WO2014103061A1 - ボリュームデータ解析システム及びその方法 - Google Patents
ボリュームデータ解析システム及びその方法 Download PDFInfo
- Publication number
- WO2014103061A1 WO2014103061A1 PCT/JP2012/084207 JP2012084207W WO2014103061A1 WO 2014103061 A1 WO2014103061 A1 WO 2014103061A1 JP 2012084207 W JP2012084207 W JP 2012084207W WO 2014103061 A1 WO2014103061 A1 WO 2014103061A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- cluster
- data
- target
- learning
- volume data
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims description 54
- 238000007405 data analysis Methods 0.000 title claims description 16
- 239000013598 vector Substances 0.000 claims abstract description 91
- 230000006870 function Effects 0.000 claims abstract description 47
- 238000012937 correction Methods 0.000 claims description 34
- 238000004364 calculation method Methods 0.000 claims description 28
- 238000012545 processing Methods 0.000 claims description 19
- 230000008569 process Effects 0.000 claims description 18
- 238000012706 support-vector machine Methods 0.000 claims description 18
- 238000010801 machine learning Methods 0.000 claims description 10
- 238000012549 training Methods 0.000 claims description 10
- 238000005259 measurement Methods 0.000 claims description 8
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 description 15
- 238000010586 diagram Methods 0.000 description 11
- 238000002591 computed tomography Methods 0.000 description 6
- 201000010099 disease Diseases 0.000 description 4
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 4
- 238000002595 magnetic resonance imaging Methods 0.000 description 4
- 238000002600 positron emission tomography Methods 0.000 description 4
- 206010028980 Neoplasm Diseases 0.000 description 3
- 201000011510 cancer Diseases 0.000 description 3
- 230000001605 fetal effect Effects 0.000 description 3
- 210000003754 fetus Anatomy 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 208000037062 Polyps Diseases 0.000 description 1
- 210000004204 blood vessel Anatomy 0.000 description 1
- 210000000988 bone and bone Anatomy 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004195 computer-aided diagnosis Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 210000004185 liver Anatomy 0.000 description 1
- 201000007270 liver cancer Diseases 0.000 description 1
- 208000014018 liver neoplasm Diseases 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 239000000523 sample Substances 0.000 description 1
- 230000007480 spreading Effects 0.000 description 1
- 238000003892 spreading Methods 0.000 description 1
- 210000001519 tissue Anatomy 0.000 description 1
- 238000002604 ultrasonography Methods 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
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
-
- 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
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
- A61B6/52—Devices using data or image processing specially adapted for radiation diagnosis
- A61B6/5211—Devices using data or image processing specially adapted for radiation diagnosis involving processing of medical diagnostic data
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
- A61B6/52—Devices using data or image processing specially adapted for radiation diagnosis
- A61B6/5211—Devices using data or image processing specially adapted for radiation diagnosis involving processing of medical diagnostic data
- A61B6/5217—Devices using data or image processing specially adapted for radiation diagnosis involving processing of medical diagnostic data extracting a diagnostic or physiological parameter from medical diagnostic data
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
- A61B6/52—Devices using data or image processing specially adapted for radiation diagnosis
- A61B6/5211—Devices using data or image processing specially adapted for radiation diagnosis involving processing of medical diagnostic data
- A61B6/5229—Devices using data or image processing specially adapted for radiation diagnosis involving processing of medical diagnostic data combining image data of a patient, e.g. combining a functional image with an anatomical image
- A61B6/5247—Devices using data or image processing specially adapted for radiation diagnosis involving processing of medical diagnostic data combining image data of a patient, e.g. combining a functional image with an anatomical image combining images from an ionising-radiation diagnostic technique and a non-ionising radiation diagnostic technique, e.g. X-ray and ultrasound
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/48—Diagnostic techniques
- A61B8/483—Diagnostic techniques involving the acquisition of a 3D volume of data
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S15/00—Systems using the reflection or reradiation of acoustic waves, e.g. sonar systems
- G01S15/88—Sonar systems specially adapted for specific applications
- G01S15/89—Sonar systems specially adapted for specific applications for mapping or imaging
- G01S15/8906—Short-range imaging systems; Acoustic microscope systems using pulse-echo techniques
- G01S15/8993—Three dimensional imaging systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T15/00—3D [Three Dimensional] image rendering
- G06T15/08—Volume rendering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
- G06T17/20—Finite element generation, e.g. wire-frame surface description, tesselation
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
- A61B6/02—Arrangements for diagnosis sequentially in different planes; Stereoscopic radiation diagnosis
- A61B6/03—Computed tomography [CT]
- A61B6/032—Transmission computed tomography [CT]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
- A61B6/02—Arrangements for diagnosis sequentially in different planes; Stereoscopic radiation diagnosis
- A61B6/03—Computed tomography [CT]
- A61B6/037—Emission tomography
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
- A61B6/46—Arrangements for interfacing with the operator or the patient
- A61B6/461—Displaying means of special interest
- A61B6/466—Displaying means of special interest adapted to display 3D data
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/46—Ultrasonic, sonic or infrasonic diagnostic devices with special arrangements for interfacing with the operator or the patient
- A61B8/461—Displaying means of special interest
- A61B8/466—Displaying means of special interest adapted to display 3D data
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/52—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/5215—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving processing of medical diagnostic data
- A61B8/5223—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving processing of medical diagnostic data for extracting a diagnostic or physiological parameter from medical diagnostic data
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/52—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S15/00
- G01S7/52017—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S15/00 particularly adapted to short-range imaging
- G01S7/52023—Details of receivers
- G01S7/52036—Details of receivers using analysis of echo signal for target characterisation
-
- 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
-
- 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
- G06T2207/10081—Computed x-ray tomography [CT]
-
- 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
- G06T2207/30056—Liver; Hepatic
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2210/00—Indexing scheme for image generation or computer graphics
- G06T2210/41—Medical
Definitions
- the present invention relates to a volume data analysis system and method for extracting a target from three-dimensional volume data.
- medical image processing devices for visualizing information inside the human body are rapidly spreading and are used for various diagnoses.
