146 results about "Radiomics" patented technology

Systems and methods for multi-modal, multi-resolution deep learning neural networks for segmentation, outcomes prediction and longitudinal response monitoring to immunotherapy and radiotherapy are detailed herein. A structure-specific Generational Adversarial Network (SSGAN) is used to synthesize realistic and structure-preserving images not produced using state-of-the art GANs and simultaneously incorporate constraints to produce synthetic images. A deeply supervised, Multi-modality, Multi-Resolution Residual Networks (DeepMMRRN) for tumor and organs-at-risk (OAR) segmentation may be used for tumor and OAR segmentation. The DeepMMRRN may combine multiple modalities for tumor and OAR segmentation. Accurate segmentation is may be realized by maximizing network capacity by simultaneously using features at multiple scales and resolutions and feature selection through deep supervision. DeepMMRRN Radiomics may be used for predicting and longitudinal monitoring response to immunotherapy. Auto-segmentations may be combined with radiomics analysis for predicting response prior to treatment initiation. Quantification of entire tumor burden may be used for automatic response assessment.

The invention relates to a radiotherapy dosage verification method and system. The method includes steps: S1, receiving a CT image obtained before radiotherapy, and sketching a target area in the CT image; S2, extracting a plurality of radiomics characteristics of the target area; S3, screening out the radiomics characteristics in remarkable relevance to TRR (tumor regression ratio) from the multiple extracted radiomics characteristics; S4, presetting a dosage value, recognizing the CT image processed in the step S3 by a pre-trained radiomics model, and outputting to obtain the TRR under the given dosage value. A novel idea is provided, the radiomics model is obtained by training based on a machine learning method, the given dosage value is verified by virtue of the TRR outputted by the radiomics model, a prediction result is insusceptible to single dosage, and high reliability of a verification result is realized.

The invention relates to an auxiliary assessment method for prognosis of nasopharynx cancer based on enhanced MRI radiomics. The auxiliary assessment method comprises the steps of: (1), performing MRIimage processing; (2), extracting imaging features; (3), screening the imaging features; (4), establishing a radiomics scoring formula; (5), screening clinical risk factors; and (6), establishing a prognostic survival model: establishing a prognostic observation model through combination of a radiomics score and the clinical risk factors of a patient with nasopharynx cancer, performing qualitative and quantitative prediction on the PFS (progression free survival) of the patient, and furthermore, assessing the performance of the prognostic survival model. The auxiliary assessment method in theinvention has little harm to the image examination of the patient; qualitative and quantitative analysis on the survival time of a specific patient is carried out; therefore, a doctor is assisted tomake an individualized treatment and follow-up visit scheme; furthermore, the doctor is assisted to assess the survival and recurrence time of the patient; simultaneously, the performance of the obtained prognostic survival model is verified; and thus, the accuracy of a prognostic prediction model is ensured.

The invention provides an NSCLC patient postoperative short-period recurrence metastasis risk evaluating system which comprises a preprocessing module, a characteristic extracting module, an AI network cluster risk evaluating module and an ANFIS network integrated risk evaluating module. The preprocessing module is used for collecting and preprocessing multidimensional patient clinical data. The multidimensional patient clinical data comprise patient preoperative CT radiomics data, tumor histopathology data and tumor clinical. The characteristic extracting module is used for extracting the characteristic of the data sample after preprocessing. The AI network cluster risk evaluating module is used for performing recurrence risk probability evaluation on the extracted sample characteristic.The ANFIS network integrated risk evaluating module is used for using a recurrence risk probability evaluation result as an input characteristic vector, establishes an ANFIS network, uses a recurrence/metastasis risk quantification probability and dangerous degree as the output, and performs final integrated risk evaluation. The system can supply a reference for selecting a radial curing postoperative comprehensive treatment plan by a lung cancer patient and has high predicting accuracy.

The invention discloses a method for quantifying a tumor immune state based on radiomics, wherein the method comprises the following steps: S1, collecting clinical information, immunohistochemical characteristics, overall survival OS and disease-free survival DFS of tumor patients; S2, obtaining a CT image sample group of solid tumors; S3, randomly dividing the patients into a training set and a test set in proportion; S4, calculating the immunoscore by an immune marker, and dividing the patients into a set with high immune state and a set with low immune state; S5, segmenting an interest region, and extracting radiomics characteristics of the solid tumors in the interest region; S6, screening the radiomics characteristics related to the immune state from the radiomics characteristics in the training set, and training a prediction model; and S7, predicting the image immunoscore of the tumor patients, obtaining the image immune state, and then predicting the DFS and the OS in the patients. The immune state of the solid tumors such as colorectal cancer, lung cancer, gastric cancer and breast cancer is predicted by the CT image characteristics, and the analysis of pathological sections and immunohistochemistry is reduced or avoided.

The invention provides an analysis method and device for multi-mode radiomics, and a terminal. The method comprises steps of obtaining multiple mode images, and preprocessing the multiple mode images; performing regional segmentation to the preprocessed mode images to obtain an area of interest corresponding to each mode image; for each area of interest of each mode image, performing high-pass feature extraction, and obtaining characteristics corresponding to each area of interest; grouping the characteristics corresponding to each area of interest of the multiple mode images into source characteristics, and using a preset clustering algorithm to perform characteristic clustering of the source characteristics; and according to the result of the characteristic clustering, constructing radiomics markers. Therefore, the problem that image characteristics cannot be extracted from multiple aspects in the radiomics research in the prior art is solved, the number of the source characteristics for the characteristic clustering and the representative characteristic kind after clustering are enriched, and the medical image information can be excavated to the greatest extent.

