WO2023035459A1 - 一种磁共振图像处理方法、终端设备及计算机存储介质 - Google Patents

一种磁共振图像处理方法、终端设备及计算机存储介质 Download PDF

Info

Publication number
WO2023035459A1
WO2023035459A1 PCT/CN2021/137626 CN2021137626W WO2023035459A1 WO 2023035459 A1 WO2023035459 A1 WO 2023035459A1 CN 2021137626 W CN2021137626 W CN 2021137626W WO 2023035459 A1 WO2023035459 A1 WO 2023035459A1
Authority
WO
WIPO (PCT)
Prior art keywords
lesion area
magnetic resonance
resonance image
processing method
image processing
Prior art date
Application number
PCT/CN2021/137626
Other languages
English (en)
French (fr)
Inventor
张娜
郑海荣
刘新
胡战利
梁栋
李烨
周缘
Original Assignee
中国科学院深圳先进技术研究院
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 中国科学院深圳先进技术研究院 filed Critical 中国科学院深圳先进技术研究院
Publication of WO2023035459A1 publication Critical patent/WO2023035459A1/zh

Links

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
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/05—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves 
    • A61B5/055—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves  involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • 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]
    • G06T2207/10096—Dynamic contrast-enhanced magnetic resonance imaging [DCE-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
    • G06T2207/20—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
    • G06T2207/30096—Tumor; Lesion

