CN112927187A - Method for automatically identifying and positioning focal cortical dysplasia epileptic focus - Google Patents
Method for automatically identifying and positioning focal cortical dysplasia epileptic focus Download PDFInfo
- Publication number
- CN112927187A CN112927187A CN202110113427.2A CN202110113427A CN112927187A CN 112927187 A CN112927187 A CN 112927187A CN 202110113427 A CN202110113427 A CN 202110113427A CN 112927187 A CN112927187 A CN 112927187A
- Authority
- CN
- China
- Prior art keywords
- epileptic
- data
- fcd
- focus
- modal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10088—Magnetic resonance imaging [MRI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10104—Positron emission tomography [PET]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20092—Interactive image processing based on input by user
- G06T2207/20104—Interactive definition of region of interest [ROI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention discloses a method for automatically identifying and positioning focal cortical dysplasia epileptic focus, which comprises the following steps: acquiring multi-modal neuroimaging data including MRI and PET; forming a large number of labeled data sets through manual labeling; deeply excavating a multi-mode biomarker of FCD (brain-responsive disease) seizure lesion from the angles of brain structure, brain metabolism and the like; establishing and training a CNN focus recognition segmentation model based on multi-modal data original features for automatic positioning diagnosis of FCD; aiming at common pathological FCD of epileptic surgery, magnetic resonance of a part of cases is negative and cannot be distinguished through vision, intelligent identification is carried out on an epileptic focus aiming at the FCD, so that the experience dependence of subjective film reading is greatly reduced, the time cost and the labor cost are reduced, the diagnosis and treatment efficiency of the epileptic focus is finally improved, and the operation prognosis is improved.
Description
Technical Field
The invention relates to the technical field of electroencephalogram detection, in particular to a method for automatically identifying and positioning focal cortical dysplasia epileptic lesions.
Background
Epilepsy is a recurrent chronic disease of the nervous system that seriously threatens human health. WHO statistics show that the prevalence rate of epilepsy is between 5 per thousand and 11.2 per thousand, about 5000 ten thousand patients with epilepsy exist all over the world, about 900 ten thousand patients with epilepsy in China, including 600 ten thousand patients with active epilepsy, and about 40 ten thousand new cases are newly generated each year. About 20-30% of all epileptic patients are drug refractory epilepsy and need preoperative assessment and operative treatment.
From a pathological point of view, cortical dysplasia (MCD) is the most common pathological basis for drug-refractory focal epilepsy. MCD is pathologically represented by migration, layering and structural abnormality of cerebral cortical neurons, and subtypes of the MCD mainly comprise Focal Cortical Dysplasia (FCD), nodular sclerosis, megacephalia, anencephalia, multiple cerebellar gyrus, subcortical gray ectopy, paraventricular nodular gray ectopy, cerebral fissure malformation and the like, wherein FCD is the most common subtype of MCD, and in the process of FCD positioning diagnosis, neuroimaging technology, particularly Magnetic Resonance Imaging (MRI), plays an important role in positioning diagnosis, however, although the existing MRI technology is rapidly promoted, a considerable part of FCD lesions are represented as MRI negative. MRI negative epilepsy means that epileptic lesions can not be seen on MRI of epileptic patients, and compared with MRI positive epilepsy, MRI negative epilepsy has hidden focus, difficult positioning and easy missed diagnosis. The direct consequences of positioning difficulties include low operative rate, poor operative treatment, etc. statistics show that only 15% of MRI negative epileptic patients eventually receive operative treatment, and this proportion reaches 73% in MRI positive cases, even if operative treatment is received, the postoperative remission rate of MRI positive patients reaches 76% -90%, and MRI negative patients only reaches about 40%. Therefore, the focus positioning problem of FCD epilepsy is one of the most critical and troublesome problems in epilepsy surgery, and is also the key and difficult point for further improving the level of epilepsy diagnosis and treatment.
Therefore, accurate identification of FCD epileptogenic focus and improvement of diagnosis rate are one of the most important challenges of current epileptic surgery, and the current clinical main diagnosis method depends on artificial visual interpretation and has the defects of low result accuracy, strong experience dependence and the like.
