60 results about "Thyroid ultrasound" patented technology

The invention provides a thyroid ultrasound image nodule automatic diagnosis system based on a multi-scale convolutional neural network. The system includes a thyroid nodule coarse-to-fine classification module, a thyroid nodule region automatic detection module, and a thyroid nodule fine classification module. The size features of different sensing regions are extracted through a multi-scale feature fusion convolutional neural network, and then, the context semantic features of a thyroid nodule can be extracted according to local and global information, and the thyroid nodule can be automatically located. Through multi-scale coarse-to-fine feature extraction based on a neural network and the design of a multi-scale fine classification AlexNet of a pyramid structure, the position of a focus and the probability that the focus is benign or malignant can be accurately predicted, doctors can be assisted in diagnosing a thyroid focus, and the objectivity of diagnosis is improved. The systemhas the characteristics of good real-time performance and high accuracy.

The invention discloses a thyroid ultrasound image nodule automatic positioning and identifying method based on USFaster R-CNN, and belongs to the field of artificial intelligence and deep learning. The method comprises the steps of preprocessing a thyroid ultrasound image, building a deep neural network model, and training and optimizing the network model, wherein the deep neural network model comprises a bottom convolution feature extraction network, a candidate box generation network, a feature map pooling layer and a classification and candidate box regression network. A deep learning method is used to realize thyroid ultrasound image feature automatic extraction, candidate box automatic generation, screening and position correction. An automatic positioning and identification functionof thyroid nodules is realized. The method can effectively assist doctors in thyroid ultrasound image diagnosis, improve the objectivity and accuracy of diagnosis, and effectively reduce the workloadof doctors and the omission ratio of small target nodules.

The invention discloses a thyroid nodule diagnosis method based on a deep learning network, and belongs to the field of image processing and artificial intelligence aided disease diagnosis. The methodcomprises the following steps: searching an ultrasonic original image and a pathological report of a thyroid nodule of a thyroid patient, and constructing a thyroid nodule database; preprocessing theultrasonic image; performing semantic segmentation on the ultrasonic image preprocessed in the step 2 through a Deeplab v3+ method based on Xception-JFT, and forming a semantic segmentation result graph; judging benign and malignant thyroid nodules based on a deep learning network; and forming a thyroid nodule diagnosis information report. According to the method, the Deeplab v3+ algorithm basedon Xeption-JFT is adopted to establish the thyroid ultrasound image segmentation network model, the optimal segmentation effect is achieved by continuously improving the backbone network Xception, nodule information can be automatically and rapidly recognized under high accuracy and high robustness, image features are automatically extracted for accurate segmentation to obtain a better diagnosis result, and an objective reference is provided for clinical diagnosis.

The invention discloses a thyroid nodule automatic detection model construction method, system and device based on a convolutional neural network, and the thyroid nodule automatic detection model construction method comprises the steps: carrying out the noise reduction of thyroid ultrasonic image data, and obtaining a thyroid ultrasonic image training data set; based on the training data set, using a Yolov3 network to train a thyroid nodule detection model; based on the training data set, using a Resnet network to train a thyroid nodule benign and malignant identification model; and fusing thethyroid nodule detection model and the thyroid nodule benign and bad recognition model to generate a thyroid nodule automatic detection model.

The invention discloses a thyroid nodule focus region generation data enhancement method based on a deep convolutional generative adversarial network. According to the method, thyroid ultrasound images of a real patient are classified according to benign nodules and malignant nodules; a focus area image is generated by using a deep convolutional generative adversarial network; According to the method for generating image fusion in the lesion area, the image which is generated from the lesion area and is close to the real lesion is selected and fused with the thyroid image of a normal person toachieve the purpose of data enhancement, and the method for generating image fusion in the lesion area improves the quality and diversity of enhanced data, enables the generated image to be close tothe real image to the maximum extent and is more credible.

The invention discloses a thyroid nodule Tirads grading automatic identification model construction method and device. The method comprises the following steps: S1, carrying out data preprocessing on a thyroid ultrasound image; s2, performing ellipse fitting conversion on the preprocessed thyroid ultrasound image to generate a segmentation mask as a training label; and S3, building a multi-task model based on a convolutional neural network, inputting the training label into the multi-task model, and performing training by adopting a preset loss function to obtain a trained thyroid nodule Tirads grading automatic identification model. According to the invention, in thyroid nodule TI-RADS grading automatic identification, the latest convolutional neural network framework is used for carrying out automatic identification tasks on thyroid ultrasound images in an end-to-end mode, TI-RADS grading is completed, detection work can be completed in the early stage of thyroid cancer, and therefore expensive work such as puncture does not need to be used; and a doctor is assisted to complete thyroid cancer screening.