Definitions

  • the present application relates to the field of magnetic resonance application technology, in particular to a magnetic resonance image processing method, terminal equipment and computer storage medium.
  • the differential diagnosis of benign and malignant lesions of the breast is generally performed by analyzing the morphological manifestations of the images, the signal intensity change curve and the internal structure of the lesions.
  • MRI Magnetic Resonance Imaging
  • the local time-signal intensity curve (TIC) is a qualitative method to analyze the hemodynamic characteristics, so as to assist in the graded diagnosis of breast cancer lesions.
  • TIC analysis is often based on the mean value of signal intensity changes in the region of interest (ROI) manually divided by doctors.
  • ROI region of interest
  • the morphological complexity and heterogeneity of the lesion, the size of the artificially divided region of interest and other internal and external factors will increase the time cost of the doctor's clinical diagnosis and lead to certain errors.
  • the present application provides a magnetic resonance image processing method, a terminal device and a computer storage medium.
  • the application provides a magnetic resonance image processing method, the magnetic resonance image processing method comprising:
  • the acquisition of time-series data of pixel signal values of each pixel in each lesion area in the magnetic resonance image includes:
  • a data set is established for each lesion area, and the data set includes a plurality of pixels and their pixel signal value time series data, wherein the pixel signal value time series data is a plurality of signal intensities of the pixels arranged according to the scanning acquisition signal time data.
  • the acquiring the distance matrix of the time-series data of pixel signal values between two pixels in each lesion area includes:
  • a distance matrix of time-series data of pixel signal values between two pixels in each lesion area is calculated by using a dynamic time warping technique.
  • the magnetic resonance image processing method also includes:
  • the magnetic resonance image processing method further includes:
  • multiple lesion areas are divided into first-level lesion areas, second-level lesion areas, and third-level lesion areas;
  • the average number of clustering centers of the third-level lesion area is greater than the average number of clustering centers of the second-level lesion area, and the average number of clustering centers of the second-level lesion area is greater than the average clustering center of the first-level lesion area quantity.
  • the outputting the pathological result of each lesion area based on the number of the cluster centers includes:
  • the number of optimal cluster centers of each lesion area is used as the characteristic information of the lesion area
  • the pathological results of the lesion area are output based on the proportion of the broken line type and the broken line analysis result.
  • the present application also provides a terminal device, and the terminal device includes:
  • a time-series data module configured to obtain time-series data of pixel signal values of each pixel in each lesion area in the magnetic resonance image
  • a distance matrix module configured to obtain a distance matrix of time-series data of pixel signal values between two pixels in each lesion area
  • a clustering center module configured to cluster the pixels in each lesion area based on the distance matrix, and obtain the number of clustering centers
  • a pathological result module configured to output the pathological result of each lesion area based on the number of cluster centers.
  • the present application also provides another terminal device, where the terminal device includes a memory and a processor, wherein the memory is coupled to the processor;
  • the memory is used for storing program data
  • the processor is used for executing the program data to realize the above-mentioned magnetic resonance image processing method.
  • the present application also provides a computer storage medium, the computer storage medium is used for storing program data, and when the program data is executed by a processor, it is used to realize the above magnetic resonance image processing method.
  • the terminal device acquires the time-series data of pixel signal values of each pixel in each lesion area in the magnetic resonance image; acquires the distance matrix of the time-series data of pixel signal values between two pixels in each lesion area; The pixels in each lesion area are clustered based on the distance matrix to obtain the number of cluster centers; and the pathological results of each lesion area are output based on the number of cluster centers.
  • the magnetic resonance image processing method of the present application directly analyzes the lesion area pixel by pixel, fully reveals the heterogeneity of the lesion, and improves the accuracy of pathological analysis.
  • Fig. 1 is a schematic flow chart of an embodiment of a magnetic resonance image processing method provided by the present application
  • Fig. 2 is a schematic flow chart of the pixel extraction method provided by the present application.
  • FIG. 3 is a schematic structural diagram of an embodiment of a terminal device provided by the present application.
  • FIG. 4 is a schematic structural diagram of another embodiment of a terminal device provided by the present application.
  • Fig. 5 is a schematic structural diagram of an embodiment of a computer storage medium provided by the present application.
  • signal intensity-time (TIC) curve analysis is a valuable measurement tool, which dynamically reflects the hemodynamic changes of the scanning site through the change of signal intensity before and after enhancement.
  • Semi-quantitative parameters are mainly obtained by describing the shape and structure of the tissue signal intensity-time curve in the region of interest, without the need to select a pharmacokinetic model that matches the tissue. Commonly used semi-quantitative parameters include initial area under the curve, time to peak, maximum signal intensity, maximum slope, washout rate, etc.
  • the horizontal axis is time, and the vertical axis is signal strength (which can be understood as pixel value).
  • the existing TIC analysis is generally based on the semi-quantitative analysis of the shape of the TIC curve.
  • the time-signal intensity curve TIC is therefore divided into three categories: ascending type (the intensity shows a slow and continuous increase, which is common in benign lesions); , malignancy is possible); and outflow type (dynamic early signal intensity reaches the highest peak and then decreases, the probability of malignant lesions is high).
  • TIC analysis generally follows the following steps. The doctor divides and selects a region of interest (ROI) of a certain MRI lesion, and then the software gives the average "signal intensity-time curve" of the region of interest, and observes the shape of the TIC dynamic enhancement curve, combined with morphology, etc. Methods and indicators for further diagnosis.
  • ROI region of interest
  • the current research on TIC generally analyzes the distribution of three types of TIC to distinguish benign and malignant tumors, without further classification of benign and malignant tumors. And it is often based on semi-quantitative parameters such as peak value, time to peak, and area under the broken line to define features and classify the TIC broken line.
  • FIG. 1 is a schematic flowchart of an embodiment of a magnetic resonance image processing method provided in the present application.
  • the magnetic resonance image processing method of the present application is applied to a terminal device, wherein the terminal device of the present application may be a server, or may be a system in which the server and electronic devices cooperate with each other.
  • the terminal device of the present application may be a server, or may be a system in which the server and electronic devices cooperate with each other.
  • various parts included in the terminal device such as various units, subunits, modules, and submodules, may all be set in the server, or may be set in the server and the terminal device separately.
  • the above server may be hardware or software.
  • the server When the server is hardware, it can be implemented as a distributed server cluster composed of multiple servers, or as a single server.
  • the server When the server is software, it can be implemented as multiple software or software modules, such as software or software modules used to provide a distributed server, or as a single software or software module, which is not specifically limited here.
  • the magnetic resonance image processing method in the embodiment of the present application may be implemented in a manner in which a processor invokes computer-readable instructions stored in a memory.
  • the magnetic resonance image processing method of the embodiment of the present application specifically includes the following steps:
  • Step S11 Obtain time-series data of pixel signal values of each pixel in each lesion area in the magnetic resonance image.
  • the terminal device acquires time-series data of pixel signal values of each pixel in each lesion area in the magnetic resonance image.
  • FIG. 2 is a schematic flowchart of a pixel extraction method provided in the present application.
  • a DCE-MRI breast scan is performed on the patient using a magnetic resonance instrument, and a contrast agent is injected into the scan result.
  • the terminal device After 2 or 3 minutes, the terminal device performs N acquisitions at equal time intervals, and each acquired acquisition set consists of m slices. For each slice, radiologists and professionals mark the lesion of each slice of the magnetic resonance image, thereby dividing multiple lesion regions.
  • the terminal device extracts the signal intensity data obtained by each pixel on each slice at the time point of signal acquisition in each scan of the magnetic resonance, and performs denoising.
  • the terminal device sorts the signal intensity data of the pixels according to the time of scanning and collecting signals, and obtains the time series data of the pixel signal values of all pixels in the lesion area. After obtaining the time-series data of pixel signal values of all pixels in the lesion area, each pixel in the lesion area is used as a research sample, and a data set is created for each lesion area.
  • each row in the data set corresponds to the time-series data of the pixel signal value of a single pixel
  • each column in the data set corresponds to the signal intensity data of all pixels measured during a single scanning acquisition signal time.
  • the terminal device creates an m*n data set, where the lesion area Li corresponding to the data set contains m pixels, and collects different signal intensity data during n scan signal acquisition times.
