93 results about "Lesion Identification" patented technology

The invention discloses a diabetic foot early-stage detection method combining infrared and visible light imaging, and relates to the technical field of diabetic foot medical imaging diagnos.The diabetic foot early-stage detection method comprises the steps that multi-mode image collection and standardization processing are conducted, infrared images and visible light images are synchronously collected, and standardization processing such as size normalization is conducted; image registration and space alignment are carried out, marking points are set based on foot anatomical features, and feature extraction, mismatching point elimination and transformation matrix calculation are carried out; extracting and screening multi-dimensional features, extracting temperature and structural features, and screening by using a Relief-F algorithm; feature lesion recognition and classification are fused, and lesion probability is output through a double-branch convolutional neural network; carrying out detection result verification and feedback optimization, and comparing a clinical diagnosis optimization model; and generating a detection report and storing data, and generating a report containing the fused image. The early lesion detection precision is improved through multi-modal fusion, individual and environment differences are adjusted and adapted in a personalized mode, and reliable technical support is provided for clinic.

The invention discloses an intracranial SEEG electroencephalogram data analysis method and system based on artificial intelligence, and relates to the technical field of data analys.The method comprises the steps that original electroencephalogram signals are obtained, and electrode space coordinates and imaging information are synchronized; extracting robust features in a time-frequency domain through synchronous extrusion in combination with learnable chirp and multi-resolution attention; according to the electrode coordinates and the anatomical topology, topology sensing alignment is executed, and an anatomical function diagram is constructed; performing continuous time evolution modeling on a node state through a graph neural continuous dynamic structure, and introducing an energy conservation and event jump mechanism to realize focus transmission chain inference; estimating orientation information in different physiological contexts, and screening a stable direction as a prior constraint; and outputting a continuous risk curve according to propagation embedding, and generating a coverage-controllable confidence interval through online conformal calibration. Energy consistency, causal stability and space-time continuous modeling of SEEG signals can be realized, and the accuracy and interpretability of electroencephalogram focus recognition and clinical risk assessment are improved.

The invention discloses a magnetic resonance image target identification method based on multi-modal feature fusion, and the method comprises the steps: firstly obtaining multi-sequence magnetic resonance image data of the same object, the data comprising a structure sequence, a function sequence and a quantization parameter sequence, and carrying out the synchronous preprocessing; the pre-processed data is input to a feature extraction network to obtain a multi-dimensional feature representation. A cross-modal interaction map is constructed based on feature representation, interaction and fusion of different modal features are realized through an attention mechanism, and a fusion feature vector is obtained. And generating a focus candidate region set, and correcting the candidate region in combination with the related information of the patient to obtain a corrected focus region set. And inputting the corrected target area set into a discrimination network, and outputting a positioning result and a type result of the target. According to the method, through multi-modal feature fusion and map modeling, the accuracy and robustness of focus recognition are effectively improved, misjudgment caused by single-modal limitation is reduced, and the method has high clinical application value.

The application provides a crop disease precision identification method and system, and relates to the technical field of crop disease identification. The method comprises the following steps: establishing standard lesion feature image datasets of different crops and different organs; constructing a background feature database of common crop diseases; forming a crop type and organ identification algorithm model and an organ lesion identification algorithm model; determining the crop type and the organ of the crop in an image to be identified based on the crop type and organ identification algorithm model; determining the disease range of the image to be identified based on the organ of the crop and using the organ lesion identification algorithm model corresponding to the organ of the crop, and outputting a lesion disease with similar features; outputting a matched crop disease type based on the background feature database of common crop diseases; and obtaining a common disease type in the determined lesion disease and the determined crop disease type, so as to realize the precision identification of the disease. The application classifies and grades to form a structured disease identification process, and improves the identification precision.

The present application relates to the technical field of lesion recognition, in particular to an automatic lesion recognition ultrasonic system for real-time monitoring, which comprises a delay path discrimination module, a gray level trend detection module, a texture feature screening module, a feature fusion sorting module and a region highlight labeling module. Based on continuous ultrasonic frames, the collected deep tissue echo path is analyzed, and the echo arrival time of each pixel point in the continuous frame is detected. The present application supports multi-type data fusion judgment through a comprehensive judgment process supported by multi-dimensional parameter collaborative screening, penetration behavior, gray level trend and texture aggregation. The regional abnormal priority sorting mode improves the hierarchy of lesion feature discrimination, provides partition directional recognition for local structural abnormalities and early micro-variation, and converts the feature judgment result into a high confidence region label by a weight aggregation method. The image output process automatically completes the real-time visual presentation of the high-risk area, improving the clarity and pertinence of the lesion region presentation.