- modalities used in medical image processing apparatuses include X-ray CT (Computed Tomography), PET (Positron Emission Tomography), MRI (Magnetic Resonance Imaging), US (Ultrasonography), etc. These modalities are three-dimensional. Volume data can be obtained.
- a doctor can diagnose the presence and progression of liver cancer.
- an automatic detection device (hereinafter referred to as CAD: Computer Aided Diagnosis) that automatically detects a disease typified by cancer or the like from volume data for the purpose of reducing the burden on the doctor. )
- CAD Computer Aided Diagnosis
- CAD Computer Aided Diagnosis
- specific noise may be mixed in the volume data depending on the modality, and depending on the noise, it is difficult to automatically detect a disease with an automatic detection device. Therefore, development of an automatic detection device that can detect a disease as accurately as a doctor is now strongly desired.
- the technique which detects a target object from volume data is disclosed by the following patent documents 1, 2, for example.
- JP 2009-22342 A Japanese Patent No. 4388121
- Patent Document 1 when extracting a target from volume data, threshold processing for determining the size of the target with a threshold is performed, so that noise components included in the volume data can be removed.
- the target is present in the non-noise region, there is a problem that the target is removed if the size is equal to or smaller than the threshold value.
- the work of setting the threshold is complicated, and the burden on doctors and engineers (hereinafter referred to as users) who are users of the apparatus is large.
- a threshold value must be set for each of a plurality of different types of scales.
- the feature amount for performing polyp detection has a three-dimensional rotation dependency, and it is necessary to perform learning for each different rotation. That is, if the rotation pattern is not increased, the detection performance may be reduced. Moreover, there is a concern about an increase in calculation cost required for pattern rotation. Furthermore, since it is necessary to calculate the feature amount directly from the voxel value of the three-dimensional volume data, the calculation cost is high.
- the present invention has been made in view of the problems of the conventional technique, and an object thereof is to provide a volume data analysis system and method capable of reducing the calculation cost.
- the present invention converts the three-dimensional volume data generated by the modality into polygon data and converts it into polygon data, and divides the converted polygon data into a plurality of clusters to form each cluster.
- the L2 norm vector of the spherical harmonic function is calculated for each cluster as a feature vector based on the polygon data to be processed, and based on the calculated feature vector and learning data obtained by machine learning using training data. , Identifying whether each of the clusters is a target, and displaying an image of the cluster identified as the target on at least a screen.
- the calculation cost can be reduced.
- FIG. 1 It is a block diagram of a volume data analysis system.
- (A) is a schematic diagram of three-dimensional volume data
- (b), (c), and (d) are configuration diagrams of a two-dimensional image obtained from the three-dimensional volume data.
- It is a flowchart for demonstrating the process of a main processor.
- It is a display example of an image obtained by converting volume data into polygons, and is a diagram illustrating a display example of an image of only a target. It is a figure which shows the example of a display of the image obtained by converting volume data into a polygon, Comprising: The example of a display containing the target and a non-target.
- (A) is a figure which shows the example of a display of the three-dimensional image of the cluster identified as noise
- (b) calculates the L2 norm vector of an N-dimensional spherical harmonic function from the polygon data of the cluster which shows noise. It is the figure which visualized the calculation result as a histogram.
- (A) is a figure which shows the example of a display of the three-dimensional image of the cluster which rotated the cluster shown to Fig.6 (a) 90 degree
- (b) is from the polygon data of the cluster shown to Fig.7 (a).
- FIG. 1 is a configuration diagram of a volume data analysis system.
- the volume data analysis system includes a plurality of modalities 10, 12, 14, and 16, a display 18, a network processor 20, and an automatic detection device 22, and automatically detects diseases such as cancer from the volume data. Can be analyzed.
- Each modality 10 to 16 is connected to a network processor 20, and the network processor 20 is connected to a display 18 and an automatic detection device 22.
- the automatic detection device 22 includes a main processor 24, a main memory 26, an auxiliary memory 28, an input device 30, and an output device 32.
- the main processor 24 is connected to the network processor 20, connected to the output device 32 via the main memory 26, and further connected to the auxiliary memory 28 via the input device 30.
- Each modality 10 to 16 is configured as a detector that transmits and receives signals to and from an analysis object or a measurement object, and generates three-dimensional volume data related to a target (hereinafter sometimes referred to as a target).
- the modality 10 is configured by X-ray CT
- the modality 12 is configured by MRI
- the modality 14 is configured by US
- the modality 16 is configured by PET.
- the modality (US) 14 is used to avoid the effects of exposure, and when the target is a lesion such as cancer, precise volume data is obtained.
- modality (X-ray CT) 10 modality (MRI) 12, or modality (PET) 16 is used.
- the modality 14 scans the ultrasonic wave from the ultrasonic probe to the tomographic plane (scanning plane) inside the fetus, receives the reflected echo signal returning from each part of the biological tissue on the tomographic plane, and digitizes it.
- the modality 14 has a function of generating frame data (volume data of each tomographic image) obtained by scanning a plurality of scan planes as three-dimensional volume data . That is, the modality 14 has a function of generating a three-dimensional volume data by receiving a reflected echo signal (signal under measurement) obtained by scanning a measurement object.
- the 3D volume data generated by the modality 14 is transferred to the main processor 24 via the network processor 20.
- the three-dimensional volume data is stored in the main memory 26 as a cube-shaped three-dimensional original array, for example, a voxel value of 512 ⁇ 512 ⁇ 512 voxels (a value corresponding to a luminance value of a two-dimensional image).
- the main processor 24 generates image data (for example, image data related to a three-dimensional image obtained by projecting a three-dimensional shape determined based on the value of the three-dimensional volume data onto a two-dimensional plane) from the received three-dimensional volume data.
- image data for example, image data related to a three-dimensional image obtained by projecting a three-dimensional shape determined based on the value of the three-dimensional volume data onto a two-dimensional plane.
- the generated image data is transferred to the display 18 via the network processor 20, and a three-dimensional image based on the transferred image data is displayed on the screen of the display 18.
- the display 18 is a display device having a touch panel function, and constitutes a display device that displays the processing result of the main processor 24, and inputs information such as correction information in response to a user operation, for example, correction operation. It has a function as a device.