The early occurrence of epilepsy caused by computer-aided recognition is a neural image post-processing technology and a multi-modal fusion technology, and aims to provide a reference for clinical decision by computer-aided and quantitative feature evaluation. Voxel-based morphological analysis (VBM) and surface-based morphological analysis (SBM) quantitatively measure and analyze gray matter, white matter signals, metabolic intensity, special morphological characteristics and the like of neural images by means of voxels and reconstructed surface morphology, and based on the measurement and analysis, characteristic changes of FCD such as grey-white mass junction confusion, grey matter structure and metabolic abnormalities and the like are highlighted by comparing quantitative characteristics of individual patients with normal images, so that the detection rate of FCD seizure-induced seizure lesions is improved to a certain extent.
Although the detection rate of the FCD is improved to a certain extent by the existing multi-modal image post-processing method, the existing multi-modal image post-processing method still faces the disadvantages of single model, insufficient intelligence, strong artificial dependency and the like, and the continuous development of the artificial intelligence algorithm in recent years brings the eosin for further improving the FCD positioning level.
In the past, the value of the multi-modal image post-processing method on FCD positioning diagnosis is verified by evaluating the multi-modal image post-processing method. Including (including MAP techniques, PET/MRI registration fusion techniques, and SPM-PET techniques, among others.
In the machine learning algorithm, an Artificial Neural Network (ANN) is an operation model that simulates the structure and function of a biological central nervous system, and can cope with complex nonlinear classification tasks and the like. On the basis of the ANN, a deep learning algorithm based on a Convolutional Neural Network (CNN) realizes an end-to-end model from an original image to a label, and an efficient image recognition and segmentation method system has been developed at present. The machine learning technology is combined with clinical data, so that the data value can be deeply mined, and objective basis is provided for FCD positioning diagnosis. The characteristics of MRI image gray values, morphology and the like of FCD patients are extracted in the past, a classifier is constructed for lesion detection, and the sensitivity is about 70% as shown by a preliminary small sample research result.
At present, the artificial intelligence method is widely applied to image segmentation or classification, but unlike intracranial tumors, blood vessels and other diseases, FCD is a common epileptogenic focus in the epileptic surgery, neuroimaging performance of about 30% of patients is not obvious, even negative, and clinical missed diagnosis is easily caused, and the automatic identification accuracy of FCD epileptogenic focus by machine learning in the current thesis is generally low, including the past work of an applicant, and the accuracy is about 68-73%.
At present, for research work aiming at FCD automatic identification, neural image post-processing software such as SPM or FreeSchfer is generally applied to carry out quantification and pretreatment, and then an artificial intelligence model is added to carry out classification or prediction, but the time consumption is long, the time required by model research and development or maintenance is excluded according to the work before, although codes are already sourced, software/system/environment preparation, image pretreatment (FreeSchfer v 5.3), data quality control, data analysis and the like are required during rerun, the semi-automatic processing method takes about 10 hours per example, the steps are multiple, the source tracing is difficult when errors occur, and the clinical use requirement is far from being reached.
FCD code is automatically identified by publishing it on Github. It can be seen that the processing flow is semi-automatic and involves processing conversion of different languages, and the steps are numerous, and although the steps are spanned from pure artificial visual recognition to machine-assisted recognition, the defects of insufficient practicability still exist, and the steps are the main bottleneck of applying machine learning to epileptic surgery at present.
The accuracy rate of the previous series of research reports by combining machine learning with a neural image post-processing technology is 70%, and the accuracy rate of the model constructed by the model is further improved. In addition, the existing methodology relies on the existing clinical experience to extract various empirical image features, so that the machine learning accuracy is limited by human visual analysis and experience summary, and the breakthrough is not achieved. The deep learning method is used for fully utilizing the advantages of machine image recognition, so that the seizure focus recognition rate is broken through, the accuracy reaches 83.3%, the specificity is 100%, and the method has the prospect of assisting a clinician in making a decision.
The former work is used as semi-automatic processing, firstly, the software is long in time consumption for post-processing, secondly, the steps are multiple, manual monitoring, operation and quality control are needed, and the possibility of error accumulation and failure in tracing is caused.
Disclosure of Invention
The invention aims to provide a method for automatically identifying and positioning focal cortical dysplasia epileptic focus so as to solve the problems in the background technology.
In order to achieve the purpose, the invention provides the following technical scheme:
a method of automatically identifying localized focal corticogenesis epileptic foci, the method comprising:
acquiring multi-modal neuroimaging data including MRI and PET;
forming a large number of labeled data sets through manual labeling;
deeply excavating a multi-mode biomarker of FCD (brain-responsive disease) seizure lesion from the angles of brain structure, brain metabolism and the like;
establishing and training a CNN focus recognition segmentation model based on multi-modal data original features for automatic positioning diagnosis of FCD;
training a convolutional neural network by using a large sample with a label, visualizing the test effect of a verification set in the training process, and adjusting the hyper-parameters of the model so as to obtain a more generalized model;
after the model training is finished, inputting an image containing a target to be detected, and outputting a rectangular bounding box containing the target and a pixel-level target segmentation label.
As a further scheme of the invention: the multi-modal neuroimaging data is acquired FCD epileptic focus confirmed by previous pathology.
As a still further scheme of the invention: the multi-modal biomarkers provide data support for training artificial intelligence models.
As a still further scheme of the invention: the CNN focus identification segmentation model combines two-dimensional single-layer images at corresponding positions of the registered multi-modal neuroimage to serve as a multi-channel input image of the feature extraction network, so that complementary fusion of multi-modal data is realized.
As a still further scheme of the invention: and performing data amplification (data augmentation) on the multi-modal neuroimaging data, namely acquiring a larger number of training data sets to improve the training effect through processing such as rotation, translation, scaling and the like based on the collected image data, and also improving the training effect by combining a transfer learning mode.
As a still further scheme of the invention: the object segmentation label includes:
generating candidate regions using the region suggestion network, adapting the network structure to the multimodal data input;
further obtaining a feature map of each ROI area in the image, and performing pixel correction on each ROI by using ROI alignment (ROI alignment);
and predicting the category of each pixel point in the ROI area by using a designed FCN frame for each ROI, and finally obtaining a focus segmentation result.
As a still further scheme of the invention: the area suggestion network selects a ResNet-101 network as a multi-mode image feature extraction network.
As a still further scheme of the invention: and after obtaining the feature map of each ROI area, predicting each ROI, and judging whether the ROI belongs to the seizure lesion and a corresponding bounding box (bounding box).
Compared with the prior art, the invention has the beneficial effects that:
1. aiming at common pathological FCD of epileptic surgery, magnetic resonance of a part of cases is negative and cannot be distinguished through vision, intelligent identification is carried out on an epileptic focus aiming at the FCD, so that the experience dependence of subjective film reading is greatly reduced, the time cost and the labor cost are reduced, the diagnosis and treatment efficiency of the epileptic focus is finally improved, and the operation prognosis is improved.
2. According to the invention, by using a more quantitative and intelligent means, a deep learning model and multi-modal imagery data characteristics, the diagnosis accuracy of the FCD epileptogenic focus can be improved, the diagnosis process is simplified, and finally an epileptic patient benefits from the diagnosis process, unnecessary repeated image examination is reduced, and good operation benefit is obtained.
3. The three-dimensional neural network is trained through multi-channel model training, namely the metabolic image PET mode is important neuroimaging data in the epilepsy surgical diagnosis and treatment process, and structural magnetic resonance (T1-MPRAGE, T2-FLAIR sequence) and metabolic images are combined.
4. By constructing an automatic platform, the intelligent identification can be carried out by directly putting the patient in a corresponding mode in the using method, so that the learning curve of a user is reduced, and the initial use difficulty of a clinician is reduced.