  • the terminal device generates a TIC line chart for each data set as shown in Figure 2, where each line in the TIC line chart corresponds to a pixel, the abscissa of the TIC line chart is the relative acquisition time, and the ordinate of the TIC line chart is is the relative signal strength.
  • Step S12 Obtain the distance matrix of the time-series data of pixel signal values between two pixels in each lesion area.
  • the lesion area Li is taken as an example.
  • the terminal device directly uses the time-series data of pixel signal values in the lesion area Li as training data.
  • the terminal device utilizes dynamic time warping technology (DTW, Dynamic Time Warping) is used as a distance metric to judge the similarity between the time series data of pixel signal values, and calculate the distance matrix. Assume that there are N pixels in the lesion area Li, corresponding to N broken lines in the TIC line chart. Finally, the terminal device calculates an N*N distance matrix, and each element dist(x, y) in the distance matrix represents the distance between the TIC polyline of pixel x and the TIC polyline of pixel y, that is, the similarity.
  • DTW Dynamic Time Warping
  • Euclidean distance may also be used to calculate the above similarity.
  • LCSS Longest-Common-Subsequence, longest common subsequence problem
  • Step S13 clustering the pixels in each lesion area based on the distance matrix, and obtaining the number of cluster centers.
  • fuzzy c-means clustering Fuzzy can also be used c-Means (FCM), improved discrete K-median clustering Modified Clustering models such as Discrete k-Median Clustering (DKM-S).
  • FCM c-Means
  • DKM-S Discrete k-Median Clustering
  • the terminal device can also pass the silhouette coefficient, Calinski and Harabasz score, Davis-Bouldin Index and other internal evaluation indicators of clustering models to evaluate the clustering effect.
  • the terminal device determines the optimal clustering iteration number k value from multiple clustering results according to the preset clustering model evaluation index, so that when the optimal k value is met, the clustering results have high inter-class dispersion And the intra-class cohesion is strong, that is, the shape of the cluster center of each cluster category is representative, and can be distinguished from the shape of the cluster center of other cluster categories.
  • the terminal device may obtain the optimal number Ki of cluster centers for each lesion area through the above process.
  • the larger the number Ki of optimal clustering centers in the lesion area the more complex the blood perfusion situation in the lesion area, the more trend shapes of signal intensity changes, and the higher the heterogeneity of the lesion area.
  • multiple focus areas can be divided into first-level focus areas, second-level focus areas, and third-level focus areas according to the degree of malignancy, wherein the average number of cluster centers of the third-level focus areas is greater than that of the second-level focus areas.
  • the average number of cluster centers, the average number of cluster centers of the secondary lesion area is greater than the average number of cluster centers of the first-level lesion area.
  • the terminal device compares the internal evaluation indicators of each lesion area and the corresponding optimal number of cluster centers, and selects the average number of cluster centers to reflect the differences between the first-level lesion area, the second-level lesion area, and the third-level lesion area
  • the optimal evaluation index of each lesion area can be obtained by using the best evaluation index, and the number of optimal clustering centers of the first-level lesion area, the second-level lesion area, and the third-level lesion area increases successively. , can be distinguished and have large differences.
  • the terminal device can perform the above process for selecting different clustering models, arrange and combine the clustering results of each clustering model, and select the average clustering center that can reflect the first-level lesion area, the second-level lesion area, and the third-level lesion area The best combination with the largest difference in quantity and increasing, where the best combination includes the type of clustering model and the internal evaluation index.
  • Step S14 output the pathological results of each lesion area based on the number of cluster centers.
  • the terminal device records the optimal number Ki of cluster centers of each lesion area as the new feature of the lesion area, that is, the broken line shape type that signs each lesion area will produce the optimal number Ki of cluster centers.
  • the terminal device performs statistical analysis on all broken line shapes, uses DBA (dynamic barycenter averaging) to obtain Ki types of cluster centers, and maps Ki broken line types to rising, platform, and outflow types, and counts the different proportions of the three basic types. According to the early enhancement rate, peak (ER max) and other information, the analysis results are also used as the characteristics of each lesion area.
  • DBA dynamic barycenter averaging
  • the terminal device establishes a data set for all lesion areas, and uses each lesion area as a training sample.
  • the terminal device acquires the time-series data of pixel signal values of each pixel in each lesion area in the magnetic resonance image; acquires the distance matrix of the time-series data of pixel signal values between two pixels in each lesion area; The pixels in each lesion area are clustered based on the distance matrix to obtain the number of cluster centers; and the pathological results of each lesion area are output based on the number of cluster centers.
  • the magnetic resonance image processing method of the present application directly analyzes the lesion area pixel by pixel, fully reveals the heterogeneity of the lesion, and improves the accuracy of pathological analysis.
  • the magnetic resonance image processing method of the embodiment of the present application can directly perform pixel-by-pixel dynamic time-signal intensity curve (TIC) analysis on the dynamic enhanced MR image lesion area, and realize the pathological grading of malignant breast tumors from the perspective of heterogeneity, and realize For the purpose of predicting lymph node metastasis, predicting the molecular type of cancer, predicting chemotherapy sensitivity, and predicting survival time; compared with traditional machine learning, the magnetic resonance image processing method of the embodiment of the application is innovatively applied in the classification of DCE-MRI signal value intensity broken line
  • the distance measure of DTW dynamic time warping is used to calculate the similarity and perform global measurement instead of Euclidean distance, Mahalanobis distance, Manhattan, etc.; DTW dynamic time warping method is proposed for sequence matching, especially when there is a certain drift in the sequence, Euclidean distance When the distance measure fails, DTW has a good effect on classifying time series data according to the shape.
  • the writing order of each step does not mean a strict execution order and constitutes any limitation on the implementation process.
  • the specific execution order of each step should be based on its function and possible
  • the inner logic is OK.
  • FIG. 3 is a schematic structural diagram of an embodiment of the terminal device provided in the present application.
  • the terminal device 300 includes a time series data module 31 , a distance matrix module 32 , a cluster center module 33 and a pathological result module 34 .
  • the time-series data module 31 is configured to obtain time-series data of pixel signal values of each pixel in each lesion area in the magnetic resonance image.
  • the distance matrix module 32 is configured to obtain a distance matrix of time-series data of pixel signal values between two pixels in each lesion area.
  • a clustering center module 33 configured to cluster the pixels in each lesion area based on the distance matrix, and obtain the number of clustering centers.
  • a pathological result module 34 configured to output the pathological result of each lesion area based on the number of cluster centers.
  • FIG. 4 is a schematic structural diagram of another embodiment of the terminal device provided in the present application.
  • the terminal device 400 in this embodiment of the present application includes a memory 41 and a processor 42, where the memory 41 and the processor 42 are coupled.
  • the memory 41 is used for storing program data
  • the processor 42 is used for executing the program data to realize the magnetic resonance image processing method described in the above-mentioned embodiments.
  • the processor 42 may also be referred to as a CPU (Central Processing Unit, central processing unit).
  • the processor 42 may be an integrated circuit chip with signal processing capability.
  • the processor 42 can also be a general-purpose processor, a digital signal processor (DSP, Digital Signal Process), an application-specific integrated circuit (ASIC, Application Specific Integrated Circuit), Field Programmable Gate Array (FPGA, Field Programmable Gate Array) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
  • DSP digital signal processor
  • ASIC Application Specific Integrated Circuit
  • FPGA Field Programmable Gate Array
  • the general purpose processor can be a microprocessor or the processor 42 can be any conventional processor or the like.
  • the present application also provides a computer storage medium.
  • the computer storage medium 500 is used to store program data 51.
  • the program data 51 is executed by the processor, it is used to realize the magnetic resonance image as described in the above-mentioned embodiments. Approach.
  • the present application also provides a computer program product, wherein the computer program product includes a computer program, and the computer program is operable to cause a computer to execute the magnetic resonance image processing method as described in the embodiment of the present application.
  • the computer program product may be a software installation package.
  • the magnetic resonance image processing methods described in the above-mentioned embodiments of the present application may be stored in a device, such as a computer-readable storage medium, when implemented in the form of a software function unit and sold or used as an independent product.
  • a device such as a computer-readable storage medium
  • the technical solution of the present application is essentially or part of the contribution to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium , including several instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor (processor) execute all or part of the steps of the methods described in various embodiments of the present invention.
  • the aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk, and other media that can store program codes. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Quality & Reliability (AREA)
  • Probability & Statistics with Applications (AREA)
  • High Energy & Nuclear Physics (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Magnetic Resonance Imaging Apparatus (AREA)

