13 results about "Lung segmentation" patented technology

The invention discloses a chronic obstructive pulmonary disease detection method and system based on multi-instance learning and multi-task learning, and the method comprises the steps: collecting and processing the multi-omics data of a subject, and manually marking the CT slice images of a part of the subject to construct a first training set; constructing a lung segmentation model based on a U-Net network, and training by using the first training set; using the trained lung segmentation model to obtain CT slice images containing lungs corresponding to all subjects, and constructing a second training set; improving the DSMIL network based on a multi-instance learning method and a multi-task learning method to construct a chronic obstructive pulmonary disease detection model, and training by using the second training set; utilizing the trained lung segmentation model to obtain a CT slice image containing the lung of the subject, and then utilizing the trained chronic obstructive pulmonary disease detection model to obtain a corresponding chronic obstructive pulmonary disease detection result. According to the method, the feature extraction capability of the model can be improved, the generalization capability of the model is enhanced, and high accuracy of chronic obstructive pulmonary disease detection is ensured.

The present invention discloses a method, device and computing equipment for pushing image reports based on RPA and AI. The method includes: using a preset lung segmentation model to segment the lung medical image to be detected sent by the RPA robot to obtain a target medical image containing only the lung area, using a preset lung nodule detection model to detect the target medical image to obtain attribute information of each suspected lung nodule, using a preset lung nodule segmentation model to segment the area where each suspected lung nodule is located to obtain three-dimensional contour information of each suspected lung nodule; generating a suspected lung nodule image report based on the attribute information and three-dimensional contour information of each suspected lung nodule and sending it to the hospital platform through the RPA robot. In this way, the lung nodules are detected by AI image analysis technology to obtain a suspected lung nodule image report, and the report is sent to the hospital platform through the RPA robot, which reduces the time doctors spend identifying lung nodules and improves efficiency.

The present invention relates generally to assessing abnormal patterns associated with a disease from x-ray images, and in particular to assessing abnormal patterns associated with COVID-19 (coronavirus disease 2019) from x-ray images using a machine learning network trained on DRRs (digital reconstructed radiographs) and ground truth derived from CT (computed tomography). Systems and methods of assessing a disease are provided. An input medical image of a first modality is received. A lung is segmented from the input medical image using a trained lung segmentation network, and an abnormal pattern associated with the disease is segmented from the input medical image using a trained abnormal pattern segmentation network. The trained lung segmentation network and the trained abnormal pattern segmentation network are trained based on 1) synthetic images of the first modality generated from training images of a second modality and 2) target segmentation masks for the synthetic images generated from training segmentation masks for the training images. An assessment of the disease is determined based on the segmented lung and the segmented abnormal pattern.

The application discloses a lung ventilation-perfusion development region image fusion method, relates to the technical field of image fusion, and comprises the following steps: arranging lung medical image data according to slice sequences to obtain a first image data set; performing data processing on the first image data set to generate a second image data set, inputting the second image data set into a U-Net model to perform lung segmentation processing, and obtaining a lung region mask; performing fusion processing on ventilation development images and perfusion development images after data processing to obtain a first fusion image, and generating a plurality of lung sub-regions; extracting ventilation development intensity and perfusion development intensity of each lung sub-region respectively, calculating ventilation-perfusion ratio values of the lung sub-regions, performing color coding mapping on the lung sub-regions according to the ventilation-perfusion ratio values, and generating a second fusion image. The application solves the image fusion problem of accurate color coding and visualization of ventilation-perfusion ratio values of lung adaptive sub-regions.

The invention provides a lung CT intelligent optimization and focus accurate display method based on free breathing. Relates to the technical field of medical image processing. The method comprises the following steps: acquiring a CT image of a target patient, and performing standardized preprocessing on a low-dose CT image and a standard-dose chest CT image to obtain a first preprocessed image and a second preprocessed image; performing full-automatic lung segmentation on the first preprocessed image by adopting a U-Net convolutional neural network model, and outputting a left lung region and a right lung region; and taking the second pre-processed image as a gold standard, training a deep convolutional neural network CycleGAN to obtain an image enhancement model, and performing image quality enhancement on the left and right lung regions in the first pre-processed image by using the image enhancement model to obtain an enhanced image. The chest image quality can be improved while the radiation dose of the patient is reduced.

The invention provides an nnunet-based lung eighteen-segment segmentation and post-processing correction method and system, which are applied to the technical field of medical image processing, and are characterized in that left and right lung segmentation is carried out on CT image data by adopting a lightweight left and right lung segmentation model, the left and right lungs are independently processed through region cutting, and finally, a hierarchical post-processing correction mechanism is introduced, so that the correction accuracy of the lung eighteen-segment segmentation and post-processing correction is improved. And the anatomical rationality and accuracy of the segmentation result are ensured. In conclusion, by optimizing the model structure and introducing the refined post-processing correction strategy, the lung eighteen-segment segmentation efficiency is improved, the resource consumption is reduced, the accuracy and reliability of the segmentation result are remarkably improved, and a better technical solution is provided for clinical medical image analysis.