The invention discloses a thyroid ultrasound detection device and a detection method thereof. The detection device comprises a main ultrasound machine and a detection bed. The main ultrasound machineis placed adjacent to the detection bed, and detection device further includes a supporting moving device, a scanning device and an auxiliary device; and the supporting moving device is fixedly installed on the main ultrasound machine, the scanning device is disposed on the end part of the supporting moving device and is suspended from the detection bed, the auxiliary device is placed on the detection bed, the auxiliary device conforms to the human body mechanical design, and the supporting moving device and the scanning device are electrically connected with the main ultrasound machine separately. According to the thyroid ultrasound detection device and the detection method thereof, the cost of professional human resources is reduced, scanning is easier and faster, the waiting time of a subject is shortened, meanwhile, the quality of image information obtained by the scanning is higher, the diagnosis is more accurate, and an ultrasonic detection result is more standard and more versatile.

The invention relates to a thyroid ultrasound image nodule analysis method based on a deep learning network and shallow texture feature fusion, and belongs to the field of computer thyroid ultrasoundimage auxiliary analysis. The method comprises the following steps: preprocessing a thyroid ultrasound image; performing deep feature extraction based on a GooLeNet deep learning network; shallow texture feature extraction based on a rotation invariance local binary pattern; fusing the depth features and the shallow features; and classifying thyroid ultrasound images based on a cost-sensitive random forest. The thyroid nodule benign and malignant classification method can rapidly and accurately classify thyroid nodules by establishing the thyroid ultrasound image nodule analysis method.

The invention belongs to the technical field of image processing, and particularly relates to a thyroid nodule invasiveness prediction method based on target detection. The method comprises the following steps: S1, preprocessing a thyroid ultrasound image obtained clinically to obtain an original data set; S2, constructing a positioning network model of a network structure based on a traditional Faster RCNN, and pre-training the positioning network model; S3, extracting nodule form information in the ultrasonic image by using a positioning network, and obtaining aspect ratio information of the nodules and context information of gland tissues; S4, constructing a classification network model; S5, establishing a multi-model fusion thyroid nodule invasiveness prediction network, and predicting the thyroid nodule invasiveness in the ultrasonic image; and S6, training and updating the classification network model in the fused network model, and storing the model with the highest accuracy in the verification set. According to the method, end-to-end full-automatic auxiliary diagnosis can be achieved, and the defects that a traditional method is insufficient in accuracy and low in detection rate are overcome.

The invention discloses a thyroid nodule ultrasonic image classification method based on a capsule network, and the method comprises the following steps: carrying out the preprocessing of a thyroid ultrasonic image, and recognizing an artificial mark in the thyroid nodule ultrasonic image through the combination of a neural network and image differentiation; obtaining a rectangular frame for defining a thyroid nodule position through boundary adjustment; constructing a capsule network, applying the capsule network to classification of thyroid nodule ultrasonic images, and adjusting the structure of the capsule network; and adding Dropout into adjusted capsule network, so that training process is stabilized, and improving classification effect. According to the method, the problems that important information is lost in the pooling process and a trained model does not have translation and rotation invariance in the traditional convolutional neural network classification technology are effectively solved, and the accuracy of a capsule network in a thyroid ultrasound image classification task is improved.

The invention discloses a method and a device for automatically diagnosing benign and malignant thyroid nodules. The method comprises the following steps: step 1, carrying out noise reduction preprocessing on an original thyroid ultrasound image; 2, performing data enhancement on the preprocessed thyroid ultrasound image; 3, the thyroid ultrasound image after data enhancement being respectively put into three convolutional neural network models which are constructed through ResNet, DenseNet or ResNext in advance for training; and 4, carrying out model integration on images trained by different models in the step 3. According to the method, the problem that the automatic classification effect of benign and malignant thyroid nodules is poor is solved, the deep learning image automatic recognition technology is applied, the latest framework is used for extracting image features, the automatic classification task of benign and malignant thyroid nodules is completed in an end-to-end mode, and meanwhile high accuracy is guaranteed.