The invention belongs to the field of predicting model and nomograph construction technology, and discloses a constructing method of a predicting model and the nomograph for HCC recurrence and RFS and application thereof. The method comprises the steps of by means a minimum redundancy maximum relevance algorithm (MRMRA), respectively extracting 647 radiomics characteristics from liver 3-period enhanced CT scanning images of all patients; confirming a radiomics label by means of a least absolute shrinkage and selection operator (LASSO)-Cox regression model; and establishing a predicting model and a predicting naomograph for HCC recurrence after therapeutic ablation treatment and teh RFS by means of clinical and pathological factors and the radiomics label. According to the method of the invention, the predicting model and the predicting nomograph are constructed through the radiomics method. The method is used for predicting tumor recurrence and RFS of an HCC patient after therapeutic ablation treatment of a focus.

The invention provides a multi-center brain tumor prognosis lifetime prediction method and system based on federated learning, and brain tumor prognosis lifetime prediction is carried out by using multi-scale information such as a multi-center multi-mode brain tumor image and omics information thereof, clinical medical record information of a patient and the like. The invention provides a multi-center federated learning mechanism based on active learning and reinforcement learning. According to the invention, a comprehensive brain tumor prognosis lifetime classification model is established by combining the patient electronic medical record information stored in each center in a distributed manner with the radiomics features and the deep learning features, and a reliable brain tumor prognosis lifetime prediction system with higher accuracy is realized on the basis of ensuring patient image data privacy.

The invention discloses a non-invasive liver epithelioid vascular smooth muscle lipoma image classification device based on radiomics, and belongs to the technical field of medical image processing. The non-invasive liver epithelioid vascular smooth muscle lipoma image classification device includes: a sampling module which for collecting CT/MRI images of liver epithelioid vascular smooth muscle lipoma, liver cancer and liver focal nodule hyperplasia meeting requirements; a focus area extraction module which is used for extracting a focus area; a feature extraction module which is used for carrying out radiomics feature extraction on the focus area; a feature screening module which is used for screening radiomics features; a random forest network training module which is used for training a random forest model to obtain a radiomics label; a clinical index fusion module which is used for fusing the radiomics prediction label and the clinical indexes of the patient and training a multivariate logistic regression model; and a classification module which is used for obtaining a final prediction label in combination with a random forest network and a multivariate logistic regression model so as to realize the classification of the liver epithelioid vascular smooth muscle lipoma images. The device is high in recognition precision, high in recognition speed, safe and stable.

The invention provides a radiomics-based hepatic venous pressure gradient (rHVPG) calculation model construction method, on this basis, a more advantageous hepatic venous pressure gradient calculation model can be constructed, and a new way is provided for the calculation of the early stage non-invasive index of a patient with portal hypertension. The invention also provides a radiomics-based hepatic venous pressure gradient calculation system. The invention overcomes the many limitations of current non-invasive measurement methods. The rHVPG calculation system proposed by the invention is feasible, is expected to provide a new idea for noninvasive measurement of portal pressure, and plays a positive role in improving the quality of life of the patient with portal hypertension and alleviating the disease burden of families and society.

The invention relates to a kidney lump classification method based on random projection. The kidney lump classification method comprises the steps: acquiring N pieces of target object data describingkidney lumps; carrying out target area sketching on each CT flat scanning image according to each mask image to obtain a region of interest of each CT flat scanning image, and carrying out radiomics characteristic data extraction on each region of interest to obtain N pieces of radiomics characteristic data; projecting the N radiomics characteristic data through L random projection matrixes to obtain L sets of projection characteristic data; performing multiple classifier training on the L sets of projection characteristic data to obtain a prediction matrix of each classifier and each trainedclassifier, and determining the weight of each classifier; and performing fusion processing on the to-be-classified data by adopting the trained classifiers according to the corresponding weights so as to determine corresponding categories. According to the method, the robustness of the to-be-classified data in the category identification process can be improved, so that the reliability of the classification result is improved.

The invention is a method of tumorology, imaging, and computer -assisted medicine, involving a method of building a model of malignant potential classification models based on radiological gastric and intestinal tumor.The construction method described in the present invention includes: data collection: abdominal enhancement phase of thin -layer CT image collection; extracting radiation group characteristics; statistical analysis; feature selection and radiation group characteristics model construction; radiology feature model verification and calibration.The method built by the method described by the present invention can classify the vicious potential classification model of gastrointestinal gastrointestinal tumor tumor tumor, which can accurately classify the GIST with different malignant potential, which is a non -invasive technology without increasing the additional cost of patients. Operations operation. Operations.Simple, convenient for clinical promotion.

An all-digital mammary gland imaging image radiomics method based on deep learning includes the following specific steps: obtaining all-digital mammary gland imaging image data [mu]<datset>, preprocessing the all-digital mammary gland imaging image data [mu]<datset> to obtain preprocessed data [mu]<hdf5>, inputting the preprocessed data [mu]<hdf5> to a deep learning network Alexnet to perform training, establishing a classification network model M<alexnet>, inputting the preprocessed data [mu]<hdf5> to the classification network model M<alexnet> to perform feature extraction, obtaining a high-dimensional feature vector FeatureMap, inputting the high-dimensional feature vector FeatureMap to a random forest RF to perform training to obtain a high-performance tumor identification classifier. The method provided by the invention uses a deep learning network Alexnet framework to extract image features for identifying a tumor, and combined with radiomics, the random forest is adopted to learn the extracted features, thereby realizing research of all-digital mammary gland imaging image radiomics based on deep learning.