the normalized chest CT image into the detection module to obtain a detection result of the chest CT image; wherein the detection module comprises a rib detection unit, a rib fracture segmentation unit, arib numbering and segmenting unit and a display module, and the display module is used for displaying an output result of the chest CT image. According to the CT rib fracture auxiliary diagnosis system based on the deep learning algorithm, false positive and false negative of rib fracture detection are effectively reduced, the detection result provides position information of suspected rib fracture, and a doctor can be assisted in diagnosis.

The invention belongs to the application field of computer analysis technology of medical images, and is a method and system for three-dimensional segmentation and feature extraction of pulmonary nodules based on chest CT. The present invention uses a three-dimensional space-based level set method to segment pulmonary nodules in a three-dimensional region of interest, and extracts nodule features in combination with the pulmonary nodule boundary surface and the characteristics of the region of interest. The invention comprises a lung region segmentation module, a three-dimensional interested region extraction module, a pulmonary nodule three-dimensional segmentation module, a nodule feature extraction and quantification module and an output module. The invention makes full use of the three-dimensional space information of the HRCT image, can reduce the interference of normal tissues around the nodules on its segmentation and feature extraction, and improves the accuracy of feature extraction.

The invention discloses a cruising locator for pulmonary minimal focuses under a thoracoscope. A navigation probe is arranged at the tip of a pair of lung grasping forceps, and a micro-electromagnetic sensor is arranged inside the navigation probe. The cruising locator is matched with an in-vitro electromagnetic locating panel, an electromagnetic patch and an electromagnetic navigation host, and is used for performing chest CT scanning and three-dimensional reconstruction on the scanned image to obtain a chest three-dimensional analog image, a doctor marks location and size of a doubtable focus position in the analog image and inputs the marked image to the electromagnetic navigation host, and the system is used for automatically simulating an optimized location navigation path, so that a pulmonary minimal focus can be easily located by utilizing the navigation probe under guidance of a real-time navigation system in the thoracoscopy. The cruising locator is simple in structure, can be used for accurately and quickly reaching a target focus and avoiding the operation risk generated when location is carried out in other operations, has relatively low requirement for doctors and patients, is safe and reliable to operate, and has a good practical value.

The invention discloses a method for producing a funnel chest correction plate. The method comprises the following steps: in accordance with a chest CT image of a patient with funnel chest, reconstructing a chest skeleton three-dimensional model by virtue of three-dimensional software; in accordance with a normal human thoracic model having close thoracic dimensions as well as curvature change of the thoracic model, designing a three-dimensional model of the correction plate; by virtue of finite element software, analyzing a correction shape of thoracic cavity after the correction plate is implanted in the thoracic skeleton of the patient with the funnel chest; and in comparison with the thorax of normal people, judging whether the correction plate can meet requirements or not, if so, determining design parameters of the correction plate, otherwise, re-designing the three-dimensional model of the correction plate. According to the method disclosed by the invention, the correction plate with individual shape and dimensions can be produced in accordance with the chest skeleton shape of the patient with the funnel chest and the optimum parameters of the correction plate can be obtained by virtue of a finite element method, so that harm to the patient caused by repeatedly regulating the shape of a steel sheet can be effectively prevented; and by manufacturing the correction plate before an operation, the method is conducive to shortening an operating time, reducing an operative wound and intra-operative and post-operative bleeding amounts and guaranteeing a post-operative correction effect.

The invention discloses an EGFR gene mutation detection method and system based on a chest CT image and a computer readable storage medium. The method comprises the following steps: receiving a chest CT image to be processed; extracting feature variables of a plurality of target imageomics features from the chest CT image, wherein each target imageomics feature does not have a saliency difference between chest plain-scan CT and chest enhanced CT on a first saliency level and has a saliency difference between EGFR mutation positive and EGFR mutation negative on a second saliency level; according to a regression model and the feature variables, obtaining a score value corresponding to the chest CT image to be processed; according to the score value, determining an EGFR gene mutation detection result of a lung cancer patient to which the chest CT image to be processed belongs, wherein the detection result reflects the probability of EGFR gene mutation. According to the method, EGFR gene mutation can be detected based on a chest plain-scan CT image or an enhanced CT image, and the clinical application range is wide.

The invention provides a non-small cell lung cancer risk prediction method based on machine learning, and the method comprises the steps of carrying out the typical feature extraction of a chest CT image of a patient, and recording the typical feature as a CT image feature; obtaining CT image features of all patients in the sampling sample, forming a sample data set by combining clinical pathologyfeatures of all patients in the sampling sample, and preprocessing the CT image features in the sample data set; dividing the preprocessed sample data set into a test set and a training set; and performing feature screening on CT image features and clinical pathology features of patients in the divided test set and training set, and based on the training set subjected to feature screening, performing prediction training on high and low risks of non-small cell lung cancer by adopting a joint Nomogram model to obtain a non-small cell lung cancer risk prediction result. The invention can accumulate the experience of excellent doctors and experts, so as to copy the experience to other small cities and hospitals for popularization and application, improve the risk prediction accuracy, and improve the treatment effect of patients.