- the network processor 20 exchanges data or information with the modalities 10 to 16, exchanges information or data with the display 18, and further exchanges data or information with the main processor 24.
- the automatic detection device 22 is configured as a general medical workstation.
- the main processor 24 has, for example, a CPU (Central Processing Unit) and functions as a controller that controls the entire automatic detection device 22. At this time, the main processor 24 also functions as a controller that processes the three-dimensional volume data generated by the modality 14 and information input from the display 18 or the input device 30.
- a CPU Central Processing Unit
- the main memory 26 is configured by, for example, DRAM (Dynamic Random Access Memory), and is configured as a storage device that stores three-dimensional volume data and the like.
- the auxiliary memory 28 is configured by, for example, an SSD (Solid State Drive) and functions as an auxiliary storage device.
- the input device 30 includes, for example, a mouse, a trackball, a keyboard, and the like, and has a function of receiving instructions from the user.
- the output device 32 includes, for example, a DVD (Digital Versatile Disc) -RW (ReWritable) drive.
- the input device 30 can also be used as an input device for inputting correction information by a user operation.
- Fig. 2 (a) is a schematic diagram of three-dimensional volume data
- Figs. 2 (b), (c), and (d) are configuration diagrams of two-dimensional images obtained from the three-dimensional volume data.
- the three-dimensional volume data 100 is three-dimensional volume data generated by the modality 14 and is composed of voxel values in a cube-shaped three-dimensional array.
- volume data 101, 102, and 103 are extracted from the three-dimensional volume data 100.
- images 201, 202, and 203 are two-dimensional images obtained from the volume data 101, 102, and 103, respectively, and represent fetal cross-sectional images. That is, the images 201, 202, and 203 correspond to cross-sectional views obtained by slicing fetal images in the horizontal direction. At this time, the images 201, 202, and 203 are visualized so that the higher the value of the voxel value is, the whiter the color is.
- the main processor 24 determines the distribution of the voxel values of the three-dimensional volume data 100 to execute processing for detecting the target.
- FIG. 3 is a flowchart for explaining the processing of the main processor.
- the main processor 24 acquires 3D volume data from the modality 14 (S11), converts the acquired 3D volume data into polygons and converts them into polygon data, and divides the converted polygon data into a plurality of clusters. (S12). Thereafter, the main processor 24 calculates, for each cluster, the L2 norm vector of the spherical harmonic function as a feature vector based on the polygon data constituting each cluster (S13), and compares the calculated feature vector with the learning data. Then, it is identified whether or not each cluster is a target (S14), and each identification result is displayed on the screen of the display 18, and each identification result is determined (S15).
- the main processor 24 ends the processing in this routine.
- the user determines that the cluster displayed on the display 18 is a non-target, that is, the identification result is “false” and is a noise, and a tag 403 is added to the cluster. Is input, the relearning process is executed (S16), and then the process returns to step S14 to repeat the processes of steps S14 to S15.
- step S ⁇ b> 11 the main processor 24 acquires three-dimensional volume data from the modality 14 and stores the acquired three-dimensional volume data in the main memory 26.
- step S12 the main processor 24 reads the three-dimensional volume data stored in the main memory 26 into the work area, converts the read three-dimensional volume data into polygons and converts them into polygon data, and the converted polygon data Is divided into multiple clusters.
- the main processor 24 employs a marching cubes algorithm, for example, from the viewpoint of reducing calculation costs, among methods such as Delaunay triangulation and marching cubes algorithm.
- the main processor 24 functions as a cluster generation unit that converts the three-dimensional volume data generated by the modality 14 into polygons and converts them into polygon data, and divides the converted polygon data into a plurality of clusters.
- the polygon is a polygon made up of a group of three-dimensional vertex coordinates with order.
- a triangle is used as the polygon.
- a cluster is a collection of polygons in which the vertices of a polygon (triangle) are connected to each other.
- Each cluster is composed of polygon data belonging to the polygons of each cluster.
- FIG. 4 is a display example of an image obtained by converting volume data into polygons, and shows a display example of an image of only a target.
- a cluster identified by the main processor 24 as a target for example, an image of the cluster 401 is schematically displayed as a three-dimensional image.
- FIG. 5 is a display example of an image obtained by converting volume data into polygons, and shows a display example of an image including a target and a non-target.
- a cluster identified by the main processor 24 as a target for example, an image of the cluster 401 is schematically displayed as a three-dimensional image.
- a cluster that is identified as a target, but corresponds to a non-target, for example, images of clusters 402 and 403 are schematically displayed as a three-dimensional image.
- the front and back can be expressed by the cyclic direction of the vertices of each cluster.
- the counterclockwise tour is the front and the clockwise tour is the back.
- a tag 404 can be added to the cluster 402 as an identification label for specifying the cluster 402 as a non-target.
- the information of the tag 404 is input to the main processor 24 via the network processor 20 as correction information accompanying the user's correction operation.
- the main processor 24 executes re-learning processing using a machine learning algorithm based on the input correction information.
- step S13 the main processor 24 calculates the feature vector of each cluster, and holds each feature vector obtained by the calculation in the main memory 26 as a one-dimensional array.
- the rotation-invariant feature amount based on the spherical harmonic function is used. However, by using various other feature amounts together, more accurate identification is performed. It becomes possible.
- the volume of the cluster As a method of calculating the feature amount held by the feature vector, first, (1) the volume of the cluster, (2) the average value of the volume data included in the cluster, and (3) the variance of the volume data included in the cluster are calculated. A method is mentioned.
- a method for calculating the feature amount held by the feature vector (4) a method for calculating a histogram of volume data included in the cluster can be cited. When this method is used, the histogram can be added to the feature vector by quantizing the voxel value of the volume data included in the cluster into 25 levels.
- a technique for calculating the feature amount held by the feature vector (5) an average curvature of polygon data included in the cluster, and (6) a technique of calculating Gaussian curvature of the polygon data included in the cluster can be cited. Furthermore, (7) a method of calculating a vector of L2 norm that is a coefficient of the spherical harmonic function from the polygon data of the cluster can be mentioned.