5. The automatic platform is constructed to achieve the end-to-end purpose, original data are placed in a target folder to achieve the result directly, the learning curve used by a clinician is reduced, and the automatic platform has landing value.
6. The invention gives full play to the potential of the artificial intelligence technology in the medical field by fully utilizing clinical multi-modal neuroimaging data, and the expected result not only can provide objective quantitative basis for clinical decision and optimize the current medical resources, but also can improve the prognosis and the life quality of patients, accords with the general target of the development of the national medical health industry, and has good clinical application prospect.
Drawings
Fig. 1 is a schematic diagram of automatic location of an epileptogenic focus by combining multi-modal data with a deep learning model.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, in an embodiment of the present invention, a method for automatically identifying and locating an epileptic focus of focal cortical dysplasia includes:
acquiring multi-modal neuroimaging data including MRI and PET; forming a large number of labeled data sets through manual labeling; deeply excavating a multi-mode biomarker of FCD (brain-responsive disease) seizure lesion from the angles of brain structure, brain metabolism and the like; establishing and training a CNN focus recognition segmentation model based on multi-modal data original features for automatic positioning diagnosis of FCD; training a convolutional neural network by using a large sample with a label, visualizing the test effect of a verification set in the training process, and adjusting the hyper-parameters of the model so as to obtain a more generalized model; after the model training is finished, inputting an image containing a target to be detected, and outputting a rectangular bounding box containing the target and a pixel-level target segmentation label.
The multi-modal neuroimaging data is acquired FCD epileptic focus confirmed by previous pathology.
The multi-modal biomarkers provide data support for training artificial intelligence models.
The CNN focus identification segmentation model combines two-dimensional single-layer images at corresponding positions of the registered multi-modal neuroimage to serve as a multi-channel input image of the feature extraction network, so that complementary fusion of multi-modal data is realized.
And performing data amplification (data augmentation) on the multi-modal neuroimaging data, namely acquiring a larger number of training data sets to improve the training effect through processing such as rotation, translation, scaling and the like based on the collected image data, and also improving the training effect by combining a transfer learning mode.
The object segmentation label includes:
generating candidate regions using the region suggestion network, adapting the network structure to the multimodal data input; further obtaining a feature map of each ROI area in the image, and performing pixel correction on each ROI by using ROI alignment (ROI alignment); and predicting the category of each pixel point in the ROI area by using a designed FCN frame for each ROI, and finally obtaining a focus segmentation result.
The area suggestion network selects a ResNet-101 network as a multi-mode image feature extraction network.
And after obtaining the feature map of each ROI area, predicting each ROI, and judging whether the ROI belongs to the seizure lesion and a corresponding bounding box (bounding box).
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be considered to be within the technical scope of the present invention, and the technical solutions and the inventive concepts thereof according to the present invention should be equivalent or changed within the scope of the present invention.
Claims (8)
1. A method for automatically identifying and locating focal cortical dysplasia epileptic foci, the method comprising:
acquiring multi-modal neuroimaging data including MRI and PET;
forming a large number of labeled data sets through manual labeling;
deeply excavating a multi-mode biomarker of FCD (brain-responsive disease) seizure lesion from the angles of brain structure, brain metabolism and the like;
establishing and training a CNN focus recognition segmentation model based on multi-modal data original features for automatic positioning diagnosis of FCD;
training a convolutional neural network by using a large sample with a label, visualizing the test effect of a verification set in the training process, and adjusting the hyper-parameters of the model so as to obtain a more generalized model;
after the model training is finished, inputting an image containing a target to be detected, and outputting a rectangular bounding box containing the target and a pixel-level target segmentation label.
2. The method for automatically identifying and locating focal cortical dysplasia epileptic foci according to claim 1, wherein said multi-modal neuroimaging data is an acquisition of a past pathologically confirmed FCD epileptic focus.
3. The method for automatically identifying and locating focal cortical dysplasia epileptic lesions as claimed in claim 1, wherein said multi-modal biomarker provides data support for training artificial intelligence models.