An adaptable deep learning method is provided that delivers sound hepatic lesion identification in NETs, while significantly reducing human effort for data annotation and improving model generalizability for PET image quantification. A region-guided GAN (RGGAN) model conducts image-to-image translation between list-mode simulated PET images and real-world clinical data, while preserving semantic content of interest, e.g., lesions. The RG-GAN model is integrated with a lesion detection model into an end-to-end, unified framework for joint-task learning, such that the two models can benefit from each other. The RG-GAN translates the list-mode simulated data into real world-style images, which appear to be drawn from the real clinical PET image dataset, and feeds the translated images into the lesion detection model for training. In order to deal with the limited diversity of list mode-simulated PET image data, a specific data augmentation module is incorporated into the unified framework to improve model training.

The application provides a liver hydatid disease screening method and system based on plain CT and deep learning, belongs to the technical field of intelligent medical treatment, and realizes liver hydatid disease identification based on plain CT by combining a liver segmentation model, a liver hydatid lesion identification model and a liver hydatid screening model, reduces the risk of patients of intake of more radiation due to enhanced CT, reduces the risk of patients of allergy due to injection of contrast agent, and has strong clinical applicability; can realize patient-level diagnosis on the basis of slice-level identification, classification and segmentation, and has interpretability; in combination with the real situation of hydatid disease in clinical diagnosis and treatment and the clinical characteristics of hydatid disease, the identification accuracy is greatly improved, and in addition, the population including healthy people can be identified and screened, and the use range is wide.

The invention belongs to the technical field of artificial intelligence and medical image processing, particularly relates to an intelligent auxiliary diagnosis system for radiographic images, and aims to solve the problems of missed diagnosis and misdiagnosis caused by overload of doctors, large subjective difference and low early focus recognition sensitivity in the prior art. The system integrates a multi-scale feature extraction engine, a cross-modal alignment network, a pathological semantic reasoning unit and a diagnosis consensus generator, and realizes multi-modal image fusion analysis, dynamic knowledge reasoning and man-machine collaborative decision. A thermodynamic diagram and a feature contribution degree map are output through an interpretability visualization module, and a structured report conforming to clinical specifications is generated by a report automation engine, so that the diagnosis accuracy and efficiency are improved, the report period is shortened, and credible deployment of AI auxiliary diagnosis in a complex medical scene is supported.

The present invention relates to an integrated diagnostic system for deep endometriosis and / or adenomyosis, comprising three interconnected platforms and a method for analysing images generated by algorithms. The first platform combines patient medical history data with transvaginal ultrasound and / or pelvic magnetic resonance images, using algorithms to generate clear and detailed images of lesions. The method for analysing these images includes pre-processing, merging of clinical data and images, lesion identification, and comparison with a clinical database, resulting in a visual and descriptive analysis of the lesions, facilitating early and assertive diagnosis. The second platform is a communication network among specialist physicians, who perform a second reading of the images and issue a validated final report. The third platform is educational, offering personalised training and mentoring for imaging physicians using the images generated by the first platform to improve their diagnostic skills. The system reduces diagnostic time, increases accuracy, and contributes to the continuous professional development of physicians involved in the management of deep endometriosis and / or adenomyosis.

The present disclosure relates to a lumbar vertebra lesion identification method, device, medium and electronic equipment, wherein the method comprises: determining a lumbar vertebra image of a user, the lumbar vertebra image being a T2 weighted magnetic resonance image; inputting the lumbar vertebra image into a target detection model to obtain a lesion category of the lumbar vertebra of the user, wherein the target detection model comprises a down-sampling layer and a pooling layer, the down-sampling layer is used to extract feature data of the lumbar vertebra image, and the pooling layer is used to classify the feature data to obtain the lumbar vertebra lesion category of the user. Compared with the way of judging the lumbar vertebra lesion by artificial judgment and identifying the lumbar vertebra lesion by machine learning method in the related art, the error of identifying the lumbar vertebra lesion type can be reduced, and thus the accuracy of identifying the lumbar vertebra lesion can be improved.