Abstract

本申请提供了一种磁共振图像处理方法、终端设备以及计算机存储介质。该磁共振图像处理方法包括：获取磁共振图像中每一病灶区域中每一像素的像素信号值时序数据；获取每一病灶区域中两两像素之间的像素信号值时序数据的距离矩阵；基于距离矩阵对每一病灶区域中的像素进行聚类，获取聚类中心的数量；基于聚类中心的数量输出每一病灶区域的病理结果。通过上述方式，本申请的磁共振图像处理方法直接对病灶区域进行逐像素分析，充分地展现病灶的异质性，提供病理分析的准确性。

Description

一种磁共振图像处理方法、终端设备及计算机存储介质 技术领域
本申请涉及磁共振应用技术领域，特别是涉及一种磁共振图像处理方法、终端设备以及计算机存储介质。
 背景技术
现有技术一般是通过对影像的形态学表现，以及病变的信号强度变化曲线及内部结构进行分析来鉴别诊断乳腺良恶性病变。其中可通过观测MRI（Magnetic Resonance Imaging，磁共振成像）的局部时间-信号强度折线（TIC）这一定性方法进行血流动力学特点分析，以此辅助乳腺癌病灶分级诊断。然而TIC分析往往基于医生手动划分的感兴趣区域（ROI）整体的信号强度变化平均值折线。病灶形态学上的复杂性，异质性，人工划分的感兴趣区域大小等内外因素，都会增加医生临床诊断的时间成本，并导致一定误差。
 技术解决方案
本申请提供了一种磁共振图像处理方法、终端设备以及计算机存储介质。
本申请提供了一种磁共振图像处理方法，所述磁共振图像处理方法包括：
获取所述磁共振图像中每一病灶区域中每一像素的像素信号值时序数据；
获取所述每一病灶区域中两两像素之间的像素信号值时序数据的距离矩阵；
基于所述距离矩阵对所述每一病灶区域中的像素进行聚类，获取聚类中心的数量；
基于所述聚类中心的数量输出所述每一病灶区域的病理结果。
其中，所述获取所述磁共振图像中每一病灶区域中每一像素的像素信号值时序数据，包括：
按照用户输入信息在所述磁共振图像上划分多个病灶区域；
提取连续多张切片的病灶区域中每个像素的信号强度数据；
为每一病灶区域建立一个数据集，所述数据集中包括多个像素及其像素信号值时序数据，其中，所述像素信号值时序数据为所述像素按照扫描采集信号时间排列的多个信号强度数据。
其中，所述获取所述每一病灶区域中两两像素之间的像素信号值时序数据的距离矩阵，包括：
利用动态时间规整技术计算所述每一病灶区域中两两像素之间的像素信号值时序数据的距离矩阵。
其中，所述磁共振图像处理方法，还包括：
按照不同的迭代次数，获取不同的聚类结果；
按照预设聚类模型评价指标从多个聚类结果中，确定最佳聚类结果及其聚类中心数量；
基于所述最佳聚类结果的聚类中心数量输出所述每一病灶区域的病理结果。
其中，所述基于所述距离矩阵对所述每一病灶区域中的像素进行聚类，获取聚类中心的数量之后，所述磁共振图像处理方法还包括：
按照恶性程度将多个病灶区域划分为一级病灶区域、二级病灶区域以及三级病灶区域；
从所述多个病灶区域的最佳评价指标中选择最佳平均评价指标，以使在所述最佳平均评价指标下，一级病灶区域、二级病灶区域以及三级病灶区域的最佳聚类中心数量依次递增。
其中，所述三级病灶区域的平均聚类中心数量大于所述二级病灶区域的平均聚类中心数量，所述二级病灶区域的平均聚类中心数量大于一级病灶区域的平均聚类中心数量。
其中，所述基于所述聚类中心的数量输出所述每一病灶区域的病理结果，包括：
将每一病灶区域的最佳聚类中心数量作为该病灶区域的特征信息；
基于所述特征信息分析每一聚类中心的折线类型；
统计每种折线类型的占比；
分析所述每一聚类中心的折线早期强化率和折线峰值，得到折线分析结果；
基于折线类型的占比以及折线分析结果输出该病灶区域的病理结果。
本申请还提供了一种终端设备，所述终端设备包括：
时序数据模块，用于获取所述磁共振图像中每一病灶区域中每一像素的像素信号值时序数据；
距离矩阵模块，用于获取所述每一病灶区域中两两像素之间的像素信号值时序数据的距离矩阵；
聚类中心模块，用于基于所述距离矩阵对所述每一病灶区域中的像素进行聚类，获取聚类中心的数量；
病理结果模块，用于基于所述聚类中心的数量输出所述每一病灶区域的病理结果。
本申请还提供了另一种终端设备，所述终端设备包括存储器和处理器，其中，所述存储器与所述处理器耦接；
其中，所述存储器用于存储程序数据，所述处理器用于执行所述程序数据以实现上述的磁共振图像处理方法。
本申请还提供了一种计算机存储介质，所述计算机存储介质用于存储程序数据，所述程序数据在被处理器执行时，用以实现上述的磁共振图像处理方法。
 有益效果
本申请的有益效果是：终端设备获取磁共振图像中每一病灶区域中每一像素的像素信号值时序数据；获取每一病灶区域中两两像素之间的像素信号值时序数据的距离矩阵；基于距离矩阵对每一病灶区域中的像素进行聚类，获取聚类中心的数量；基于聚类中心的数量输出每一病灶区域的病理结果。通过上述方式，本申请的磁共振图像处理方法直接对病灶区域进行逐像素分析，充分地展现病灶的异质性，提供病理分析的准确性。
 附图说明
为了更清楚地说明本发明实施例中的技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。其中：
图1是本申请提供的磁共振图像处理方法一实施例的流程示意图；
图2是本申请提供的像素提取方法的流程示意图；
图3是本申请提供的终端设备一实施例的结构示意图；
图4是本申请提供的终端设备另一实施例的结构示意图；
图5是本申请提供的计算机存储介质一实施例的结构示意图。
 本发明的实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅是本申请的一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
在DCE-MRI动态增强影像分析中，信号强度-时间（TIC）曲线分析是极具价值的测量工具，它通过强化前后信号强度的变化，动态反映了扫描部位血流动力学变化。半定量参数主要通过描述感兴趣区内组织信号强度-时间曲线的形状和结构来获得，不需要选择与组织相匹配的药代动力学模型。常用的半定量参数有初始曲线下面积、达峰时间、最大信号强度、最大斜率、廓清速率等。