4. The method for automatically identifying and locating the focal cortical dysplasia epileptic focus according to claim 1, wherein the CNN focus identification segmentation model combines two-dimensional single-layer images at corresponding positions of the registered multimodal neuroimaging as a multi-channel input image of the feature extraction network, so as to realize the complementary fusion of multimodal data.
5. The method for automatically identifying and locating the focal cortical dysplasia epileptic focus according to claim 1, wherein the multi-modal neuroimaging data is subjected to data amplification, that is, based on the collected image data, a larger number of training data sets are obtained through processing such as rotation, translation, scaling and the like, so as to improve the training effect, and the training effect can also be improved by combining a migration learning mode.
6. The method of claim 1, wherein the target segmentation label comprises:
generating candidate regions using the region suggestion network, adapting the network structure to the multimodal data input;
further obtaining a feature map of each ROI area in the image, and performing pixel correction on each ROI by using ROI alignment;
and predicting the category of each pixel point in the ROI area by using a designed FCN frame for each ROI, and finally obtaining a focus segmentation result.
7. The method for automatically identifying and locating focal cortical dysplasia epileptic lesions according to claim 6, wherein the regional suggestion network selects the ResNet-101 network as a multi-modal image feature extraction network.
8. The method for automatically identifying and locating focal cortical dysplasia epileptic foci according to claim 6, wherein after obtaining the feature map of each ROI area, each ROI is predicted to determine whether it belongs to epileptogenic lesion and corresponding bounding box.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110113427.2A CN112927187A (en) | 2021-01-27 | 2021-01-27 | Method for automatically identifying and positioning focal cortical dysplasia epileptic focus |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110113427.2A CN112927187A (en) | 2021-01-27 | 2021-01-27 | Method for automatically identifying and positioning focal cortical dysplasia epileptic focus |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112927187A true CN112927187A (en) | 2021-06-08 |
Family
ID=76167279
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110113427.2A Withdrawn CN112927187A (en) | 2021-01-27 | 2021-01-27 | Method for automatically identifying and positioning focal cortical dysplasia epileptic focus |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112927187A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113284126A (en) * | 2021-06-10 | 2021-08-20 | 安徽省立医院(中国科学技术大学附属第一医院) | Method for predicting hydrocephalus shunt operation curative effect by artificial neural network image analysis |
| CN113902751A (en) * | 2021-11-10 | 2022-01-07 | 南京大学 | Intestinal neuron dysplasia identification method based on Swin-Unet algorithm |
| CN115005802A (en) * | 2022-07-21 | 2022-09-06 | 首都医科大学宣武医院 | Method, system and device for positioning onset part of brain network disease |
| WO2023168728A1 (en) * | 2022-03-10 | 2023-09-14 | 中国科学院深圳先进技术研究院 | Multimodal radiomics-based epilepsy drug treatment outcome prediction method and apparatus |
| CN117152128A (en) * | 2023-10-27 | 2023-12-01 | 首都医科大学附属北京天坛医院 | Method and device for recognizing focus of nerve image, electronic equipment and storage medium |
| TWI838742B (en) * | 2022-05-05 | 2024-04-11 | 臺北醫學大學 | Brain edema image processing system and its operation method |
| WO2024098449A1 (en) * | 2022-11-11 | 2024-05-16 | 中国科学院深圳先进技术研究院 | Epileptic focus zone positioning system and method based on deep learning and electrophysiological signal |
-
2021
- 2021-01-27 CN CN202110113427.2A patent/CN112927187A/en not_active Withdrawn
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113284126A (en) * | 2021-06-10 | 2021-08-20 | 安徽省立医院(中国科学技术大学附属第一医院) | Method for predicting hydrocephalus shunt operation curative effect by artificial neural network image analysis |
