17 results about "Pathology Report" patented technology

The invention discloses an interpretable cancer survival prediction method based on multi-modal data fusion, and the method comprises the following steps: carrying out the preprocessing of a full-section pathological image, and obtaining an image feature matrix; preprocessing a pathological report text corresponding to the pathological image, and constructing a text feature matrix; preprocessing the high-dimensional gene expression data of the patient to generate a plurality of survival-related gene modules and feature vectors thereof; performing dynamic fusion on the obtained multi-modal features through a self-adaptive multi-modal expert hybrid module to obtain final fusion feature representation; and training a deep learning model by using fusion feature representation in combination with a negative log-likelihood loss function and a Cox proportional risk model, and performing prognosis analysis on the cancer patient. According to the explainable cancer survival prediction method based on multi-modal data fusion, the prediction accuracy is improved, good model interpretability is achieved, and biomarkers closely related to cancer prognosis can be automatically recognized.

The invention provides a cell pathology intelligent auxiliary method based on a CLIP-RETRIEVAL model, and relates to the technical field of medical image processing, and the method comprises the following steps: S10, obtaining data of a cervical cell pathology image and a pathology text, and constructing a structured data set of an image-text pair; s20, preprocessing the cervical cell pathological image, and respectively generating corresponding feature vectors according to the preprocessed cervical cell pathological image and pathological text; s30, constructing a unified medical semantic embedding space, aligning feature vectors of the image-text pairs through the dynamic projection matrix P, and outputting alignment features; and S40, inputting the alignment features into an LLM model, and generating a structured pathology report. Through data acquisition, preprocessing, multi-modal feature extraction, semantic alignment and vertical question and answer function adaptation, the problems of high misuse rate of medical terms and semantic segmentation of images and texts are solved, and the accuracy and interpretability of pathological diagnosis are improved.

The invention discloses a myelofibrosis grading prediction method and system of a full-slice image and a storage medium. The method comprises the steps of obtaining the full-slice image and a corresponding pathological detection report; inputting the full-slice image and the pathological detection report into a grading prediction model; through the hierarchical prediction model, according to the full-slice image, obtaining a full-slice image feature and a first block image feature; through the hierarchical prediction model, obtaining a first report feature and a first word segmentation feature according to the pathological detection report; reconstructing the first word segmentation feature according to the first block image feature to obtain a second word segmentation feature; obtaining a second report feature according to the second word segmentation feature; fusing the full-slice image features with the second report features to obtain fused features; and performing myelofibrosis grading prediction on the full-slice image according to the fusion features. By combining the full-slice image and the pathological detection report, the accuracy of grading prediction of myelofibrosis is improved.

The invention provides a full-process pathological image analysis method and device, and belongs to the technical field of medical artificial intelligence, and the method comprises the steps: obtaining a full-view digital WSI slice image from a gastric cancer patient, and carrying out the analysis of the obtained WSI slice image through a multi-task deep learning model, and obtaining pathological morphology data; based on the image features of a large number of WSI slice images and the semantic features of a large number of pathological texts, constructing a multi-modal feature library containing image text pairs, and based on the multi-modal feature library, performing multi-modal retrieval of pathological information on the to-be-retrieved texts or the to-be-retrieved images; and integrating the obtained pathological morphology data and the pathological information, and performing content filling according to a preset template to generate a pathological report. Therefore, through pathology image analysis based on multi-task deep learning, cross-modal interactive image-text retrieval and pathology report automatic generation, full-process intelligence of pathology analysis is realized, and efficiency and accuracy of gastric cancer pathology analysis are improved.

The invention relates to a pathological image semantic segmentation method based on multi-modal feature fusion and boundary enhancement. The method comprises the following steps: firstly, acquiring an original nephropathy image and an associated pathology report text, and preprocessing the original nephropathy image and the associated pathology report text; then, a multi-modal segmentation network is constructed, and the multi-modal segmentation network comprises an image encoder, a text encoder, an image-text multi-modal fusion and enhancement module, a shared decoder, a segmentation prediction head and an auxiliary boundary detection head; then, training the multi-modal segmentation network by adopting the preprocessed original nephropathy image and the associated pathological report text; and finally, inputting a to-be-segmented pathological image and an associated pathological report text into the trained multi-modal segmentation network, and outputting a target semantic segmentation result. A nephropathy image and a matched pathology report text are effectively and deeply fused and applied to a pixel-level semantic segmentation task, and a more accurate, robust and fine segmentation result is provided.

The invention relates to a visual template generation method for a medical pathology report, and the method comprises the steps: obtaining a specimen image of a target case and an associated material block information list, carrying out the semantic segmentation model reasoning based on the content of the specimen image, recognizing and extracting a key anatomical structure region with morphological stability and clinical significance, and carrying out the recognition of the key anatomical structure region; according to the method, a group of sparse and reasonably distributed semantic anchor point coordinate sets are generated, the semantic anchor point coordinate sets are correlated with the original size of an image and then are stored in a structured manner, and the spatial topological relation between graphic elements and anchor points is used as a dynamic calibration basis, so that accurate positioning and consistent reproduction of annotations are realized. According to the scheme, the reducibility, the accuracy and the interaction consistency of pathological image annotation in a cross-equipment scene can be improved, and the efficiency and the reliability of pathological auxiliary diagnosis and remote cooperation are improved.

The invention provides a clinical application level pathology large model construction method and system based on weak supervised learning. The method comprises the steps of performing weak supervised training on each historical clinical pathology image, adjusting an upstream learning parameter of a large model according to a training result, constructing an upstream task of the large model, and performing segmentation decoding on each weak supervised training result, determining a segmentation result, adjusting downstream learning parameters of the large model, constructing a large model downstream task, obtaining upstream fine features and downstream fine features, constructing an upstream and downstream cooperative task of the large model, and generating a pathological large model in combination with the large model upstream task and the large model downstream task. And inputting the current clinical pathological image into the large pathological model for pathological information identification, obtaining a plurality of pathological identification tags corresponding to the patient, and generating a pathological report of the patient, thereby reducing the calculation complexity when processing a high-resolution image, optimizing a collaboration mechanism among different tasks in the large model, and improving the efficiency of the large model. The method can be widely applied to pathological recognition technologies.

The invention provides a pathology agent system with reasoning ability. The pathology agent system comprises a pathology host agent, a pathology physician agent, a pathology section viewer, a pathology multi-mode large language model and a pathology report agent. The pathology host agent is responsible for coordinating a diagnostic pathology analysis process and strategically planning a pathological section area for detailed examination; the pathology main agent plans and coordinates the plurality of pathologist agents to investigate different pathological section tissue areas at the same time according to the formulated preliminary pathological diagnosis hypothesis and strategy; and each pathologist intelligent agent interacts with the pathological section viewer and the pathological multi-mode large language model, analyzes a specific pathological section area and feeds back a result to the pathologist intelligent agent. According to the invention, different pathological section areas can be studied step by step in a layered manner, and important background information such as tissue morphology and peripheral features can be captured based on analysis of a full pathological section level.

The invention discloses a curative effect prediction device, a curative effect prediction method, a storage medium and a processor. In the case, a data acquisition module acquires a full-slice image and a pathological report text; the feature extraction module extracts global tissue texture features and local cell structure features based on the full slice image; the analysis module determines the cell category and the target weight of the pathological tissue based on the pathological report text; the fusion module performs weighting processing on the local cell space structure features based on the cell category and the target weight, and performs feature fusion on the weighted local cell space structure features and the global tissue texture features to obtain fusion features; and the prediction module predicts the probability that the target object reaches the expected curative effect after being treated by the target therapy based on the fusion features of all the image blocks by a pre-trained curative effect prediction model. According to the application, high-precision curative effect prediction is realized through a multi-scale feature fusion mechanism guided by a pathological report, and the auxiliary value of a curative effect prediction result on medication and treatment of medical workers is effectively improved.

The present invention discloses an interpretable cancer survival prediction method based on multimodal data fusion, comprising the following steps: preprocessing full-slice pathology images to obtain an image feature matrix; preprocessing the pathology report text corresponding to the pathology image to construct a text feature matrix; preprocessing the patient's high-dimensional gene expression data to generate multiple survival-related gene modules and their feature vectors; dynamically fusing the obtained multimodal features through an adaptive multimodal expert hybrid module to obtain a final fusion feature representation; using the fusion feature representation, combining the negative log-likelihood loss function with the Cox proportional hazard model to train a deep learning model to perform prognosis analysis on cancer patients. The present invention adopts the above-mentioned interpretable cancer survival prediction method based on multimodal data fusion, which not only improves the prediction accuracy, but also has good model interpretability and can automatically identify biomarkers closely related to cancer prognosis.

This invention provides a management system for WSI pathology reports based on OFD-H. It includes a presentation layer, a business layer, a data storage layer, and a standard interface for interoperability with third-party medical systems. The presentation layer includes an editor for editing OFD reports and a WSI image reader capable of coordinate backtracking of WSI images. The business layer includes: module one for automatically generating pathology image heatmaps and textual diagnostic suggestions; module two for OFD format container encapsulation; module three for three-level streaming image backtracking; and module four for adding digital signatures and watermarks. The data storage layer is used for hierarchical storage of WSI files, ROI images, and OFD documents to meet the retrieval needs of massive multimodal data. This invention can achieve "intelligent assisted generation" of reports, "two-way interoperability" between text and images, "lightweight encapsulation" of data, and "compliant traceability" of documents.