The present invention discloses a lung field image segmentation method, device, and storage medium. The method comprises: determining the gradient amplitude of a preprocessed chest image and integrating the gradient amplitude; determining a chest image based on the integration result, and segmenting the chest image into a left chest image and a right chest image based on a left lung segmentation point and a right lung segmentation point; performing costal margin segmentation, lung apex segmentation, and transverse and mediastinum segmentation on the left chest image and the right chest image, respectively, to obtain left and right costal margin segmentation images, left and right lung apex segmentation images, and left and right transverse and mediastinum segmentation images; determining lung apex region boundaries, transverse mediastinum region boundaries, and costal margin region boundaries based on the left and right costal margin segmentation images, the left and right lung apex segmentation images, and the left and right transverse and mediastinum segmentation images, and connecting the lung apex region boundaries, transverse mediastinum region boundaries, and costal margin region boundaries to obtain a lung field image. The present invention segments and connects the costal margin, lung apex, and transverse mediastinum regions of the chest, thereby improving segmentation accuracy.

The application discloses a honeycomb lung segmentation method based on class imbalance and multi-stage down-sampling feature fusion, and relates to the field of medical image processing. The method comprises the following steps: S1, acquiring multiple honeycomb lung CT images, and dividing a training set and a test set; S2, constructing a honeycomb lung segmentation network model; S3, training the honeycomb lung segmentation network model in S2 by using the training set to obtain a trained honeycomb lung segmentation network model; and S4, inputting the test set into the trained honeycomb lung segmentation network model in S3 to obtain segmented honeycomb lung CT images. The method is helpful to improve the segmentation precision of the model on the basis of increasing a small amount of model parameters.

The invention provides an AI-based medical respiratory system diagnosis auxiliary method and system, and the method comprises the steps: obtaining a chest CT image from a PACS system, carrying out the preprocessing of lung segmentation, window level adjustment and the like, and generating a concept score vector through a concept activation encoder; the method specifically comprises the steps that a backbone network extracts a preprocessed image feature map, similarity is calculated according to the feature map and a predefined concept prototype library to generate an activation map, a high activation area of each concept prototype is determined, and average similarity is calculated to obtain a concept score; a diagnosis probability is generated by a microcosmic concept inference device, that is, output is generated by applying logic atoms (weight is set based on medical priori knowledge), and the diagnosis probability is generated through a classification layer in combination with a concept score vector; and finally, generating a structured report according to the diagnosis probability and the concept score vector. According to the method, diagnosis is decoupled into concept discovery, quantification and reasoning, interpretable decision making is achieved, the black box problem is solved, and a doctor can conveniently trace the AI judgment basis.

The invention discloses a lung ventilation-perfusion development area image fusion method, and relates to the technical field of image fusion, and the method comprises the steps: arranging lung medical image data according to a slice sequence, and obtaining a first image data set; performing data processing on the first image data set to generate a second image data set, and inputting the second image data set into a U-Net model for lung segmentation processing to obtain a lung region mask; performing fusion processing on the ventilation development image and the perfusion development image after data processing to obtain a first fusion image, and generating a plurality of lung sub-regions; respectively extracting the ventilation development intensity and the perfusion development intensity of each lung sub-region, and calculating the ventilation perfusion ratio of each lung sub-region; and performing color coding mapping on each lung sub-region according to the ventilation perfusion ratio to generate a second fusion image. According to the method, the problem of accurate color coding and visual image fusion of the ventilation perfusion ratio of the lung self-adaptive sub-region is solved.

Devices, methods, and systems are provided for analyzing lung imaging data. A server computing device receives imaging data of a lung over a network from a client computing device and analyzes the imaging data to identify lung, airways, and blood vessels, segment the lung into lobes, subtract airways, calculate volumes, calculate emphysema scores, identify fissure locations, calculate fissure completeness scores. A reconstruction of the fissures indicating locations where the fissures are incomplete and a report comprising fissure scores, volumes, and emphysema scores are created.

The present disclosure relates to a lung segmentation, lung disease evaluation method and device, electronic equipment and storage medium, and relates to the technical field of DR lung image segmentation. The segmentation method comprises: acquiring a segmentation model of a preset convolutional neural network, a DR lung region label image used for training the segmentation model, and a plurality of DR lung images to be segmented at multiple moments in a breathing process or a breath-holding state; wherein the method for determining the DR lung region label image used for training the segmentation model comprises: respectively performing rib edge boundary, lung apex boundary, and mediastinum and diaphragm edge detection on left chest images and right chest images of the plurality of DR lung region images to obtain the DR lung region label image; training the segmentation model by using the DR lung region label image; and completing left lung and / or right lung segmentation of the plurality of DR lung images to be segmented based on the trained segmentation model. Lung region segmentation and lung disease evaluation of DR lung images can be realized.

The application provides a kind of based on double knowledge graph's pneumoconiosis X-ray chest radiograph classification method, device, equipment and storage medium, it is related to medical image processing technical field.The method comprises: after lung segmentation is carried out to chest radiograph, utilize the frozen CLIP visual encoder and its embedded multilevel residual adapter output multistage enhanced visual features to strengthen subtle lesion expression;Based on medical classification standard, construct text and visual double knowledge subgraph, carry out graph convolution interaction and fusion to original text features and subgraph, obtain optimized text features;Visual and text features are mapped to unified semantic space, the similarity of hierarchical alignment is calculated, and the classification result is obtained by attention mechanism fusion.The present application introduces medical priori through knowledge graph, combined with multilevel visual enhancement, can significantly improve the recognition ability of pneumoconiosis lesion and cross-modal semantic alignment accuracy, realize high-accuracy automatic classification.