The application relates to the field of image analysis, in particular to an AKI diagnosis method fusing multi-modal image features, which obtains kidney area MRI and CT images, extracts kidney pelvis candidate response boundary change information, completes area alignment and multi-scale response fusion through path aggregation, priority judgment and directionality adjustment, outputs a diagnosis graph and combines abnormal expansion direction to optimize an image set. According to the application, a high-response pixel set of a kidney pelvis area in an MRI image is extracted, pixel density and edge gradient change trend are combined, boundary change feature groups are established, the spatial distribution characteristics of potential lesion areas in the image can be finely described, and the initial accuracy of lesion identification is enhanced. The connectivity stability and path consistency of a response block are taken as the measurement basis, a path sorting mechanism in path construction is introduced, and the structural integrity and boundary controllability are improved in the feature path screening stage.

The application relates to the technical field of image recognition, and discloses an electronic laryngoscope image recognition method and system based on artificial intelligence. The method comprises the following steps: acquiring a continuous image sequence of a throat region, and recognizing a suspected lesion candidate region in the image through an identification model; acquiring a first image sequence, and judging whether each first candidate region in the first image exists in the sequence; if yes, calculating a motion vector consistency and rigidity degree parameter of the region based on the first image sequence, and judging whether the region is a lesion region according to the parameters; if no, acquiring a subsequent second image sequence, and judging whether the disappeared candidate region appears again; for the candidate region appearing again, calculating related parameters of the candidate region by comprehensively considering the first and second image sequences, and finally judging the candidate region by combining image features. The application can improve the accuracy and reliability of electronic laryngoscope image lesion recognition.

The invention relates to the technical field of biomedical engineering, and provides a tongue body motion form reconstruction and neuromuscular lesion recognition method, which comprises the following steps: S1, acquiring a multi-modal original signal by using a flexible electrode array attached to a lingual surface, and recording a system timestamp; s2, performing targeted filtering, artifact removal and feature extraction processing on the multi-mode original signal; s3, reconstructing a local dynamic curved surface of the lingual surface by adopting a preset model, and inferring an overall position and posture sequence of the tongue body in combination with an IMU signal; s4, a joint optimization objective function is constructed, and a tongue body three-dimensional dynamic form sequence is obtained through iterative solution; s5, performing motion unit action potential MUAP decomposition and feature extraction on the surface electromyogram signal sEMG; and S6, constructing lesion feature vectors based on the multi-dimensional abnormal features, and analyzing and generating a tongue lesion risk map TDM through a machine learning model. The defects that real motion reconstruction, form inference and neuromuscular lesion positioning of the tongue cannot be achieved through an existing tongue detection technology are overcome.

The present application relates to a kind of based on multimodal ultrasound time series data lesion identification method.It includes: obtaining the multimodal target time series data generated by ultrasound scanning to organ, the multimodal target time series data includes at least a group of multimodal data to be examined, multimodal target time series data is loaded to lesion instance segmentation model, to utilize the lesion instance segmentation model to carry out lesion segmentation identification processing to multimodal target time series data, and output the lesion instance state information corresponding to current multimodal target time series data;When it is determined that there is lesion in multimodal target time series data by lesion segmentation identification processing, then the lesion instance state information at least includes the category of each lesion.The present application can realize the continuous detection identification to lesion, improve the precision and reliability of lesion identification.

The invention discloses a bladder cancer early screening method based on plain-scan CT and MRI time sequence fusion, which realizes high-precision lesion recognition and positioning under the condition of low-dependence enhanced scanning through knowledge distillation and bimodal feature fusion. The method comprises the following steps: constructing a multi-modal data set including plain scanning, arteriovenous phase enhanced CT and MRI; a key frame sparse annotation and linear interpolation complementation strategy is adopted to reduce the annotation cost; performing preprocessing such as window level adjustment, normalization and zooming on the image; a CT / MRI recognition and positioning network is constructed, each sub-network comprises a 2D positioning branch and a 3D classification branch, and the characterization capability of plain scanning data is enhanced through a teacher-student knowledge distillation mechanism; and finally, integrating the bimodal information through a feature fusion module, and outputting a classification and positioning result. According to the method, time sequence consistency and modal consistency constraints are introduced, the recognition robustness of small focus and unclear boundary areas is improved, the manual film reading burden is remarkably reduced, the screening efficiency and safety are improved, and the method is suitable for early large-scale screening scenes of bladder cancer and has important clinical popularization value.

The invention relates to an AI-assisted lumbar intervertebral disc resection visual navigation device with a postoperative care monitoring function, and belongs to the technical field of medical equipment.The AI-assisted lumbar intervertebral disc resection visual navigation device comprises an AI assisting module, a visual navigation module, a postoperative care monitoring module and a data processing and transmission module, and the AI assisting module is internally provided with an image recognition model and a risk prediction model; the system is used for performing lesion recognition and positioning on image data of a patient before an operation, performing operation risk assessment in combination with multi-source information of the patient, and overlapping and displaying the operation risk in a three-dimensional model to realize real-time navigation in the operation, the AI auxiliary module realizes accurate lesion positioning and risk quantitative assessment, and the visual navigation module provides real-time guidance with the precision smaller than or equal to 0.5 mm. The postoperative nursing monitoring module achieves all-around real-time monitoring of physiological parameters, wound states and rehabilitation training, and the abnormity early warning subunit discovers the risks in time and treats suggestions in a grading mode to improve nursing pertinence.

The application belongs to the technical field of image recognition, and discloses a weakly supervised interstitial lung disease lesion recognition method based on multiple example learning, which comprises the following steps: step 1: acquiring CT image samples; step 2: selecting part of the CT images, and manually labeling the lung parenchyma in the images; step 3: establishing a lung parenchyma segmentation model through a saliency segmentation algorithm, inputting the manually labeled CT image samples to perform training and testing, and obtaining a trained lung parenchyma segmentation model; step 4: training a lesion recognition model using a multiple example learning algorithm; step 5: acquiring a to-be-recognized CT image sample, inputting the lung parenchyma segmentation model to perform segmentation, and then inputting the segmented data sample into the lesion recognition model to obtain a lesion position. The application can realize the visual labeling of interstitial lung disease through a small amount of labeling, and greatly improves the recognition efficiency.

The application provides a key frame extraction and report generation method for a light reflection-oriented endoscopic video, comprising: evaluating the aggregation degree of the visible arc length at the lesion edge, determining the attention priority order of the lesion edge area according to the aggregation degree; dynamically adjusting the key frame screening condition by combining the attention priority order with the contour concave-convex degree, preferentially retaining the image frames with obvious local morphology of the lesion, and obtaining a candidate key frame set; positioning the abnormal area in each frame according to the candidate key frame set, matching the abnormal area with a reference template of an anatomical position description, and evaluating the specific position of the abnormal area in the anatomical structure; combining the anatomical position description and the abnormal area positioning information according to the final output key frame combination, generating a structured endoscopic video analysis report, and completing the lesion identification and description task under the interference of light reflection.

This invention belongs to the field of image processing, specifically disclosing a precise gastrointestinal lesion identification system based on image processing. The system includes an image acquisition module, an image preprocessing module, a lesion identification module, and a result output module. This solution achieves precise noise removal from gastrointestinal images while fully preserving the detailed features of lesion areas by constructing a standardized noisy dataset through noise simulation, employing a 4-level feature pyramid fully convolutional encoder-decoder architecture, incorporating deep-width residual blocks, multi-head attention, and convolutional gating mechanisms, and using L1 pixel reconstruction loss to train the model. A customized deep learning model is built based on EfficientNetB5, incorporating multiple regularization constraints, and employing an Adamax optimizer and classification cross-entropy loss for training, achieving efficient extraction of fine-grained features of gastrointestinal lesions, effectively suppressing model overfitting, and accelerating training convergence.

This application provides a comprehensive non-destructive detection method for internal lesions in potatoes based on multispectral feature screening and deep learning. The method includes: S1 collecting potato samples and obtaining their spectral data; S2 selecting a combination of CARS-VIP and CARS-SPA screening adapted to absorbance spectra, and a dual-wavelength correlation coefficient method adapted to energy spectra, based on the spectral data, to obtain a subset of characteristic wavelengths; S3 constructing a three-level discrimination model based on the subset of characteristic wavelengths to identify various internal lesions in potatoes. This model sequentially distinguishes between healthy and diseased samples, further subdivides lesion types in diseased samples, and enhances discrimination through an improved residual neural network for early or boundary lesion samples. This method, through multi-strategy feature screening and a three-level discrimination model, achieves high-precision non-destructive detection of various internal lesions in potatoes, significantly improving the sensitivity of early, subtle lesion identification, and providing efficient technical support for quality control in the potato industry.

The invention discloses a liver single-cell necrosis intelligent identification method and system based on deep learning and a storage medium, and the method comprises the steps: obtaining and segmenting a WSI labeling file obtained after labeling a single-cell necrosis lesion in a liver WSI image, and obtaining Tile and Mask files; generating a training set and a test set; constructing a ResUNet model based on an encoder-decoder symmetric architecture, and training the ResUNet model to obtain an identification model; a weight based on a positive pixel ratio is introduced in the training process and is used for amplifying positive sample loss; evaluating and verifying the identification model by using the test set to obtain a trained identification model; and packaging and deploying the trained recognition model, performing intelligent recognition on the input WSI image, and outputting a single-cell necrosis lesion recognition result. By means of the method, interactive instant reasoning feedback and result visualization can be achieved, and an efficient, accurate and standardized technical solution is provided for toxicity pathology assessment.

The application relates to the technical field of medical image processing, and particularly discloses a cavity image real-time lesion recognition and boundary segmentation system. A cross-sectional image sequence in a cavity is acquired through an interventional imaging catheter, and after shared features are extracted by using a deep network, pixel-level tissue boundary segmentation and overall lesion property recognition are synchronously performed. The core lies in that a dynamic gradient coordination mechanism is introduced, gradient direction conflicts between the segmentation and recognition double tasks are detected and resolved in real time during the training process, and adaptive projection transformation based on the task convergence state is used to fuse the gradients, so that the shared network can learn balanced features which are optimal for both tasks; the method realizes end-to-end real-time processing from image input to synchronous output of the segmentation and recognition results, and significantly improves the synergy, accuracy of lesion analysis and real-time auxiliary efficiency of clinical operation navigation.

The application provides a fluorescence microscopic endoscopy system and method capable of adjusting liquid inflow and concentration in real time, and belongs to the field of microscopic imaging. The system comprises an integrated image transmission fiber probe (integrated liquid inflow channel and image transmission fiber bundle), a double-channel electric infusion unit (stain pump and diluent pump), an optical path control module (switchable filter and wide spectrum light source), and an image feedback module. A high-sensitivity camera is used to collect fluorescence images, and a main control board dynamically adjusts the concentration (mixed stain and diluent ratio) and spraying amount of the stain based on image brightness / contrast features, thereby solving the problems of traditional split-type operation complexity, uncontrollable stain amount, and poor imaging quality. Advantages include: single insertion for spraying and imaging, shortening detection time; image feedback closed-loop control improves staining accuracy; and expandable multi-spectral imaging. The application is suitable for early cancer screening of the digestive tract and significantly improves lesion recognition rate.

The application discloses a two-stage radiomics lesion identification and positioning method and device, adopts a multilayer perceptron network to analyze multi-modal image data, detects FCD by extracting features taking the gray matter as a region of interest by using a radiomics method, and the features combine shape, first-order statistics and texture features from multi-modal and wavelet images. In addition, the application also introduces asymmetric features of left and right hemispheres, avoids potential interference in the contralateral area of the unilateral FCD patient caused by the compensatory mechanism of the left and right hemispheres of epilepsy. According to the rich high-dimensional features and asymmetric features of radiomics, the sensitive features of FCD are fully explored, the two-stage detection method is combined to identify FCD abnormalities, the extracted features are more perfect, false positive results are avoided, and the accuracy and sensitivity of detecting FCD are improved from different scales from coarse to fine granularity.

The invention discloses a CBCT image condyle lesion recognition method based on multi-modal information. The CBCT image, patient quantitative data and a medical knowledge corpus are preprocessed; extracting an image depth feature vector from the preprocessed CBCT image by using a deep learning model, and obtaining a structured quantitative data feature vector; performing preliminary fusion on the image features and the quantized data features to generate a comprehensive query vector; searching in a pre-coded medical knowledge text embedding library by using the query vector so as to dynamically obtain knowledge most related to the current case and form a knowledge feature vector; and finally fusing the image features, the quantized data features and the knowledge features to form a final fusion feature vector, and inputting the final fusion feature vector into a lesion recognition model to output lesion probability. The invention further provides a system for recognizing the condylar lesion in the CBCT image based on the multi-modal information. According to the method, the accuracy, robustness and interpretability of condylar lesion recognition are remarkably improved.

A method and apparatus for identifying fundus lesions are provided. The method includes: acquiring OCT fundus images of a specific patient; using a first neural network model to determine the boundary information of retinal tissue in the OCT fundus images, and segmenting the OCT fundus images into multiple image blocks based on the boundary information of the retinal tissue; providing the multiple image blocks to a second neural network model to obtain the fundus lesion type of the specific patient; the second neural network model calculating the feature vector of each image block and the attention weight of each image block for the specified fundus lesion type, and obtaining the fused feature vector through weighted calculation to obtain the identification result for the specified fundus lesion type. The second neural network model combines spatial saliency and scanning position saliency to enhance important information in the image block group and filter out unimportant information, thereby achieving low-cost model training and application.

This invention discloses an automatic rating method, device, equipment, and storage medium for maize gray spot disease. The method involves acquiring an RGB image of the ear-side leaves of maize to be rated against a solid-color background, preprocessing the RGB image to obtain a binarized image of the leaf area, using the binarized image as a mask to separate the channel values ​​of the target RGB image within the leaf area to obtain a binary image of the lesion, filtering the binary image of the lesion, extracting the outline of the filtered lesion area, generating a final binary image of the lesion based on the outline, and inputting the final binary image of the lesion into an automatic gray spot disease rating model to obtain the gray spot disease level of the ear-side leaves to be rated. This method enables high-precision, high-efficiency, and low-cost automatic rating of maize gray spot disease, improves the accuracy of lesion identification, has good mobile device adaptability, meets the needs of rapid field diagnosis, and has strong model generalization ability, thus improving the speed and efficiency of automatic rating of maize gray spot disease.

The present application relates to a kind of multispectral imaging methods for human electronic angioscope, belong to medical instrument technical field.Therein, the method includes: by ultramicro infrared enhanced image sensor, photoelectric signal conversion is carried out to target near-infrared spectrum, based on focusing lens, the imaging light in blood vessel is focused, the overall specification is calibrated to adapt to the intervention demand of most blood vessels of human body;Based on the requirement of avoiding blood interference, preset three kinds of near-infrared spectrum wave band, corresponding multispectral light source is built and is coupled with endoscope optics;Under the assistance of image device, endoscope is pushed to target blood vessel, and image is collected according to wave band and is transmitted to external terminal;Terminal passes through adaptive algorithm, and noise reduction, signal enhancement, generates visual image.It realizes real-time optical direct imaging in blood vessel, solves the problems such as no direct optical imaging, blood interference, low resolution of traditional technology, improves the lesion recognition clarity and diagnostic efficiency, provides support for intravascular precise diagnosis and treatment.

The invention provides a medical endoscope image intelligent enhancement and focus recognition system, and relates to the technical field of data processing. The region segmentation module is used for detecting and distinguishing a suspected microvascular network region and a background region; the structure maintaining module is used for executing smoothing and color correction processing on the background area and executing texture maintaining processing on the suspected microvascular network area; the difference highlighting module is used for calculating the pixel difference between the suspected microvascular network region and the adjacent background region and executing local contrast enhancement in the region; the dynamic correction module is used for analyzing the brightness distribution relation between the suspected microvascular network area and the adjacent background area and executing self-adaptive brightness compression or lifting; the lesion recognition module is used for extracting complete blood vessel texture and a spatial distribution mode thereof, generating a lesion recognition result and outputting a suspicious lesion area; according to the invention, the accuracy of endoscope image focus identification is improved.

The invention provides a pathology image diagnosis method based on GDKAN and multi-order context interaction gating, and belongs to the technical field of medical image processing and artificial intelligence crossing. The method comprises the following steps: step 1, collecting a full-view pathology full-slice image; step 2, constructing a self-adaptive packet dynamic Kolmogorov-Arnold network, and constructing a dynamic Kolmogorov-Arnold network; step 3, constructing a GDKannform coding module of adaptive grouping; step 4, establishing a multi-order context interaction gating mechanism; step 5, a GDKannform model is obtained through training; and step 6, outputting a pathological image lesion identification result. According to the method, automatic diagnosis of the pathological image is realized through a full-process design of preprocessing, dynamic feature optimization, multi-order context interaction gating fusion and classification optimization.