The invention provides a pet pathology auxiliary diagnosis system based on visual language artificial intelligence. The computer vision AI pathological analysis module analyzes the whole pathological section map, detects a tumor area and identifies key pathological features; the intelligent pathological report generation module automatically generates a pathological diagnosis report based on a large language model in combination with an information retrieval enhancement generation (RAG) technology; the cloud digital section management system provides a remote virtual section reading function, supports pathological specialists to access and manage pathological sections anytime and anywhere, integrates AI auxiliary diagnosis and report generation capabilities, and realizes whole-process information tracking and management. According to the auxiliary diagnosis system, the manual film reading time can be shortened, the pathological resource configuration is optimized, and the overall working efficiency is improved.

The invention relates to a specimen-level pathological report automatic generation method based on artificial intelligence, and the method comprises the steps: obtaining a plurality of digital slide images belonging to the same pathological specimen, inputting the plurality of digital slide images into a pre-trained specimen-level artificial intelligence diagnosis model, receiving a diagnosis analysis result outputted by the specimen-level artificial intelligence diagnosis model, the diagnosis analysis result comprises at least one diagnosis element obtained by comprehensively analyzing the plurality of digital slide images, matching and calling a corresponding structured diagnosis template from a preset diagnosis template library according to the diagnosis analysis result, obtaining an initial diagnosis report draft based on the matched structured diagnosis template and the diagnosis analysis result, and sending the initial diagnosis report draft to a server; and presenting the initial diagnosis report draft and the diagnosis options capable of being adjusted by the user on an interactive interface, and updating the content of the initial diagnosis report draft in response to the interactive operation of the user to obtain a final pathological diagnosis report.

The invention discloses an image-text pathology recognition method and device based on multi-modal fusion, and the method comprises the steps: inputting an image block sequence into a UNI-Patch encoder, outputting the microscopic features of a fixed dimension, segmenting the image block sequence into a plurality of instances, inputting the instances into an MIL mask generator, learning and generating a feature mask, and carrying out the recognition of the feature mask. The method comprises the following steps of: performing point product operation on a feature mask block and a microscopic feature to generate an image block feature, aggregating the image block feature by adopting a Perceeiver-Slide aggregator to generate a slice-level global feature, inputting a text sequence into a BERT-Text encoder, outputting a text semantic feature, and performing point product operation on the microscopic feature, the slice-level global feature and the text semantic feature through a projection layer to generate an image block feature. And after uniformly mapping to a semantic space with the same dimension, inputting into a CoCa trainer for training, and generating a pathological report. According to the method, deep semantic fusion of the pathological image and the clinical text can be realized, and the diagnosis efficiency and accuracy are improved.

This invention discloses a method and system for intelligent coding review and automated data entry of medical records, relating to the field of data processing technology. The method acquires structured data from the electronic medical record system, including the patient's discharge diagnosis, treatment process, pathology report, medication records, and cost information, through multi-source acquisition. The data is then classified and preprocessed according to medical record type. A pre-trained natural language processing model analyzes the text data in the medical record homepage to understand the doctor's diagnostic descriptions and treatment process. Combined with a constructed coding rule base and knowledge graph, a recommended coding list and anomaly alerts are output through rule mapping and reasoning. A human-machine collaboration mechanism is established to automatically enter the reviewed medical record coding data into the medical record management system. This solves the problems of excessive reliance on manual operation, insufficient coding accuracy, and low efficiency in existing medical record coding review methods, forming a precise medical record coding working mode based on human-machine collaboration.

The invention discloses a rectal cancer CT lymph node pixelated marking method based on a pathological report, and relates to the technical field of crossing of medical image processing and artificial intelligence, and the method comprises the following steps: carrying out image standardization preprocessing operation on an abdomen enhanced CT sequence of a rectal cancer patient, including space standardization adjustment and gray normalization processing; performing structured analysis processing on the postoperative pathological report, and extracting lymph node metastasis state information, anatomical position coordinate data and morphological characteristic parameters; implementing pixel-level accurate positioning marking by adopting a three-plane collaborative verification mechanism, and synchronously performing target area identification and boundary refinement on a sagittal plane, a coronal plane and a horizontal plane; according to the rectal cancer CT lymph node pixelated marking method based on the pathological report, the postoperative pathological report is analyzed in a structured mode, and gold standard information such as the lymph node metastasis state, the precise anatomical position, the short diameter numerical value and the morphological characteristics is directly extracted and used as initialization parameters for image marking.

The invention discloses a pathological report generation method and system based on a multi-instance generation model and a medium, and the method comprises the steps: obtaining a pathological full-field graph, inputting the pathological full-field graph into a preset multi-instance generation model for data processing, and carrying out the data slicing and feature extraction of the pathological full-field graph, so as to obtain a visual feature sequence; performing feature extraction and enhancement based on the visual feature sequence to obtain a target feature sequence; and based on the target feature sequence, performing interaction processing of a visual mode and a text mode to obtain a pathology report corresponding to the pathology full-field graph. According to the method, the pathological features can be automatically extracted from the full-field graph, the detailed pathological report is generated, the workload of pathologists is reduced, the medical diagnosis information contained in the generated report is close to the real result, and the method has the characteristics of high accuracy, high universality, high automation degree and the like.