| CN113284126B (en) * | 2021-06-10 | 2022-06-24 | 安徽省立医院(中国科学技术大学附属第一医院) | Method for predicting hydrocephalus shunt operation curative effect by artificial neural network image analysis |
| CN113902751A (en) * | 2021-11-10 | 2022-01-07 | 南京大学 | Intestinal neuron dysplasia identification method based on Swin-Unet algorithm |
| WO2023168728A1 (en) * | 2022-03-10 | 2023-09-14 | 中国科学院深圳先进技术研究院 | Multimodal radiomics-based epilepsy drug treatment outcome prediction method and apparatus |
| TWI838742B (en) * | 2022-05-05 | 2024-04-11 | 臺北醫學大學 | Brain edema image processing system and its operation method |
| CN115005802A (en) * | 2022-07-21 | 2022-09-06 | 首都医科大学宣武医院 | Method, system and device for positioning onset part of brain network disease |
| WO2024098449A1 (en) * | 2022-11-11 | 2024-05-16 | 中国科学院深圳先进技术研究院 | Epileptic focus zone positioning system and method based on deep learning and electrophysiological signal |
| CN117152128A (en) * | 2023-10-27 | 2023-12-01 | 首都医科大学附属北京天坛医院 | Method and device for recognizing focus of nerve image, electronic equipment and storage medium |
| CN117152128B (en) * | 2023-10-27 | 2024-02-27 | 首都医科大学附属北京天坛医院 | Method and device for recognizing focus of nerve image, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112927187A (en) | Method for automatically identifying and positioning focal cortical dysplasia epileptic focus | |
| CN108464840B (en) | Automatic detection method and system for breast lumps | |
| CN111028206A (en) | Prostate cancer automatic detection and classification system based on deep learning | |
| AU2011344876B2 (en) | Method for delineation of tissue lesions | |
| WO2022256063A9 (en) | Predictive prognosis based on multimodal analysis | |
| CN111179227B (en) | Mammary gland ultrasonic image quality evaluation method based on auxiliary diagnosis and subjective aesthetics | |
| CN104414636B (en) | Cerebral microbleeds computer-aided detection system based on MRI | |
| CN110991536B (en) | Training method of early warning model of primary liver cancer | |
| CA2752370A1 (en) | Segmentation of structures for state determination | |
| Dai et al. | A novel meibomian gland morphology analytic system based on a convolutional neural network | |
| CN111738997A (en) | Method for calculating new coronary pneumonia lesion area ratio based on deep learning | |
| CN110390678B (en) | Tissue type segmentation method of colorectal cancer IHC staining image | |
| Yang et al. | Diagnosis of Parkinson’s disease based on 3D ResNet: The frontal lobe is crucial | |
| CN118262875A (en) | Medical image diagnosis and contrast film reading method | |
| Ong et al. | Detection of subtle white matter lesions in MRI through texture feature extraction and boundary delineation using an embedded clustering strategy | |
| CN104463885A (en) | Partition method for multiple-sclerosis damage area | |
| CN114847922A (en) | Brain age prediction method based on automatic fiber bundle identification | |
| CN114332910A (en) | Human body part segmentation method for similar feature calculation of far infrared image | |
| CN112200810A (en) | Multi-modal automated ventricular segmentation system and method of use thereof | |
| CN112837807A (en) | Early intelligent high-precision auxiliary diagnosis method for T2DM brain aging cognitive impairment | |
| CN116645389A (en) | Individualized vascular thrombosis three-dimensional structure modeling method and system | |
| CN114445399A (en) | Animal pathological liver pathological image identification method based on deep learning | |
| Abd Hamid et al. | Incorporating attention mechanism in enhancing classification of alzheimer’s disease | |
| CN117934416B (en) | CTA internal carotid artery segmentation method and system based on machine learning | |
| CN118154975B (en) | Tumor pathological diagnosis image classification method based on big data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20221026 Address after: 119 South Fourth Ring Road West, Fengtai District, Beijing 100070 Applicant after: BEIJING TIANTAN HOSPITAL, CAPITAL MEDICAL University Address before: 119 South Fourth Ring Road West, Fengtai District, Beijing 100070 Applicant before: Zhang Kai Applicant before: Mo Jiajie |
|
| TA01 | Transfer of patent application right | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20210608 |
|
| WW01 | Invention patent application withdrawn after publication |