其横轴为时间，纵轴为信号强度（可以理解为像素值）。现有的TIC分析普遍是基于TIC曲线形状的半定量分析。时间-信号强度曲线TIC因此被分为三类：上升型（强度呈缓慢持续增强，常见于良性病变）；平台型（动态早期信号强度到达最高峰，在延时期信号强度无明显变化，良，恶性都有可能）；及流出型（动态早期信号强度到达最高峰后降低，恶性病变几率大）。
TIC分析一般遵循以下步骤，医生划分选择MRI某病灶感兴趣区域（ROI），然后由软件给出感兴趣区域的平均“信号强度-时间曲线”，观测其TIC动态增强曲线形状，结合形态学等方法与指标作进一步诊断。目前对TIC的研究，普遍是分析三种类型TIC的分布情况来区分良性，恶性肿瘤，没有对良性和恶性进行进一步的分类。并且往往是基于峰值，达峰时间，折线下面积等半定量参数来定义特征进行对TIC折线的分类划型的。
基于上述理论基础，本申请实施例提出一种磁共振图像处理方法，围绕对乳腺DCE-MRI动态增强影像进行时间-信号强度进行分析，以辅助乳腺癌术前病理分级。具体请参阅图1，图1是本申请提供的磁共振图像处理方法一实施例的流程示意图。
其中，本申请的磁共振图像处理方法应用于一种终端设备，其中，本申请的终端设备可以为服务器，也可以为由服务器和电子设备相互配合的系统。相应地，终端设备包括的各个部分，例如各个单元、子单元、模块、子模块可以全部设置于服务器中，也可以分别设置于服务器和终端设备中。
进一步地，上述服务器可以是硬件，也可以是软件。当服务器为硬件时，可以实现成多个服务器组成的分布式服务器集群，也可以实现成单个服务器。当服务器为软件时，可以实现成多个软件或软件模块，例如用来提供分布式服务器的软件或软件模块，也可以实现成单个软件或软件模块，在此不做具体限定。在一些可能的实现方式中，本申请实施例的磁共振图像处理方法可以通过处理器调用存储器中存储的计算机可读指令的方式来实现。
具体而言，如图1所示，本申请实施例的磁共振图像处理方法具体包括以下步骤：
步骤S11：获取磁共振图像中每一病灶区域中每一像素的像素信号值时序数据。
在本申请实施例中，终端设备获取磁共振图像中每一病灶区域中每一像素的像素信号值时序数据。具体地，请参阅图2，图2是本申请提供的像素提取方法的流程示意图。
首先，利用磁共振仪器对患者进行DCE-MRI乳腺扫描，对扫描结果注射造影剂。经过2或3分钟之后，终端设备在相等的时间间隔内进行N次采集，每个获取的采集集合由m个切片组成。对于每个切片而言，放射科医生和专业人员对磁共振图像的每个切片的病变进行了标记，从而划分出多个病灶区域。
然后，终端设备提取每一张切片上每个像素在磁共振每一次扫描采集信号时间点所得到的信号强度数据，并进行去噪。终端设备将像素的信号强度数据按照扫描采集信号时间排序，得到病灶区域中所有像素的像素信号值时序数据。得到病灶区域中所有像素的像素信号值时序数据后，将病灶区域的每一个像素作为研究样本，为每一个病灶区域创建一个数据集。其中，数据集中的每一行对应单个像素的像素信号值时序数据，数据集中的每一列对应单个扫描采集信号时间所测到的所有像素的信号强度数据。例如，对于每个病灶区域Li，终端设备创建一个m*n的数据集，其中，数据集对应的病灶区域Li包含m个像素，并在n个扫描采集信号时间采集不同的信号强度数据。
最后，终端设备对每个数据集生成如图2所示的TIC折线图，其中，TIC折线图中每一条折线对应一个像素，TIC折线图的横坐标为相对采集时间，TIC折线图的纵坐标为相对信号强度。
步骤S12：获取每一病灶区域中两两像素之间的像素信号值时序数据的距离矩阵。
在本申请实施例中，以病灶区域Li为例。终端设备直接使用病灶区域Li中的像素信号值时序数据作为训练数据。
具体地，终端设备利用动态时间规整技术（DTW，Dynamic Time Warping）作为距离度量判断像素信号值时序数据之间的相似性，计算距离矩阵。假设病灶区域Li有N个像素，对应TIC折线图中的N条折线。最终，终端设备计算得到一个N*N的距离矩阵，距离矩阵中每个元素dist（x，y）表征像素x的TIC折线和像素y的TIC折线的距离，即相似度。
在其他实施例中，也可以采用欧式距离，LCSS（Longest-Common-Subsequence，最长公共子序列问题）等距离计算方式计算上述相似度。
步骤S13：基于距离矩阵对每一病灶区域中的像素进行聚类，获取聚类中心的数量。
在本申请实施例中，终端设备利用K-means或PAM（Partitioning Around Medoid，围绕中心点的划分）等聚类方法进行聚类。由于K-means以及PAM聚类模型需要提前输入k值，k值表示聚类模型的迭代次数。在本申请实施例中，终端设备可以选择一个比较大的上限值，例如100等。聚类模型从k=2到k=100不断迭代后，得到不同k值下不同的聚类中心，而每一个聚类中心Ci的形状代表了这一类像素的折线形状，即信号强度变化走势（Enhancement pattern，ER）。
在其他实施例中，也可以采用模糊c均值聚类Fuzzy c-Means(FCM)，改良离散K-中值聚类Modified Discrete k-Median Clustering(DKM-S)等聚类模型。
进一步地，终端设备还可以在不同k值的结果基础上，通过轮廓系数，Calinski and Harabasz score，Davis-Bouldin Index等聚类模型的内部评价指标来评价聚类效果。终端设备按照预设的聚类模型评价指标从多个聚类结果中，确定最佳的聚类迭代次数k值，以使在满足最佳的k值时，聚类结果的类间离散性高以及类内凝聚性强，即每个聚类类别的聚类中心形状均有代表性，且可以区分于其他聚类类别的聚类中心形状。
对于每个病灶区域而言，终端设备可以通过上述过程获取每个病灶区域的最佳聚类中心数量Ki。其中，病灶区域的最佳聚类中心数量Ki越大，说明该病灶区域内血流灌注情况越复杂，存在更多的信号强度变化走势形状，病灶区域的异质性越高。
其中，乳腺癌病灶的异质性与其恶性程度正向相关，即恶性程度越高，异质性越高，聚类中心数量越大。因此，本申请实施例可以按照恶性程度将多个病灶区域划分为一级病灶区域、二级病灶区域以及三级病灶区域，其中，三级病灶区域的平均聚类中心数量大于二级病灶区域的平均聚类中心数量，二级病灶区域的平均聚类中心数量大于一级病灶区域的平均聚类中心数量。