The invention discloses a thyroid nodule semi-supervised segmentation method based on an attention mechanism. The method comprises the following steps of: 1, carrying out preprocessing of a thyroid ultrasonic image, and removing an edge information region in the image; 2, constructing a semi-supervised segmentation neural network, performing classification and segmentation prediction tasks on theultrasonic image, and adjusting a network structure to adapt to a specific application scene; 3, adding an attention mechanism into the semi-supervised segmentation neural network to improve the network effect; 4, measuring the performances of a semi-supervised segmentation algorithm and an existing full-supervised segmentation algorithm in the field of thyroid nodule auxiliary diagnosis through an intersection-parallel ratio and a Dice coefficient; and 5, continuously reducing the number of the pixel-level labels, and observing the change condition of the network performance. According to theinvention, the thyroid nodule semi-supervised segmentation method based on an attention mechanism benefits from the semi-supervised effect of a small number of pixel-level labels while keeping the high segmentation performance of the semi-supervised segmentation model, learns the real benign and malignant characteristics of the nodules and improves the benign and malignant classification capacity.

The invention relates to a method for generating medical ultrasonic image data based on an adversarial network, and the method comprises the steps: enabling the adversarial network to be applied to anROI (region of interest) data set of thyroid ultrasonic data, and proving that a WGAN (generative adversarial network) can generate an ultrasonic image with better quality; expanding the benign thyroid ultrasonic data ROI data set and the malignant thyroid ultrasonic data ROI data set by using the WGAN in the data set; and comparing the performance of the VGG-16 network on the enhanced data set and the original data set, and proving the feasibility of performing thyroid ultrasound data set data enhancement through the generative adversarial network. The method is scientific and reasonable indesign, generates medical ultrasonic image data through the generative adversarial network, enriches a data set, and improves the adaptability of the model and the identification capability of a target.

The invention discloses a thyroid ultrasound image classification and segmentation network, a training method, a device and a medium, and belongs to the technical field of medical information. The network comprises a coding stage, a decoding stage and jump connection between the coding stage and the decoding stage, an attention mechanism module is arranged in the coding stage, so that the network pays more attention to the difference between a thyroid structure and surrounding tissues, and in the coding stage of the network, an output feature map is used as the input of a classification network, so that feature sharing can be carried out on the segmentation and classification tasks, the thyroid ultrasound images are classified and segmented at the same time through the same network, classification and segmentation task training is carried out at the same time in training, and the segmentation and classification accuracy is improved.

The invention provides a method for detecting nodules in a thyroid ultrasound image based on deep learning. The method comprises the steps of preprocessing the thyroid ultrasound image; extracting features of the preprocessed thyroid ultrasound image to obtain a feature image; respectively inputting the obtained feature images into corresponding classification and regression structures, and obtaining specific position information of a thyroid nodule region in each feature image; for the classification loss, the central point distance regression loss and the offset loss generated by calculation of the feature images input into the corresponding classification and regression structures, obtaining the total loss of the to-be-trained model through weighted summation calculation; and training and testing the to-be-trained model. According to the method, an anchor box does not need to be arranged, the nodule region in the thyroid ultrasound image is efficiently detected, calculation and resource waste related to the anchor box are avoided, the training speed is increased, and the generalization performance of an experimental result is enhanced. The invention further provides a device for detecting the nodules in the thyroid ultrasound image based on deep learning.

The invention discloses a thyroid nodule automatic identification model construction method, system and device based on a convolutional neural network, and the thyroid nodule automatic identificationmodel construction method comprises the steps: carrying out the data noise reduction of a thyroid ultrasonic image, and obtaining a training data set and a test data set; inputting the training data set into ResNet, DenseNet, and/or ResNext to carry out thyroid nodule malignant probability training to obtain a thyroid nodule automatic identification model; and testing the thyroid nodule automaticidentification model by adopting the test data set.

The invention relates to a method for constructing a thyroid ultrasound domain ontology, which is characterized by comprising the following steps: step 1, carrying out data preprocessing on a thyroidultrasound report; step 2, performing entity extraction; step 3, extracting a dependency relationship; 4, extracting a semantic relation; and 5, constructing a thyroid ultrasound field body. In a thyroid ultrasound report, the main concern lies in the lesion condition of thyroid and thyroid lesions, and the knowledge of other tissues or gene levels of a human body are not needed to be concerned too much, so that a medical ontology suitable for the thyroid ultrasound field is constructed on the basis of anatomy. Useful diagnosis and treatment information can be better extracted from an ultrasonic report by applying the thyroid ultrasound domain ontology, so that a doctor can be better assisted in disease diagnosis and treatment.

The invention discloses a method for dynamically identifying benign and malignant nodules based on thyroid ultrasonic video streams. The method comprises the steps of obtaining a transverse scanning section video and a longitudinal scanning section video, the transverse scanning section video and the longitudinal scanning section video are sent to an ultrasonic doctor for labeling; preprocessing the transverse scanning section video and the longitudinal scanning section video labeled by the sonographer; respectively obtaining a preprocessed transverse scanning section video and a preprocessedlongitudinal scanning section video; and respectively inputting the transverse scanning section video and the longitudinal scanning section video into the two Retina positioning networks to respectively obtain nodule related information of each frame of image in the transverse scanning section video and the longitudinal scanning section video, and performing denoising processing on the transversescanning section video and the longitudinal scanning section video. The method can overcome the technical problems that an existing thyroid nodule benign and malignant identification method has greatmisjudgment possibility, misdiagnosis and missed diagnosis are easy to occur, and correct treatment of a patient is delayed.

The invention discloses a thyroid nodule risk prediction method and system based on big data. The method comprises following steps of S1: acquisition of clinic data: collecting data of medical histories, signs and symptoms of a patient and recording the data; S2: ultrasonic examination: carrying out thyroid nodule checking of the patient, knowing the size, the shape, the nodule edge, the inner structure, the echo features, the blood flow conditions and the cervical lymph node conditions of the nodule; and S3: fine needle aspiration examination: carrying out thyroid nodule fine needle aspiration cytological examination on the patient, checking result data and carrying out recording. According to the invention, by acquiring clinic data of the patient, carrying out the ultrasonic examinationand the fine needle aspiration examination on the patient, and integrating the examination data, rapid extraction of the data is facilitated and risk prediction can be rapidly made.

The embodiment of the invention provides a thyroid nodule edge sign classification method, device and system. The method comprises the following steps: acquiring a thyroid ultrasound image training set; constructing a convolutional neural network model; taking a thyroid ultrasound image in the training set as the input of the convolutional neural network model, taking a classification result as the output of the convolutional neural network model, and training the convolutional neural network model; classifying thyroid nodule edge signs in the thyroid ultrasound image to be processed according to the convolutional neural network model. According to the thyroid nodule edge sign classification method and device, a doctor can be assisted in completing classification of thyroid nodule edge signs, unnecessary puncture operations caused by TI-RADS grading errors due to inaccurate edge sign classification are avoided, and body, money and spirit burdens of a patient are relieved.

The invention provides a method for generating a thyroid ultrasonic report by voice input of B-ultrasonic abnormal keywords. The method comprises the following steps: defining a thyroid ultrasonic semantic tree; generating a substructure according to the thyroid ultrasonic semantic tree; performing B-ultrasonic abnormal keyword voice input; positioning a description range; positioning a gland background description part and carrying out attribute positioning; positioning focal lesion description nodules and carrying out attribute positioning; supplementing necessary attributes; supplementing default information; generating text. According to the method, a B-ultrasonic doctor inputs B-ultrasonic exception description through voice while carrying out ultrasonic imaging diagnosis, and does not need to manually input an ultrasonic text report by an additional doctor, so that the manpower is greatly saved. After voice input of a doctor is finished, normal attribute values of unmentioned attributes can be automatically filled, voice description of a large amount of default redundant information by the doctor is avoided, and the integrity of a report is ensured under the condition that voice input is reduced as much as possible. Finally, a text is generated based on the generated sample tree, and the problem of report ambiguity is solved.

The invention relates to a thyroid myeloid cancer ultrasonic image identification method based on clinical prior knowledge guidance. Firstly, a thyroid nodule area is obtained by designing a cascade segmentation network, then a multi-branch classification network is designed, and according to clinical knowledge guidance of radiologists, important thyroid ultrasound image feature validity and calcification are added as prior knowledge to assist network classification. The method not only can distinguish benign and malignant nodules in the image, but also can accurately distinguish the malignant nodules into myeloid thyroid carcinoma and papillary thyroid carcinoma.