- a method of calculating a 20-dimensional L2 norm vector and adding the calculation result to the feature vector can be used.
- the feature vector of one cluster is 50 dimensions.
- the method (7) which is a method for calculating a rotation invariant feature quantity based on a spherical harmonic function, is used as a basis. Any one of methods 6) is combined.
- the number of feature vectors is 145. If volume data is converted to polygon data and converted to polygon data, and feature vectors are calculated using the converted polygon data, it is not necessary to search all volume data, so feature vectors can be calculated directly from volume data. Compared to the method, the calculation cost is greatly reduced. It should be noted that the value of each dimension of the feature vector is greatly different in value range, and thus correction processing is performed. As an example of the correction processing, by performing scaling so that the value becomes 0 to 1, it is possible to perform censoring with a constant, logarithmic scaling, or the like.
- the spherical harmonic function y l m ( ⁇ ) is given by the following equation.
- l is the order of the spherical harmonic function
- P l m is the Legendre function
- m is an integer that satisfies ⁇ l ⁇ m ⁇ l
- FIG. 6A is a diagram showing a display example of a three-dimensional image of a cluster identified as noise.
- a cluster 403 composed of a plurality of polygons is a cluster identified as noise.
- the main processor 24 performs scaling processing on the cluster 403.
- the scaling process After converting the three-dimensional coordinate values of polygon data constituting one cluster 403 into a coordinate system with the center of gravity of the cluster itself as the origin, the vertex coordinates of the polygon are enlarged (or reduced). Then, convert to a certain size. In the case of the cabling process, adjustment is performed so that the aspect ratio is not changed. By performing this scaling process, it is possible to perform a stable identification process regardless of the size of the cluster. As an example of scaling processing, in this system, the major axis (maximum cluster width) is reduced (or enlarged) to be 1. After the scaling process, the following process is executed.
- the L2 norm vector of the N-dimensional spherical harmonic function with index k is expressed by the following equation and added to the feature vector as the feature amount obtained by the method (7).
- FIG. 6 (b) shows an example in which the L2 norm vector of the N-dimensional spherical harmonic function is calculated from the polygon data by the above formula (4) and the calculation result is visualized as a histogram.
- FIG. 6B is a diagram in which the L2 norm vector of the N-dimensional spherical harmonic function is calculated from the polygon data of the cluster indicating noise, and the calculation result is visualized as a histogram.
- each histogram is a histogram obtained from polygon data of the cluster 403 indicating noise, and is a one-dimensional to twenty-dimensional histogram.
- the horizontal direction indicates the dimension n of the feature vector
- the vertical direction indicates the value of the L2 norm of the spherical harmonic function.
- FIG. 7A shows a display example of a three-dimensional image of the cluster 403 obtained by rotating the cluster shown in FIG. 6A by 90 degrees.
- FIG. 7B shows an example in which the L2 norm vector of the N-dimensional spherical harmonic function is calculated from the polygon data of the cluster 403 shown in FIG. FIG.
- the L2 norm vector of the N-dimensional spherical harmonic function is calculated from the polygon data using the equation (4), even when the cluster 403 is rotated 90 degrees, the same cluster 403 is used.
- the L2 norm vectors of the calculated spherical harmonics are equal. That is, the value of the histogram of each dimension shown in FIG. 6B is equal to the value of the histogram of each dimension shown in FIG. 7B.
- the cluster 403 is decomposed into variously shaped spheres, and a process of calculating how much the decomposed various spheres are included is performed.
- the rotational dependency can be removed by using the formula (4) and sequentially adding the dimension terms each time the dimension increases.
- the decomposition by the spherical harmonic function is equivalent to Fourier transform using a sphere.
- the main processor 24 functions as a feature vector calculation unit that calculates, for each cluster, an L2 norm vector of a spherical harmonic function as a feature vector based on polygon data constituting each cluster.
- the method for calculating the feature value by combining the marching cubes algorithm and the spherical harmonic function has the following advantages.
- the polygonization by the marching cubes algorithm has a disadvantage that a topological hole is generated.
- the method of calculating the value of the L2 norm vector of the spherical harmonic function of the cluster is adopted, the calculation result for the cluster having a topological hole and the topological Since a change appears between the calculation results regarding the cluster in which no hole is generated, the topological hole can be effectively removed as compared with the case where other feature values are used.
- the perforated clusters are displayed in a visualized state, so that the user can efficiently remove the perforated clusters.
- step 14 a process of generating identification label data from the feature vector is executed.
- the target is identified from the feature vector using machine learning.
- machine learning method for example, a neural network, AdaBoost, and support vector machine (SVM) can be used.
- SVM support vector machine
- an example of multi-level identification by SVM support vector machine will be described from the viewpoint of high identification accuracy and ease of explanation.
- the single kernel SVM definition is shown below.
- ⁇ is an optimization variable
- K (x i , x j ) is a kernel
- x is a feature vector.
- the subscripts i and j of each variable represent the feature vector index, not the dimension.
- X i that is a combination of ⁇ i ⁇ 0 and an accompanying feature vector (generally called a support vector) is taken as learning data.
- identification by machine learning requires learning in advance based on training data (a combination of ⁇ , x, and t).
- the main processor 24 executes a learning process in advance using an SVM (support vector machine), and stores the learning result as learning data together with the training data.
- the equation (5) is a single-kernel SVM, but by using a multi-kernel SVM extended to a plurality of different kernels, it is possible to perform more accurate identification.
- multi-kernel SVM it is possible to perform identification processing after learning with different weights for each dimension k of feature vectors, which is suitable when using different types of feature quantities for feature vectors. It is.
- FIG. 8 is a diagram showing a display example of an image visualizing the learning result by the support vector machine.
- FIG. 8 shows an example in which 50-dimensional feature vectors are used, only 2 dimensions are extracted, and an image of the extracted feature vectors is visualized. .
- the vertical axis in FIG. 8 is one dimension, and the horizontal axis in FIG. 8 is another dimension.
- a point drawn with a square represents the target 501
- a point drawn with a triangle represents the non-target 502.
- the points drawn in circles are training points 503 and 504 located near the boundary between the target 501 and the non-target 502, which means a support vector.