进一步地，终端设备比对各个病灶区域的内部评价指标及其对应的最佳聚类中心数量，选择在平均聚类中心数量能够体现出一级病灶区域、二级病灶区域以及三级病灶区域差异性的最佳评价指标，即利用该最佳评价指标能够得到各病灶区域的最佳聚类中心数量，一级病灶区域、二级病灶区域以及三级病灶区域的最佳聚类中心数量依次递增，可区分且差异性大。
终端设备可以对选择不同的聚类模型执行上述流程，并对各聚类模型的聚类结果进行排列组合，选择能够体现一级病灶区域、二级病灶区域以及三级病灶区域的平均聚类中心数量有最大差异，且递增的最佳组合，其中，最佳组合包括聚类模型种类以及内部评价指标。
步骤S14：基于聚类中心的数量输出每一病灶区域的病理结果。
在本申请实施例中，终端设备记录每一个病灶区域的最佳聚类中心数量Ki作为该病灶区域的新特征，即体征每个病灶区域会产生最佳聚类中心数量Ki的折线形状类型。终端设备对所有折线形状类型进行统计分析，利用DBA（dynamic barycenter averaging）得到Ki种聚类中心，把Ki种折线类型对应到上升型，平台型，流出型中，统计三种基础类型的不同占比，根据早期强化率，峰值（ER max）等信息进行分析，将分析结果也同样作为每一个病灶区域的特征。
进一步地，终端设备对所有的病灶区域建立数据集，将每个病灶区域作为训练样本。利用每个病灶区域的组学信息（病理分级，淋巴结转移等指标）作为标签，每个病灶区域的折线形状类型，和上述过程中分析得到的其他信息作为新的病灶特征，再进行分类，分析异质性，来实现病理分析临床上基于TIC折线形状的类别数量上来区分一二三级恶性病灶这一病理分级的目的，甚至于可以满足判断淋巴结转移等其他有临床诊断意义的期望。
在本申请实施例中，终端设备获取磁共振图像中每一病灶区域中每一像素的像素信号值时序数据；获取每一病灶区域中两两像素之间的像素信号值时序数据的距离矩阵；基于距离矩阵对每一病灶区域中的像素进行聚类，获取聚类中心的数量；基于聚类中心的数量输出每一病灶区域的病理结果。通过上述方式，本申请的磁共振图像处理方法直接对病灶区域进行逐像素分析，充分地展现病灶的异质性，提供病理分析的准确性。本申请实施例的磁共振图像处理方法可以直接对动态增强MR影像病灶区域进行逐像素动态时间-信号强度曲线（TIC）分析，从异质性角度来实现针对恶性乳腺肿瘤的病理分级，并实现预测淋巴结转移，预测癌症的分子分型，预测化疗敏感性，预测生存期等目的；较于传统机器学习，本申请实施例的磁共振图像处理方法在DCE-MRI信号值强度折线分类种创新应用DTW动态时间规整的距离度量来计算相似性，进行全局度量，而非欧氏距离，马氏距离，曼哈顿等；DTW动态时间规整这一方法针对序列匹配提出，尤其是当序列出现一定漂移，欧氏距离度量失效的情况下，DTW对时序数据按照形状进行分类效果良好。
本领域技术人员可以理解，在具体实施方式的上述方法中，各步骤的撰写顺序并不意味着严格的执行顺序而对实施过程构成任何限定，各步骤的具体执行顺序应当以其功能和可能的内在逻辑确定。
为实现上述实施例的磁共振图像处理方法，本申请还提出了一种终端设备，具体请参阅图3，图3是本申请提供的终端设备一实施例的结构示意图。
如图3所示，本申请提供的终端设备300包括时序数据模块31、距离矩阵模块32、聚类中心模块33以及病理结果模块34。
其中，时序数据模块31，用于获取所述磁共振图像中每一病灶区域中每一像素的像素信号值时序数据。
距离矩阵模块32，用于获取所述每一病灶区域中两两像素之间的像素信号值时序数据的距离矩阵。
聚类中心模块33，用于基于所述距离矩阵对所述每一病灶区域中的像素进行聚类，获取聚类中心的数量。
病理结果模块34，用于基于所述聚类中心的数量输出所述每一病灶区域的病理结果。
为实现上述实施例的磁共振图像处理方法，本申请还提出了另一种终端设备，具体请参阅图4，图4是本申请提供的终端设备另一实施例的结构示意图。
本申请实施例的终端设备400包括存储器41和处理器42，其中，存储器41和处理器42耦接。
存储器41用于存储程序数据，处理器42用于执行程序数据以实现上述实施例所述的磁共振图像处理方法。
在本实施例中，处理器42还可以称为CPU（Central Processing Unit，中央处理单元）。处理器42可能是一种集成电路芯片，具有信号的处理能力。处理器42还可以是通用处理器、数字信号处理器（DSP，Digital Signal Process）、专用集成电路（ASIC，Application Specific Integrated Circuit）、现场可编程门阵列（FPGA，Field Programmable Gate Array）或者其它可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件。通用处理器可以是微处理器或者该处理器42也可以是任何常规的处理器等。
本申请还提供一种计算机存储介质，如图5所示，计算机存储介质500用于存储程序数据51，程序数据51在被处理器执行时，用以实现如上述实施例所述的磁共振图像处理方法。
本申请还提供一种计算机程序产品，其中，上述计算机程序产品包括计算机程序，上述计算机程序可操作来使计算机执行如本申请实施例所述的磁共振图像处理方法。该计算机程序产品可以为一个软件安装包。
本申请上述实施例所述的磁共振图像处理方法，在实现时以软件功能单元的形式存在并作为独立的产品销售或使用时，可以存储在装置中，例如一个计算机可读取存储介质中。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的全部或部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一台计算机设备（可以是个人计算机，服务器，或者网络设备等）或处理器（processor）执行本发明各个实施方式所述方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器（ROM，Read-Only Memory）、随机存取存储器（RAM，Random Access Memory）、磁碟或者光盘等各种可以存储程序代码的介质。
以上所述仅为本申请的实施方式，并非因此限制本申请的专利范围，凡是利用本申请说明书及附图内容所作的等效结构或等效流程变换，或直接或间接运用在其他相关的技术领域，均同理包括在本申请的专利保护范围内。

Claims (10)

  1. 一种磁共振图像处理方法，其特征在于，所述磁共振图像处理方法包括：
    获取所述磁共振图像中每一病灶区域中每一像素的像素信号值时序数据；
    获取所述每一病灶区域中两两像素之间的像素信号值时序数据的距离矩阵；
    基于所述距离矩阵对所述每一病灶区域中的像素进行聚类，获取聚类中心的数量；
    基于所述聚类中心的数量输出所述每一病灶区域的病理结果。
  2. 根据权利要求1所述的磁共振图像处理方法，其特征在于，
    所述获取所述磁共振图像中每一病灶区域中每一像素的像素信号值时序数据，包括：
    按照用户输入信息在所述磁共振图像上划分多个病灶区域；
    提取连续多张切片的病灶区域中每个像素的信号强度数据；
    为每一病灶区域建立一个数据集，所述数据集中包括多个像素及其像素信号值时序数据，其中，所述像素信号值时序数据为所述像素按照扫描采集信号时间排列的多个信号强度数据。
  3. 根据权利要求1所述的磁共振图像处理方法，其特征在于，
    所述获取所述每一病灶区域中两两像素之间的像素信号值时序数据的距离矩阵，包括：
    利用动态时间规整技术计算所述每一病灶区域中两两像素之间的像素信号值时序数据的距离矩阵。
  4. 根据权利要求1所述的磁共振图像处理方法，其特征在于，
    所述磁共振图像处理方法，还包括：
    按照不同的迭代次数，获取不同的聚类结果；
    按照预设聚类模型评价指标从多个聚类结果中，确定最佳聚类结果及其聚类中心数量；
    基于所述最佳聚类结果的聚类中心数量输出所述每一病灶区域的病理结果。
  5. 根据权利要求4所述的磁共振图像处理方法，其特征在于，
    所述基于所述距离矩阵对所述每一病灶区域中的像素进行聚类，获取聚类中心的数量之后，所述磁共振图像处理方法还包括：
    按照恶性程度将多个病灶区域划分为一级病灶区域、二级病灶区域以及三级病灶区域；
    从所述多个病灶区域的最佳评价指标中选择最佳平均评价指标，以使在所述最佳平均评价指标下，一级病灶区域、二级病灶区域以及三级病灶区域的最佳聚类中心数量依次递增。
  6. 根据权利要求5所述的磁共振图像处理方法，其特征在于，
    所述三级病灶区域的平均聚类中心数量大于所述二级病灶区域的平均聚类中心数量，所述二级病灶区域的平均聚类中心数量大于一级病灶区域的平均聚类中心数量。
  7. 根据权利要求5所述的磁共振图像处理方法，其特征在于，
    所述基于所述聚类中心的数量输出所述每一病灶区域的病理结果，包括：
    将每一病灶区域的最佳聚类中心数量作为该病灶区域的特征信息；
    基于所述特征信息分析每一聚类中心的折线类型；
    统计每种折线类型的占比；
    分析所述每一聚类中心的折线早期强化率和折线峰值，得到折线分析结果；
    基于折线类型的占比以及折线分析结果输出该病灶区域的病理结果。
  8. 一种终端设备，其特征在于，所述终端设备包括：
    时序数据模块，用于获取所述磁共振图像中每一病灶区域中每一像素的像素信号值时序数据；
    距离矩阵模块，用于获取所述每一病灶区域中两两像素之间的像素信号值时序数据的距离矩阵；
    聚类中心模块，用于基于所述距离矩阵对所述每一病灶区域中的像素进行聚类，获取聚类中心的数量；
    病理结果模块，用于基于所述聚类中心的数量输出所述每一病灶区域的病理结果。
  9. 一种终端设备，其特征在于，所述终端设备包括存储器和处理器，其中，所述存储器与所述处理器耦接；
    其中，所述存储器用于存储程序数据，所述处理器用于执行所述程序数据以实现权利要求1-7中任一项所述的磁共振图像处理方法。
  10. 一种计算机存储介质，其特征在于，所述计算机存储介质用于存储程序数据，所述程序数据在被处理器执行时，用以实现权利要求1-7中任一项所述的磁共振图像处理方法。
PCT/CN2021/137626 2021-09-08 2021-12-13 一种磁共振图像处理方法、终端设备及计算机存储介质 WO2023035459A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202111049766.5 2021-09-08
CN202111049766.5A CN113808090A (zh) 2021-09-08 2021-09-08 一种磁共振图像处理方法、终端设备及计算机存储介质