- Lines drawn with curved lines are identification boundaries 505 and 506 calculated by learning data.
- the identification boundaries 505 and 506 are boundaries for identifying the target 501 and the non-target 502, and are calculated at positions where the distance becomes the maximum with reference to the support vector (training points 503 and 504). .
- the main processor 24 calculates the following expression from the learning data and the feature vector for each cluster.
- b is a bias parameter and can be calculated from learning data. Further, if the value of the equation (6) is positive, a label indicating the target is given.
- the cluster is identified as the target 501, and if the value of equation (6) is negative, the cluster is a non-target 502. Identified.
- a method of preparing a plurality of learning data and solving as a multi-cluster problem using a one-to-many method or a one-to-one method can be adopted.
- the main processor 24 can classify the clusters as targets or non-targets and classify the clusters identified as targets into two types.
- the label data one-dimensional array having tuples as elements can be used.
- a label used for identification for example, a class A label means a fetal region, a class B label means a uterine region, and a non-target label uses a label that means a non-target region. be able to.
- the label data used for identification is stored in the main memory 26 as identification label data.
- step 15 processing for visualizing each cluster is executed based on the identification label data.
- FIG. 9 is a diagram showing a display example of a three-dimensional image when an identification label is added to the cluster.
- a non-target label 601 indicating a non-target is added to the cluster 401
- a target label 402 indicating a class A target is added to the cluster 402.
- the user can input correction information for correcting the cluster 402 to a non-target by adding a tag 403 by a touch panel operation of the display 18.
- the main processor 24 corrects and displays the target label 602 as a non-target label, corrects the label data of the cluster 402 from the label data indicating the target to the label data indicating the non-target, and the corrected label. Data is stored in the main memory 26.
- the main processor 24 determines that the cluster 401 Is corrected from label data indicating a non-target to label data indicating a target, and the corrected label data is stored in the main memory 26.
- step 16 a relearning process is executed.
- the main processor 24 generates learning data as new learning data by performing relearning of the target based on the corrected label data, feature vector, and training data.
- the main processor 24 adds the corrected label data and feature vector to the training data stored at the time of pre-learning, optimizes the parameter (optimization variable) based on the equation (5),
- the learning data obtained after the optimization is stored in the main memory 26, and the process of step 14 is executed again.
- the main processor 24 calculates a feature vector calculation unit for each cluster using the L2 norm vector of the spherical harmonic function as a feature vector based on polygon data constituting each cluster, and each feature vector calculated by the feature vector calculation unit.
- an identification unit for identifying whether each cluster is a target based on learning data obtained by machine learning using training data, and among the clusters generated by the cluster generation unit, An image of a cluster identified as a target is generated from polygon data constituting each cluster, and functions as an image generation unit that displays at least the generated image on the display 18.
- the main processor 24 selects the cluster specified by the input correction information.
- the feature data of the cluster that has been corrected by the correction unit and the correction data of the correction unit related to the cluster are used to re-learn the learning data. It functions as a re-learning unit that learns and generates new learning data.
- the main processor 24 when new learning data is generated by the re-learning unit, the main processor 24 generates the feature vector of the cluster corrected by the correcting unit and the re-learning unit out of the feature vectors calculated by the feature vector calculating unit. Based on the newly learned data, the cluster corrected by the correction unit is identified as a target and functions as an identification unit that outputs the identification result to the image generation unit.
- the main processor 24 selects the cluster specified by the input correction information.
- the cluster specified by the input correction information is targeted. Functions as a correction unit for correcting.
- the main processor 24 determines each cluster based on the plurality of types of learning data and the feature vectors calculated by the feature vector calculation unit.
- the identification unit for adding the label data according to the identification result, and the cluster displayed as an image on the display 18 according to the label data added to the identification result of the identification unit It functions as an image generation unit that displays an identification label.
- the calculation cost can be greatly reduced. Further, since only the target region can be detected from the volume data and the detected target image can be visualized, false detection is reduced. Furthermore, since the feature quantity of the feature vector has no rotation dependency, both improvement in detection accuracy and reduction in calculation cost can be achieved. Moreover, since the calculation cost is low, the result can be obtained at high speed.
- this invention is not limited to the above-mentioned Example, Various modifications are included.
- the above-described embodiments have been described in detail for easy understanding of the present invention, and are not necessarily limited to those having all the configurations described.
- each of the above-described configurations, functions, etc. may be realized by hardware by designing a part or all of them, for example, by an integrated circuit.
- Each of the above-described configurations, functions, and the like may be realized by software by interpreting and executing a program that realizes each function by the processor.
- Information such as programs, tables, and files that realize each function is stored in memory, a hard disk, a recording device such as an SSD (Solid State Drive), an IC (Integrated Circuit) card, an SD (Secure Digital) memory card, a DVD ( It can be recorded on a recording medium such as Digital Versatile Disc).
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Biomedical Technology (AREA)
- Radiology & Medical Imaging (AREA)
- Pathology (AREA)
- Surgery (AREA)
- Biophysics (AREA)
- Veterinary Medicine (AREA)
- Animal Behavior & Ethology (AREA)
- Heart & Thoracic Surgery (AREA)
- Molecular Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- High Energy & Nuclear Physics (AREA)
- Remote Sensing (AREA)
- Radar, Positioning & Navigation (AREA)
- Optics & Photonics (AREA)
- Theoretical Computer Science (AREA)
- Acoustics & Sound (AREA)
- Computer Graphics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Quality & Reliability (AREA)
- Software Systems (AREA)
- Geometry (AREA)
- Physiology (AREA)
- Data Mining & Analysis (AREA)
- Primary Health Care (AREA)
- Epidemiology (AREA)
- Databases & Information Systems (AREA)
- Image Analysis (AREA)
- Ultra Sonic Daignosis Equipment (AREA)
- Pulmonology (AREA)
Abstract
Description
以下、本発明の一実施例を図面に基づいて説明する。図1は、ボリュームデータ解析システムの構成図である。図1において、ボリュームデータ解析システムは、複数のモダリティ10、12、14、16と、ディスプレイ18と、ネットワークプロセッサ20と、自動検出装置22とを有し、ボリュームデータから、癌などの疾患を自動的に解析することができる。各モダリティ10~16は、それぞれネットワークプロセッサ20に接続され、ネットワークプロセッサ20は、ディスプレイ18と自動検出装置22に接続される。
ω=(sinθcosφ,sinθsinφ,cosθ)となる方向である。
Claims (10)
- 測定対象を走査して得られた被測定信号を受信して三次元ボリュームデータを生成するモダリティと、
操作による情報を入力する入力装置と、
前記モダリティで生成された三次元ボリュームデータと前記入力装置から入力された情報を処理するコントローラと、
前記コントローラの処理結果を表示する表示装置と、を有し、
前記コントローラは、
前記モダリティで生成された三次元ボリュームデータをポリゴン化してポリゴンデータに変換し、前記変換されたポリゴンデータを複数のクラスタに分割するクラスタ生成部と、
前記各クラスタを構成するポリゴンデータを基に球面調和関数のL2ノルムベクトルを特徴ベクトルとして前記クラスタ毎に算出する特徴ベクトル算出部と、
前記特徴ベクトル算出部の算出による各特徴ベクトルと、訓練データを用いた機械学習で得られた学習データとを基に、前記各クラスタがターゲットであるか否かを識別する識別部と、
前記クラスタ生成部の生成による各クラスタのうち、前記識別部でターゲットであると識別されたクラスタの画像を、当該各クラスタを構成するポリゴンデータから生成し、少なくとも当該生成した画像を前記表示装置に表示させる画像生成部と、を含むことを特徴とするボリュームデータ解析システム。 - 前記コントローラは、
前記入力装置から、前記表示装置に画像表示されたクラスタのうちいずれかのクラスタを訂正するための訂正情報が入力された場合、当該入力された訂正情報で指定されたクラスタをターゲット又は非ターゲットに訂正する訂正部と、
前記特徴ベクトル算出部の算出による特徴ベクトルのうち、前記訂正部で訂正されたクラスタの特徴ベクトルと、当該クラスタに関する前記訂正部の訂正結果を基に前記学習データを再学習して、新規学習データを生成する再学習部と、を有し、
前記識別部は、
前記再学習部で前記新規学習データが生成された場合、前記特徴ベクトル算出部の算出による特徴ベクトルのうち、前記訂正部で訂正されたクラスタの特徴ベクトルと、前記再学習部で生成された新規学習データとを基に、前記訂正部で訂正されたクラスタが、ターゲットであるか否かを識別し、当該識別結果を前記画像生成部に出力することを特徴とする請求項1に記載のボリュームデータ解析システム。 - 前記識別部は、
前記学習データとして、前記ターゲットの種類に応じた複数種類の学習データを用いる場合、前記複数種類の学習データと前記特徴ベクトル算出部の算出による各特徴ベクトルとを基に、前記各クラスタがターゲットであるか否かを識別すると共に、前記各識別結果に、当該識別結果に応じたラベルデータを付加し、
前記画像生成部は、
前記表示装置に画像表示されたクラスタに、前記識別部の識別結果に付加されたラベルデータに応じた識別ラベルを表示させることを特徴とする請求項1に記載のボリュームデータ解析システム。 - 前記クラスタ生成部は、
ドロネー三角形分割法又はマーチングキューブスアルゴリズムのいずれかを用いて、前記モダリティで生成された三次元ボリュームデータをポリゴン化してポリゴンデータに変換することを特徴とする請求項1に記載のボリュームデータ解析システム。 - 前記識別部は、
前記機械学習に、サポートベクターマシン、マルチサポートベクターマシン、アダブースト、オンラインサポートベクターマシン、又はニューラルネットワークのうちいずれか一つを用いることを特徴とする請求項1に記載のボリュームデータ解析システム。 - 測定対象を走査して得られた被測定信号を受信して三次元ボリュームデータを生成するモダリティと、
操作による情報を入力する入力装置と、
前記モダリティで生成された三次元ボリュームデータと前記入力装置から入力された情報を処理するコントローラと、
前記コントローラの処理結果を表示する表示装置と、を有するシステムにおけるボリュームデータ解析方法であって、
前記コントローラは、
前記モダリティで生成された三次元ボリュームデータをポリゴン化してポリゴンデータに変換し、前記変換されたポリゴンデータを複数のクラスタに分割する第1のステップと、
前記各クラスタを構成するポリゴンデータを基に球面調和関数のL2ノルムベクトルを特徴ベクトルとして前記クラスタ毎に算出する第2のステップと、
前記第2のステップで算出された各特徴ベクトルと、訓練データを用いた機械学習で得られた学習データとを基に、前記各クラスタがターゲットであるか否かを識別する第3のステップと、
前記第1のステップで生成された各クラスタのうち、前記第3のステップでターゲットであると識別されたクラスタの画像を、当該各クラスタを構成するポリゴンデータから生成し、少なくとも当該生成した画像を前記表示装置に表示させる第4のステップと、を実行することを特徴とするボリュームデータ解析方法。 - 前記コントローラは、
前記入力装置から、前記表示装置に画像表示されたクラスタのうちいずれかのクラスタを訂正するための訂正情報が入力された場合、当該入力された訂正情報で指定されたクラスタをターゲット又は非ターゲットに訂正する第5のステップと、
前記第2のステップで算出された特徴ベクトルのうち、前記第5のステップで訂正されたクラスタの特徴ベクトルと、当該クラスタに関する訂正結果を基に前記学習データを再学習して、新規学習データを生成する第6のステップとを実行し、
前記第3のステップでは、
前記第6のステップで前記新規学習データが生成された場合、前記第2のステップで算出された特徴ベクトルのうち、前記第5のステップで訂正されたクラスタの特徴ベクトルと、前記第6のステップで生成された新規学習データとを基に、前記第5のステップで訂正されたクラスタが、ターゲットであるか否かを識別し、当該識別結果を前記第4のステップに反映させることを特徴とする請求項6に記載のボリュームデータ解析方法。 - 前記第3のステップでは、
前記学習データとして、前記ターゲットの種類に応じた複数種類の学習データを用いる場合、前記複数種類の学習データと前記第2のステップで算出された各特徴ベクトルとを基に、前記各クラスタがターゲットであるか否かを識別すると共に、前記各識別結果に、当該識別結果に応じたラベルデータを付加し、
前記第4のステップでは、
前記表示装置に画像表示されたクラスタに、前記第3のステップで付加されたラベルデータに応じた識別ラベルを表示させることを特徴とする請求項6に記載のボリュームデータ解析方法。 - 前記第1のステップでは、
ドロネー三角形分割法又はマーチングキューブスアルゴリズムのいずれかを用いて、前記モダリティで生成された三次元ボリュームデータをポリゴン化してポリゴンデータに変換することを特徴とする請求項6に記載のボリュームデータ解析方法。 - 前記第3のステップでは、
前記機械学習に、サポートベクターマシン、マルチサポートベクターマシン、アダブースト、オンラインサポートベクターマシン、又はニューラルネットワークのうちいずれか一つを用いることを特徴とする請求項6に記載のボリュームデータ解析方法。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2012/084207 WO2014103061A1 (ja) | 2012-12-28 | 2012-12-28 | ボリュームデータ解析システム及びその方法 |
US14/655,512 US9589344B2 (en) | 2012-12-28 | 2012-12-28 | Volume data analysis system and method therefor |
JP2014554042A JP5982726B2 (ja) | 2012-12-28 | 2012-12-28 | ボリュームデータ解析システム及びその方法 |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2012/084207 WO2014103061A1 (ja) | 2012-12-28 | 2012-12-28 | ボリュームデータ解析システム及びその方法 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2014103061A1 true WO2014103061A1 (ja) | 2014-07-03 |
Family
ID=51020206
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2012/084207 WO2014103061A1 (ja) | 2012-12-28 | 2012-12-28 | ボリュームデータ解析システム及びその方法 |
Country Status (3)
Country | Link |
---|---|
US (1) | US9589344B2 (ja) |
JP (1) | JP5982726B2 (ja) |
WO (1) | WO2014103061A1 (ja) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2018537798A (ja) * | 2015-10-02 | 2018-12-20 | トラクタブル リミテッドTractable Ltd. | データセットの半自動ラベル付け |
JP2019530062A (ja) * | 2016-08-12 | 2019-10-17 | アキフィ,インコーポレイティド | メディアドキュメントのメタデータを自動的に生成するシステム及び方法 |
JPWO2019012632A1 (ja) * | 2017-07-12 | 2019-12-26 | 株式会社ソニー・インタラクティブエンタテインメント | 認識処理装置、認識処理方法及びプログラム |
JP2021033378A (ja) * | 2019-08-15 | 2021-03-01 | キヤノン株式会社 | 情報処理装置、情報処理方法 |
US11244438B2 (en) | 2020-01-03 | 2022-02-08 | Tractable Ltd | Auxiliary parts damage determination |
CN115018904A (zh) * | 2022-06-02 | 2022-09-06 | 如你所视(北京)科技有限公司 | 全景图像的掩膜生成方法和装置 |
JP7510782B2 (ja) | 2020-04-23 | 2024-07-04 | 合同会社H.U.グループ中央研究所 | プログラム、情報処理装置、情報処理方法及び学習モデルの生成方法 |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10249053B2 (en) * | 2016-08-04 | 2019-04-02 | The Boeing Company | Method and apparatus for rapid scanning of a three dimensional surface |
US20180114368A1 (en) * | 2016-10-25 | 2018-04-26 | Adobe Systems Incorporated | Three-dimensional model manipulation and rendering |
JP6961011B2 (ja) | 2016-12-09 | 2021-11-05 | ベイジン ホライズン インフォメーション テクノロジー カンパニー リミテッド | データ管理のためのシステム及び方法 |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2005052860A2 (en) * | 2003-11-19 | 2005-06-09 | Siemens Corporate Research, Inc. | System and method for detecting and matching anatomical structures using appearance and shape |
WO2007035340A2 (en) * | 2005-09-15 | 2007-03-29 | St. Jude Medical, Atrial Fibrillation Division, Inc. | Method of rendering a surface from a solid graphical image |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7239908B1 (en) * | 1998-09-14 | 2007-07-03 | The Board Of Trustees Of The Leland Stanford Junior University | Assessing the condition of a joint and devising treatment |
US7184814B2 (en) * | 1998-09-14 | 2007-02-27 | The Board Of Trustees Of The Leland Stanford Junior University | Assessing the condition of a joint and assessing cartilage loss |
WO2001010339A2 (en) * | 1999-08-11 | 2001-02-15 | Case Western Reserve University | Method and apparatus for producing an implant |
US20050089878A1 (en) * | 2003-02-14 | 2005-04-28 | Debe Derek A. | Method for determining functional sites in a protein |
US7558413B2 (en) | 2004-10-12 | 2009-07-07 | Siemens Medical Solutions USA, Inc, | Method for detecting polyps in a three dimensional image volume |
US7672497B2 (en) * | 2005-12-29 | 2010-03-02 | Carestream Health, Inc. | Computer aided disease detection system for multiple organ systems |
JP5226978B2 (ja) | 2007-07-17 | 2013-07-03 | 日立アロカメディカル株式会社 | 超音波診断装置及び画像処理プログラム |
US9171396B2 (en) * | 2010-06-30 | 2015-10-27 | Primal Space Systems Inc. | System and method of procedural visibility for interactive and broadcast streaming of entertainment, advertising, and tactical 3D graphical information using a visibility event codec |
-
2012
- 2012-12-28 US US14/655,512 patent/US9589344B2/en active Active
- 2012-12-28 JP JP2014554042A patent/JP5982726B2/ja not_active Expired - Fee Related
- 2012-12-28 WO PCT/JP2012/084207 patent/WO2014103061A1/ja active Application Filing
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2005052860A2 (en) * | 2003-11-19 | 2005-06-09 | Siemens Corporate Research, Inc. | System and method for detecting and matching anatomical structures using appearance and shape |
WO2007035340A2 (en) * | 2005-09-15 | 2007-03-29 | St. Jude Medical, Atrial Fibrillation Division, Inc. | Method of rendering a surface from a solid graphical image |
Non-Patent Citations (1)
Title |
---|
ANDRAS KELEMEN ET AL.: "Three-dimensional Model-based Segmentation of Brain MRI", PROCEEDINGS OF THE IEEE WORKSHOP ON BIOMEDICAL IMAGE ANALYSIS, 1998, pages 4 - 13, XP010291410, DOI: doi:10.1109/BIA.1998.692374 * |
Cited By (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2018537798A (ja) * | 2015-10-02 | 2018-12-20 | トラクタブル リミテッドTractable Ltd. | データセットの半自動ラベル付け |
JP7048499B2 (ja) | 2015-10-02 | 2022-04-05 | トラクタブル リミテッド | データセットの半自動ラベル付け |
JP2019530062A (ja) * | 2016-08-12 | 2019-10-17 | アキフィ,インコーポレイティド | メディアドキュメントのメタデータを自動的に生成するシステム及び方法 |
JPWO2019012632A1 (ja) * | 2017-07-12 | 2019-12-26 | 株式会社ソニー・インタラクティブエンタテインメント | 認識処理装置、認識処理方法及びプログラム |
US11170246B2 (en) | 2017-07-12 | 2021-11-09 | Sony Interactive Entertainment Inc. | Recognition processing device, recognition processing method, and program |
JP2021033378A (ja) * | 2019-08-15 | 2021-03-01 | キヤノン株式会社 | 情報処理装置、情報処理方法 |
JP7324086B2 (ja) | 2019-08-15 | 2023-08-09 | キヤノン株式会社 | 情報処理装置、情報処理方法 |
US11257203B2 (en) | 2020-01-03 | 2022-02-22 | Tractable Ltd | Inconsistent damage determination |
US11257204B2 (en) | 2020-01-03 | 2022-02-22 | Tractable Ltd | Detailed damage determination with image segmentation |
US11250554B2 (en) | 2020-01-03 | 2022-02-15 | Tractable Ltd | Repair/replace and labour hours determination |
US11361426B2 (en) | 2020-01-03 | 2022-06-14 | Tractable Ltd | Paint blending determination |
US11386543B2 (en) | 2020-01-03 | 2022-07-12 | Tractable Ltd | Universal car damage determination with make/model invariance |
US11587221B2 (en) | 2020-01-03 | 2023-02-21 | Tractable Limited | Detailed damage determination with image cropping |
US11636581B2 (en) | 2020-01-03 | 2023-04-25 | Tractable Limited | Undamaged/damaged determination |
US11244438B2 (en) | 2020-01-03 | 2022-02-08 | Tractable Ltd | Auxiliary parts damage determination |
JP7510782B2 (ja) | 2020-04-23 | 2024-07-04 | 合同会社H.U.グループ中央研究所 | プログラム、情報処理装置、情報処理方法及び学習モデルの生成方法 |
CN115018904A (zh) * | 2022-06-02 | 2022-09-06 | 如你所视(北京)科技有限公司 | 全景图像的掩膜生成方法和装置 |
CN115018904B (zh) * | 2022-06-02 | 2023-10-20 | 如你所视(北京)科技有限公司 | 全景图像的掩膜生成方法和装置 |
Also Published As
Publication number | Publication date |
---|---|
JP5982726B2 (ja) | 2016-08-31 |
US9589344B2 (en) | 2017-03-07 |
US20160005167A1 (en) | 2016-01-07 |
JPWO2014103061A1 (ja) | 2017-01-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP5982726B2 (ja) | ボリュームデータ解析システム及びその方法 | |
US10467757B2 (en) | System and method for computer aided diagnosis | |
JP5954769B2 (ja) | 医用画像処理装置、医用画像処理方法および異常検出プログラム | |
US10102622B2 (en) | Processing apparatus, processing method, and non-transitory computer-readable storage medium | |
JP5864542B2 (ja) | 画像の異常を検出するための画像データの処理方法、システムおよびプログラム | |
JP5334692B2 (ja) | 制約されたマージナル空間学習を使用して3d解剖学的オブジェクトを検出するための方法およびシステム | |
US9218542B2 (en) | Localization of anatomical structures using learning-based regression and efficient searching or deformation strategy | |
KR102204437B1 (ko) | 컴퓨터 보조 진단 방법 및 장치 | |
US8135189B2 (en) | System and method for organ segmentation using surface patch classification in 2D and 3D images | |
US20110262015A1 (en) | Image processing apparatus, image processing method, and storage medium | |
US9014456B2 (en) | Computer aided diagnostic system incorporating appearance analysis for diagnosing malignant lung nodules | |
US10319093B2 (en) | Image processing apparatus, image processing method, and program | |
EP2846310A2 (en) | Method and apparatus for registering medical images | |
JP2012155723A (ja) | 三次元医療映像から最適の二次元医療映像を自動的に生成する方法及び装置 | |
US10706534B2 (en) | Method and apparatus for classifying a data point in imaging data | |
Li et al. | Automated measurement network for accurate segmentation and parameter modification in fetal head ultrasound images | |
US10949698B2 (en) | Image processing apparatus, image processing system, image processing method, and storage medium | |
EP3893198A1 (en) | Method and system for computer aided detection of abnormalities in image data | |
Kumar et al. | Lung nodule segmentation using 3-dimensional convolutional neural networks | |
US20120232379A1 (en) | Method and System for Regression-Based 4D Mitral Valve Segmentation From 2D+t Magnetic Resonance Imaging Slices | |
KR20130016942A (ko) | 3차원 볼륨 파노라마 영상 생성 방법 및 장치 | |
JP2017189394A (ja) | 情報処理装置および情報処理システム | |
Chen et al. | Virtual blood vessels in complex background using stereo x-ray images | |
CN115661132A (zh) | 一种用于检测肺结节良恶性的计算机存储介质及装置 | |
JP2024509039A (ja) | 分類の視覚的説明、方法、システム |
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: 12890717 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2014554042 Country of ref document: JP Kind code of ref document: A |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 14655512 Country of ref document: US |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 12890717 Country of ref document: EP Kind code of ref document: A1 |