Publications (1)

Publication Number Publication Date
WO2023035459A1 true WO2023035459A1 (zh) 2023-03-16

Family

ID=78894863

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/137626 WO2023035459A1 (zh) 2021-09-08 2021-12-13 一种磁共振图像处理方法、终端设备及计算机存储介质

Country Status (2)

Country Link
CN (1) CN113808090A (zh)
WO (1) WO2023035459A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117982106A (zh) * 2024-04-02 2024-05-07 天津市肿瘤医院（天津医科大学肿瘤医院） 一种基于mri图像的乳腺癌化疗疗效预测系统及方法

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2024108396A1 (zh) * 2022-11-22 2024-05-30 深圳先进技术研究院 一种测量血流动力学异质性的方法、装置及计算机设备
CN116740064B (zh) * 2023-08-14 2023-10-20 山东奥洛瑞医疗科技有限公司 一种核磁共振肿瘤区域提取方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100238170A1 (en) * 2007-10-15 2010-09-23 Koninklijke Philips Electronics N.V. Visualization of temporal data
CN109843176A (zh) * 2016-07-29 2019-06-04 诺瓦达克技术有限公司 用于利用机器学习来表征受验者的组织的方法和系统
CN110969618A (zh) * 2019-12-18 2020-04-07 南京航空航天大学 一种基于动态超声造影的肝肿瘤血管生成定量分析方法

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106250701B (zh) * 2016-08-10 2019-02-01 华东理工大学 计算机软件系统中实现肺结节危险程度分类的系统和方法
CN108765411B (zh) * 2018-06-05 2021-01-12 东北大学 一种基于影像组学的肿瘤分型系统
CN110610472A (zh) * 2018-06-16 2019-12-24 深圳市前海安测信息技术有限公司 实现肺结节图像分类检测的计算机装置及方法
CN111666970A (zh) * 2020-04-23 2020-09-15 杭州传化智能制造科技有限公司 生产信息的处理方法、装置、系统和计算机设备
CN111815610B (zh) * 2020-07-13 2023-09-12 广东工业大学 一种病变图像的病灶检测方法及装置

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100238170A1 (en) * 2007-10-15 2010-09-23 Koninklijke Philips Electronics N.V. Visualization of temporal data
CN109843176A (zh) * 2016-07-29 2019-06-04 诺瓦达克技术有限公司 用于利用机器学习来表征受验者的组织的方法和系统
CN110969618A (zh) * 2019-12-18 2020-04-07 南京航空航天大学 一种基于动态超声造影的肝肿瘤血管生成定量分析方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
K. NIE; P. BALTZER; B. PREIM; G. MISTELBAUER: "Knowledge‐Assisted Comparative Assessment of Breast Cancer using Dynamic Contrast‐Enhanced Magnetic Resonance Imaging", COMPUTER GRAPHICS FORUM : JOURNAL OF THE EUROPEAN ASSOCIATION FOR COMPUTER GRAPHICS, WILEY-BLACKWELL, OXFORD, vol. 39, no. 3, 18 July 2020 (2020-07-18), Oxford , pages 13 - 23, XP071551876, ISSN: 0167-7055, DOI: 10.1111/cgf.13959 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117982106A (zh) * 2024-04-02 2024-05-07 天津市肿瘤医院（天津医科大学肿瘤医院） 一种基于mri图像的乳腺癌化疗疗效预测系统及方法

Also Published As

Publication number Publication date
CN113808090A (zh) 2021-12-17

Similar Documents

Publication Publication Date Title
WO2023035459A1 (zh) 一种磁共振图像处理方法、终端设备及计算机存储介质
US10303986B2 (en) Automated measurement of brain injury indices using brain CT images, injury data, and machine learning
CN107016395B (zh) 稀疏表示的原发性脑部淋巴瘤和胶质母细胞瘤的鉴别系统
EP2158575B1 (en) Detecting haemorrhagic stroke in ct image data
CN110895817A (zh) 一种基于影像组学分析的mri影像肝纤维化自动分级方法
KR20190105452A (ko) 딥러닝 기반 뇌 질환 진단 방법
WO2007079207A2 (en) An integrated segmentation and classification approach applied to medical applications analysis
CN104143101A (zh) 一种基于超声图像的乳腺肿瘤区域自动识别方法
Mahapatra Automatic cardiac segmentation using semantic information from random forests
CN107358267A (zh) 一种基于互相关特征的乳腺超声图像多元分类系统及方法
Florez et al. Emergence of radiomics: novel methodology identifying imaging biomarkers of disease in diagnosis, response, and progression
WO2023193390A1 (zh) 在治疗前对直肠癌治疗抵抗及其分子机制的预测系统
US9483705B2 (en) Image processing device, image processing method, and image processing program
CN104835155A (zh) 基于分形的早期乳腺癌钙化点计算机辅助检测方法
Asodekar et al. Brain tumor classification using shape analysis of MRI images
CN104463830A (zh) 血管内斑块的侦测系统及方法
TW201800059A (zh) 乳房影像的分析方法以及其電子裝置
CN113033667A (zh) 一种超声图像两阶段深度学习的乳腺肿瘤分类方法和装置
CN109800820B (zh) 一种基于超声造影图像均匀程度的分类方法
TWM527991U (zh) 醫療影像處理裝置
Zheng et al. Improvement of visual similarity of similar breast masses selected by computer-aided diagnosis schemes
Sao Khue et al. Improving Brain Tumor Multiclass Classification With Semantic Features
Roy et al. Breast tumor segmentation using image segmentation algorithms
Acharya et al. Segmentation of pap smear images to diagnose cervical cancer types and stages
CN110349662A (zh) 用于过滤肺部肿块误测结果的跨图像集合的离群样本发现方法和系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21956628

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE