CN110517766A - Identify the method and device of encephalatrophy - Google Patents
Identify the method and device of encephalatrophy Download PDFInfo
- Publication number
- CN110517766A CN110517766A CN201910736591.1A CN201910736591A CN110517766A CN 110517766 A CN110517766 A CN 110517766A CN 201910736591 A CN201910736591 A CN 201910736591A CN 110517766 A CN110517766 A CN 110517766A
- Authority
- CN
- China
- Prior art keywords
- encephalatrophy
- graded index
- key frame
- key
- diameter
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
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
-
- 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
- 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
- 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
- G06T2207/10016—Video; Image sequence
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
- G06T2207/20152—Watershed 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30016—Brain
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Biomedical Technology (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Quality & Reliability (AREA)
- Radiology & Medical Imaging (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
- Apparatus For Radiation Diagnosis (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the invention provides the method and devices of identification encephalatrophy, it is related to field of artificial intelligence, this method comprises: determining the key frame in brain phantom sequence using key frame detection module, using the key point in critical point detection module detection key frame, and first kind graded index is determined according to the key point in key frame.Key frame is split using image segmentation module, determines the second class graded index.Encephalatrophy is identified further according to first kind graded index and the second class graded index later.Since the characteristics of being directed to different graded indexes is using different detection modes, to improve the detection accuracy of graded index, encephalatrophy is identified using first kind graded index and the second class graded index, also improves the precision of identification encephalatrophy.Secondly, for artificial hand dipping and calculating, manually being relied on small and high-efficient using neural network model automatic identification encephalatrophy.
Description
Technical field
The present embodiments relate to field of artificial intelligence, more particularly to the method and device of identification encephalatrophy.
Background technique
Encephalatrophy be brain tissue occur organic disease and compared with normal volume reduce Radiologic imaging, can by heredity, mind
Cause through many factors such as systemic disease, poisoning, malnutritions.Wherein, the most common are cerebral cortical atrophies, it is seen that gyrus becomes
It is flat, brain ditch is broadening, ventricles of the brain brain pond expands, brain weight mitigates, clinical manifestation be failure of memory, thinking ability decline, mood not
Stablize, can not focus on, would develop into dementia, aphasis when serious, lose intelligence etc..The whole world has about every year
15000000 people die of the relevant disease of various encephalatrophies, and the death rate increases year by year.The usual course of disease of encephalatrophy related disease is long, sends out
It is sick slow, therefore be not easy to be noticeable, once significant symptom occur just can not reverse, seriously affect the work and life of patient.Cause
This, early stage encephalatrophy diagnoses and treatment is to improving the survival rate of Brain Atrophy Patient, improve its quality of life and have important role.
Currently, the main method based on imaging diagnosis encephalatrophy has brain tissue volumetric measurement method and linear measurement method.Line
Property mensuration be the variation for reflecting encephalic cerebrospinal fluid volume by the measurement of the one-dimensional linear indexs of the ventricles of the brain, brain ditch, fissure, into
And reflect the variation of brain parenchym capacity indirectly.Its measuring point is clear, fixed, easy to implement the method, is clinically widely used.But
It is that this method relies on doctor and carries out hand dipping and calculating, subjectivity is strong and low efficiency.
Summary of the invention
Since the method currently based on imaging diagnosis encephalatrophy relies on, doctor carries out hand dipping and calculating, subjectivity are strong
And the problem of low efficiency, the embodiment of the invention provides the method and devices of identification encephalatrophy.
On the one hand, the embodiment of the invention provides a kind of methods for identifying encephalatrophy, comprising:
Key frame in brain phantom sequence is determined using key frame detection module;
Key point in the key frame is detected using critical point detection module, and according to the key point in the key frame
Determine first kind graded index;
The key frame is split using image segmentation module, determines the second class graded index;
Encephalatrophy is identified according to the first kind graded index and the second class graded index.
Optionally, the first kind graded index includes between anterior angle between maximum diameter, anterior angle between most path, choroid plexus of lateral ventricle
Outer diameter between diameter and telocoele top;
The key point detected using critical point detection module in the key frame, and according to the pass in the key frame
Key point determines first kind graded index, comprising:
Anterior angle key point, the telocoele key point in the key frame are detected using critical point detection module;
Most path between maximum diameter and the anterior angle is determined between the anterior angle according to the anterior angle key point;
Outer diameter between diameter and the telocoele top is determined between the choroid plexus of lateral ventricle according to the telocoele key point.
Optionally, the second class graded index includes the most wide diameter of three ventricles of the brain;
It is described that the key frame is split using image segmentation module, determine the second class graded index, comprising:
First area is determined according to the key point in the key frame;
Binary conversion treatment is carried out to the first area, determines second area;
The second area is split using image segmentation algorithm, determines three ventricle regions;
The most wide diameter of three ventricles of the brain is determined according to three ventricle region.
Optionally, the second class graded index includes skull maximum outside diameter and skull maximum inner diameter;
It is described that the key frame is split using image segmentation module, determine the second class graded index, comprising:
The key frame is split according to skull corresponding CT value, determines the first boundary;
First boundary is split using image segmentation algorithm, determines skull bone boundary;
The skull maximum outside diameter and the skull maximum inner diameter are determined according to the skull bone boundary.
It is optionally, described that encephalatrophy is identified according to the first kind graded index and the second class graded index, comprising:
Encephalatrophy assessment index is determined according to the first kind graded index and the second class graded index;
The encephalatrophy assessment index is inputted into encephalatrophy model, identifies encephalatrophy.
Optionally, the critical point detection module and the key frame detection module are convolutional neural networks.
On the one hand, the embodiment of the invention provides a kind of devices for identifying encephalatrophy, comprising:
Key frame detection module, for detecting the key frame in brain phantom sequence;
Critical point detection module, for detecting the key point in the key frame, and according to the key in the key frame
Point determines first kind graded index;
Image segmentation module determines the second class graded index for being split to the key frame;
Identification module, for identifying encephalatrophy according to the first kind graded index and the second class graded index.
Optionally, the first kind graded index includes between anterior angle between maximum diameter, anterior angle between most path, choroid plexus of lateral ventricle
Outer diameter between diameter and telocoele top;
The critical point detection module includes:
First detection module, for detecting the anterior angle key point in the key frame, telocoele key point;
First determining module, for being determined between the anterior angle between maximum diameter and the anterior angle most according to the anterior angle key point
Path;
Second determining module, for determining diameter and the side between the choroid plexus of lateral ventricle according to the telocoele key point
Outer diameter between ventricles of the brain top.
Optionally, the second class graded index includes the most wide diameter of three ventricles of the brain;
Described image divides module
Second detection module, for determining first area according to the key point in the key frame;
Third detection module determines second area for carrying out binary conversion treatment to the first area;
First segmentation module determines three ventricles of the brain areas for being split using image segmentation algorithm to the second area
Domain;
Third determining module, for determining the most wide diameter of three ventricles of the brain according to three ventricle region.
Optionally, the second class graded index includes skull maximum outside diameter and skull maximum inner diameter;
Described image divides module
Second segmentation module determines the first boundary for being split according to the corresponding CT value of skull to the key frame;
Third divides module, for being split using image segmentation algorithm to first boundary, determines skull bone boundary;
4th determining module, for determining that the skull maximum outside diameter and the skull are maximum according to the skull bone boundary
Internal diameter.
Optionally, the identification module includes:
5th determining module, for determining encephalatrophy according to the first kind graded index and the second class graded index
Assessment index;
6th determining module identifies encephalatrophy for the encephalatrophy assessment index to be inputted encephalatrophy model.
On the one hand, the embodiment of the invention provides a kind of computer equipment, including memory, processor and it is stored in storage
On device and the computer program that can run on a processor, the processor realize the side of identification encephalatrophy when executing described program
The step of method.
On the one hand, the embodiment of the invention provides a kind of computer readable storage medium, being stored with can be set by computer
The standby computer program executed, when described program is run on a computing device, so that the computer equipment executes identification
The step of method of encephalatrophy.
In the embodiment of the present invention, the key frame in brain phantom sequence is detected first, then detects the key in key frame
Point determines first kind graded index based on key point, the second graded index is determined by being split to key frame, for difference
The characteristics of graded index, is using different detection modes, to improve the detection accuracy of graded index.Referred to using first kind classification
Mark and the second class graded index identify encephalatrophy, also improve the precision of identification encephalatrophy.Secondly, certainly using neural network model
Dynamic identification encephalatrophy manually relies on small and high-efficient for artificial hand dipping and calculating.
Detailed description of the invention
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly introduced, it should be apparent that, drawings in the following description are only some embodiments of the invention, for this
For the those of ordinary skill in field, without any creative labor, it can also be obtained according to these attached drawings
His attached drawing.
Fig. 1 is a kind of flow diagram of method for identifying encephalatrophy provided in an embodiment of the present invention;
Fig. 2 is a kind of schematic diagram of brain phantom provided in an embodiment of the present invention;
Fig. 3 a is a kind of structural schematic diagram of key frame detection module provided in an embodiment of the present invention;
Fig. 3 b is the structural schematic diagram of rapid drop part in a kind of key frame detection module provided in an embodiment of the present invention;
Fig. 3 c is the structural schematic diagram of characteristic extraction part in a kind of key frame detection module provided in an embodiment of the present invention;
Fig. 3 d is the structural schematic diagram of feature extraction submodule in a kind of characteristic extraction part provided in an embodiment of the present invention;
Fig. 3 e is that the structure of Classification Neural part in a kind of key frame detection module provided in an embodiment of the present invention is shown
It is intended to;
Fig. 4 a is a kind of schematic diagram of key frame provided in an embodiment of the present invention;
Fig. 4 b is a kind of schematic diagram of key frame provided in an embodiment of the present invention;
Fig. 5 is a kind of flow diagram of method for detecting the second class graded index provided in an embodiment of the present invention;
Fig. 6 is a kind of schematic diagram of key frame provided in an embodiment of the present invention;
Fig. 7 is a kind of flow diagram of method for detecting the second class graded index provided in an embodiment of the present invention;
Fig. 8 is a kind of schematic diagram of key frame provided in an embodiment of the present invention;
Fig. 9 is a kind of flow diagram of the method for determining encephalatrophy disease compression levels provided in an embodiment of the present invention;
Figure 10 is a kind of structural schematic diagram of device for identifying encephalatrophy provided in an embodiment of the present invention;
Figure 11 is a kind of structural schematic diagram of computer equipment provided in an embodiment of the present invention.
Specific embodiment
In order to which the purpose of the present invention, technical solution and beneficial effect is more clearly understood, below in conjunction with attached drawing and implementation
Example, the present invention will be described in further detail.It should be appreciated that specific embodiment described herein is only used to explain this hair
It is bright, it is not intended to limit the present invention.
The method of identification encephalatrophy in the embodiment of the present invention can be applied to the scene of auxiliary diagnosis encephalatrophy, for example,
The brain CT images of patient are obtained, then using brain CT images of the method to patient for identifying encephalatrophy in the embodiment of the present invention
It is analyzed, exports the encephalatrophy recognition result of patient, doctor combines the recognition result of output to diagnose patient.
Based on above-mentioned application scenarios, the embodiment of the invention provides a kind of process of method for identifying encephalatrophy, this method
Process can be executed by the device of identification encephalatrophy, as shown in Figure 1, comprising the following steps:
Step S101 obtains brain phantom sequence.
Specifically, brain phantom sequence includes multiple brain phantoms, and the computerized tomography that brain phantom can be brain is taken the photograph
Shadow (Computed Tomography, abbreviation CT) image, nuclear magnetic resonance image etc..For example with CT images sequence, for giving
Fixed CT mode digital imaging and communications in medicine (Digital Imaging and Communications in Medicine,
Abbreviation DICOM) image sequence, each frame image information is read, interpolation zooms to fixed size (such as 512*512 pixel), and
It is adjusted to fixed window width and window level (brain window: W=80, L=40), obtains brain phantom sequence.Illustratively, brain phantom is specific
It can be as shown in Figure 2.
Step S102 determines the key frame in brain phantom sequence using key frame detection module.
Specifically, key frame detection module can be 2D convolutional neural networks (Convolutional Neural
Networks, abbreviation CNN) or 3D CNN.Key frame is predetermined for detecting the brain phantom of encephalatrophy, and key frame can
Think the maximum clearest level of Basal ganglia lenticular nucleus and two body of lateral ventricle display levels, key frame is one in brain phantom sequence
Frame or multiframe.
With key frame detection module be 2D CNN for example, introduce the network structure of key frame detection module first, have
Body is as shown in Figure 3a, and the network structure of key frame detection module includes: rapid drop part, characteristic extraction part, classification nerve
Network portion.
Rapid drop part is as shown in Figure 3b, by a convolutional layer, a batch normalization (Batch
Normalization, abbreviation BN) layer, activation primitive (Rectified Linear Unit, an abbreviation ReLU) layer and one
Pond layer is constituted.Convolutional layer convolution kernel size is 5*5, is divided into 2 pixels.Pond layer is the maximum value pond of 2*2, by fast
Speed reduces part and can reduce the area of brain phantom rapidly, and side length becomes original 1/4.
Characteristic extraction part is as shown in Figure 3c, is made of N number of feature extraction submodule, and N is the integer greater than 0.Each spy
It is as shown in Figure 3d to levy extracting sub-module, includes three bottleneck layers and a down-sampling layer.Bottleneck layer and down-sampling layer include three
A convolutional layer.
The feature for the characteristic pattern that bottleneck layer is exported rapid drop part by first convolutional layer and second convolutional layer
Map number is reduced, then the characteristic pattern number for the characteristic pattern that rapid drop part exports is increased back original spy by third convolutional layer
Figure number is levied, it is defeated after being then directly added the characteristic pattern that third convolutional layer exports with the characteristic pattern that rapid drop part exports
Out.
The characteristic pattern of rapid drop part output successively inputs down-sampling layer after three bottleneck layers carry out feature extraction.
The feature map number for the characteristic pattern that down-sampling layer is exported bottleneck layer by first convolutional layer and second convolutional layer is reduced, then
The characteristic pattern number for the characteristic pattern that bottleneck layer exports is increased into back original characteristic pattern number by third convolutional layer.Meanwhile third
Convolutional layer is while increasing characteristic pattern number, by the way that convolution step-length is set as 2, by the size reduction of characteristic pattern to half.Bottleneck
The characteristic pattern of layer output, by size reduction to original half, is finally exported third convolutional layer by the average pond of 2*2
Characteristic pattern exported with after the characteristic pattern that the bottleneck layer in average pond exports is added.
Classification Neural part is as shown in Figure 3 e, including an overall situation is averaged pond layer, a random inactivation
(dropout) layer, a full articulamentum, one softmax layers.It is defeated that the input of Classification Neural part is characterized extraction part
Characteristic pattern out exports the prediction classification for brain phantom.First by global average pond layer, characteristic pattern is extracted into one
Feature vector, then feature vector is inputted into dropout layers, full articulamentum and softmax layers, obtain a classification confidence to
Amount, each indicate a classification confidence level, and all confidence levels and be 1, export the highest position of confidence level as brain
The prediction classification of image.
The key frame in brain phantom sequence is set as 4 frames, wherein No. 1 key frame, No. 2 key frames and No. 3 key frames
For the maximum clearest display level of Basal ganglia lenticular nucleus, No. 4 key frames are that two body of lateral ventricle show level.In the above-mentioned key of training
When frame detection module, a large amount of brain CT images sequence is collected, then 4 frame of label is crucial from each CT images sequence by doctor
Frame.Then the CT images sequence that key frame is marked is expanded by data enhancement method, obtains training sample.Data increase
Strong mode includes: random to translate 0~20 pixel, Random-Rotation -20~~20 degree, 0.8~1.2 times of scaling at random up and down
Deng.Training sample is inputted in convolutional neural networks and is trained.In training, by the prediction classification of convolutional neural networks output
It is compared with the classification that training sample is marked, using cross entropy function as target loss function, and by reversed
Propagation algorithm is iterated using the optimal way of sgd, until objective function convergence, obtains key frame detection module.
When detecting the key frame in brain phantom sequence using key frame detection module, first against brain phantom sequence
In the 2nd frame to 2nd frame reciprocal every frame brain phantom, using every frame brain phantom as center frame, with each frame brain phantom in front and back
Splicing determines the brain phantom sequence including 3 frame brain phantoms.Then the brain phantom sequence including 3 frame brain phantoms is defeated
Entering key frame detection module, key frame detection module carries out 5 classification to the brain phantom sequence for including 3 frame brain phantoms and predicts,
Acquisition center frame is the confidence level of 5 classifications.5 category classifications are 0~4, wherein 0 indicates that center frame is not key frame, 1 table
Show that center frame is No. 1 key frame, 2 expression center frames are No. 2 key frames, and 3 expression center frames are No. 3 key frames, and 4 indicate center frame
It is No. 4 key frames, classification belonging to frame centered on the output maximum classification of confidence level.
It should be noted that when key frame detection module is 3D CNN, it can be for the 3rd frame in brain phantom sequence extremely
Every frame brain phantom of 3rd frame reciprocal splices with each two frames brain phantom in front and back, determines using every frame brain phantom as center frame
Brain phantom sequence including 5 frame brain phantoms.It then will include the brain phantom sequence inputting key frame inspection of 5 frame brain phantoms
5 classification prediction of survey module progress obtains center frame for the confidence level of 5 classifications, centered on the output maximum classification of confidence level
Classification belonging to frame.
In addition, key frame detection module is also possible to conventional machines learning model in addition to CNN, i.e., it is defeated to splice graphic sequence
Enter key frame detection module, key frame detection module calculates the gray feature and textural characteristics of the brain phantom in each channel, spells
It is connected into feature vector.Then using feature vector as input, divided using classifier (such as support vector machines, random forest)
Class obtains classification belonging to the frame of center.
Step S103, using the key point in critical point detection module detection key frame, and according to the key in key frame
Point determines first kind graded index.
Specifically, critical point detection module can be 2D CNN, comprising: rapid drop part, characteristic extraction part, classification
Part of neural network.
Rapid drop part is by a convolutional layer, a batch normalization (Batch Normalization, abbreviation BN)
Layer, activation primitive (Rectified Linear Unit, an abbreviation ReLU) layer and a pond layer are constituted.Convolutional layer convolution
Core size is 5*5, is divided into 2 pixels.Pond layer is the maximum value pond of 2*2, can will be crucial by rapid drop part
The area of frame reduces rapidly, and side length becomes original 1/4.
Characteristic extraction part is made of M feature extraction submodule, and M is the integer greater than 0.Each feature extraction submodule
Include three bottleneck layers and a down-sampling layer.Bottleneck layer and down-sampling layer include three convolutional layers.
The feature for the characteristic pattern that bottleneck layer is exported rapid drop part by first convolutional layer and second convolutional layer
Map number is reduced, then the characteristic pattern number for the characteristic pattern that rapid drop part exports is increased back original spy by third convolutional layer
Levy figure number.It is exported after the characteristic pattern that third convolutional layer exports directly is added with the characteristic pattern that rapid drop part exports.
The characteristic pattern of rapid drop part output successively inputs down-sampling layer after three bottleneck layers carry out feature extraction,
The feature map number for the characteristic pattern that down-sampling layer is exported bottleneck layer by first convolutional layer and second convolutional layer is reduced, then
The characteristic pattern number for the characteristic pattern that bottleneck layer exports is increased into back original characteristic pattern number by third convolutional layer.Meanwhile third
Convolutional layer is while increasing characteristic pattern number, by the way that convolution step-length is set as 2, by the size reduction of characteristic pattern to half.Bottleneck
The characteristic pattern of layer output, by size reduction to original half, is finally exported third convolutional layer by the average pond of 2*2
Characteristic pattern exported with after the characteristic pattern that the bottleneck layer in average pond exports is added.
Classification Neural part include an overall situation be averaged pond layer, one at random inactivate (dropout) layer, one entirely
Articulamentum, a linear transformation layer.The input of Classification Neural part is characterized the characteristic pattern for extracting part output, exports and is
Key point coordinate.First by global average pond layer, characteristic pattern is extracted into a feature vector, then feature vector is inputted
Dropout layers, full articulamentum and linear transformation layer, obtain a two-dimensional coordinate vector, and two-dimensional coordinate vector indicates key point in X
The position of axis and Y-axis.
In the above-mentioned critical point detection module of training, a large amount of brain CT images sequence is collected, by doctor in each CT shadow
As marking key frame in sequence, key point is then marked in key frame.By data enhancement method to be marked key frame and
The CT images sequence of key point is expanded, and training sample is obtained.Data enhancement method include: random translation 0 up and down~
20 pixels, Random-Rotation -20~~20 degree, scale 0.8~1.2 times etc. at random.Training sample is inputted in convolutional neural networks
It is trained.In training, by the key point coordinate that convolutional neural networks prediction exports and the key point that training sample is marked
Coordinate is compared, and using Mean Square Error (MSE) function as target loss function, and passes through back-propagation algorithm,
It using the optimal way of sgd, iterates, until objective function convergence, obtains critical point detection module.It is examined using key point
When surveying the key point in module detection key frame, key frame is inputted into critical point detection module, exports the key point in key frame
Coordinate.
Optionally, first kind graded index include between anterior angle between maximum diameter, anterior angle between most path, choroid plexus of lateral ventricle diameter and
Outer diameter between telocoele top.When detecting first kind graded index, closed using the anterior angle in critical point detection module detection key frame
Key point, telocoele key point.Then most path between maximum diameter and anterior angle is determined between anterior angle according to anterior angle key point, according to telocoele
Key point determines between choroid plexus of lateral ventricle outer diameter between diameter and telocoele top.
Illustratively, after setting key frame detection module detects brain phantom sequence, 4 frame key frames are determined, are closed
Key point detection module carries out critical point detection to 4 frame key frames respectively, determines the key point in every frame key frame.Setting wherein two
As shown in figures 4 a and 4b, anterior angle key point is the key point a in Fig. 4 a to frame key frame1, key point a2, key point b1, key point
b2, telocoele key point is the key point d in Fig. 4 a1, key point d2With the key point e in Fig. 4 b1, key point e2.For key
Point a1With key point a2, detect the key point a in every frame key frame1With key point a2The distance between, then compare 4 frames key
Key point a in frame1With key point a2The distance between size, maximum distance is determined as maximum diameter A between anterior angle.For key
Point b1With key point b2, detect the key point b in every frame key frame1With key point b2The distance between, then compare 4 frames key
Key point b in frame1With key point b2The distance between size, maximum distance is determined as between anterior angle most path B.For key
Point d1With key point d2, detect the key point d in every frame key frame1With key point d2The distance between, then compare 4 frames key
Key point d in frame1With key point d2The distance between size, maximum distance is determined as diameter D between choroid plexus of lateral ventricle.For
Key point e1With key point e2, detect the key point e in every frame key frame1With key point e2The distance between, then compare 4 frames
Key point e in key frame1With key point e2The distance between size, maximum distance is determined as outer diameter E between telocoele top.
By detecting the key point in each key frame, then the distance between key point is determined in more each frame key frame
First kind graded index is determined for first kind graded index compared to based on the key point in single frames key frame, improves inspection
Survey precision.
Step S104 is split key frame using image segmentation module, determines the second class graded index.
In a kind of possible embodiment, the second class graded index includes the most wide diameter of three ventricles of the brain, detection the second class classification
Index includes the following steps, as shown in Figure 5:
Step S501 determines first area according to the key point in key frame.
Specifically, three brains can be then based on using three ventricles of the brain key points in critical point detection module detection key frame
Room key point and anterior angle key point determine first area.Illustratively, as shown in fig. 6, three ventricles of the brain key points are pass shown in fig. 6
Key point c1With key point c2, in conjunction with key point b1, key point b2, key point c1With key point c2Determine first area.
Step S502 carries out binary conversion treatment to first area, determines second area.
Specifically, for pixel each in first area, when pixel intensity is greater than preset threshold, then by the pixel
It is determined as a part of three ventricles of the brain, otherwise which is determined as to a part of background, pixel intensity is big in first area
Second area is formed in the pixel of preset threshold.
Step S503 is split second area using image segmentation algorithm, determines three ventricle regions.
In specific implementation, image segmentation algorithm includes the dividing method based on threshold value, the dividing method based on edge, is based on
The dividing method in region, dividing method based on specific theory etc..
The basic thought of dividing method based on threshold value is the gray feature based on image to calculate one or more gray scales
Threshold value, and the gray value of pixel each in image is compared with threshold value, pixel is finally assigned into conjunction according to comparison result
In suitable classification.Therefore, the most key step of such method is exactly that optimum gradation threshold value is solved according to some criterion function.
Dividing method based on edge refers to the set of continuous pixel on the boundary line of two different zones in image,
It is the reflection of image local feature discontinuity, embodies the mutation of the picture characteristics such as gray scale, color, texture.Under normal conditions,
Dividing method based on edge refers to the edge detection based on gray value, it, which is built upon edge gray value, can show step
Type or roof type change the method on the basis of this observation.
Dividing method based on region is that image is divided into different regions according to similarity criterion, mainly includes seed zone
The several types such as domain growth method, regional split act of union and watershed method.Wherein, watershed method is a kind of based on topological theory
The dividing method of mathematical morphology, basic thought are topological landforms image regarded as in geodesy, each in image
The gray value of pixel indicates the height above sea level of the point, each local minimum and its influence area are known as reception basin, and collect
The boundary of basin then forms watershed.The realization of the algorithm can be modeled to the process of flood inundation on tracks, and the minimum point of image is first
It is submerged, then water gradually floods entire mountain valley.It will overflow when water level reaches certain altitude, at this moment be overflowed in water
Dykes and dams are built in place, repeat this process until the pixel in whole image is all submerged, that is at this moment established is a series of
Dykes and dams just become the watershed in separately each basin.Watershed algorithm has good response to faint edge, but in image
Noise can make watershed algorithm generate over-segmentation the phenomenon that.
Step S504 determines the most wide diameter of three ventricles of the brain according to three ventricle regions.
Specifically, the maximum width of three ventricle regions is determined as the most wide diameter C of three ventricles of the brain.In a kind of possible embodiment
In, when key frame is multiframe, three ventricle regions in every frame key frame can be detected using the above method, and determine that every frame is crucial
The maximum width of three ventricle regions in frame.Then according to sequence from big to small by the maximum of three ventricle regions in each frame key frame
Width is ranked up, and the maximum width for coming first is determined as the most wide diameter of three ventricles of the brain.By combining critical point detection and image
Partitioning algorithm detects the second class graded index, effectively improves the precision of the second class graded index of detection.
In a kind of possible embodiment, the second class graded index includes skull maximum outside diameter and skull most imperial palace
Diameter, the second class graded index of detection includes the following steps, as shown in Figure 7:
Step S701 is split key frame according to the corresponding CT value of skull, determines the first boundary.
Specifically, the corresponding different CT value of the tissue of different densities, the corresponding CT value of skull is generally higher than 400HU, close
Degree is generally higher than its hetero-organization of brain, therefore the corresponding CT value of skull can be taken to be split key frame, filters out other groups of brain
It knits, obtains the first boundary, the first boundary includes at least the inner boundary of skull and the outer boundary of skull.
Step S702 is split the first boundary using image segmentation algorithm, determines skull bone boundary.
In specific implementation, image segmentation algorithm includes the dividing method based on threshold value, the dividing method based on edge, is based on
The dividing method in region, dividing method based on specific theory etc..
Step S703 determines skull maximum outside diameter and skull maximum inner diameter according to skull bone boundary.
Illustratively, the skull bone boundary in key frame is set as shown in figure 8, then skull maximum outside diameter is distance F, and skull is most
Large diameter is distance G.It, can be every using above method detection when key frame is multiframe in a kind of possible embodiment
Then skull bone boundary in frame key frame determines skull maximum outside diameter and skull maximum inner diameter in every frame key frame.Later again
The size for comparing the skull maximum outside diameter in each key frame, using maximum skull maximum outside diameter as the second class graded index.Than
The size of skull maximum inner diameter in more each key frame, using maximum skull maximum inner diameter as the second class graded index.Due to
The density of skull is greater than its hetero-organization of brain, therefore when being split using CT value to key frame, accurate skull can be obtained
The first boundary, then be split using image segmentation algorithm, the exact boundary of skull obtained, later again based on the accurate of skull
Boundary determines skull maximum outside diameter and skull maximum inner diameter, effectively improves the precision of the second class graded index of detection.
Step S105 identifies encephalatrophy according to first kind graded index and the second class graded index.
Optionally, it when identifying encephalatrophy based on first kind graded index and the second class graded index, specifically includes following
Step, as shown in Figure 9:
Step S901 determines encephalatrophy assessment index according to first kind graded index and the second class graded index.
Specifically, encephalatrophy assessment index includes that Kazakhstan value, ventricular index, body of lateral cerebral ventricle index, body of lateral cerebral ventricle are wide
Spend index, anterior angle index, diacele width.
Kazakhstan value most the sum of path between maximum diameter and anterior angle between anterior angle, in general, the normal Kazakhstan value range of male
It is 3~6.9, the Kazakhstan value range of female normal is 2.6~5.2.
The ratio of ventricular index maximum diameter between diameter and anterior angle between choroid plexus of lateral ventricle, in general, the normal brain of male
Room index range is 1.1~3.3, and the ventricular index range of female normal is 1.1~2.9.
The ratio of body of lateral cerebral ventricle index outer diameter between skull maximum outside diameter and telocoele top, in general, male is normal
Body of lateral cerebral ventricle index range be 4.3~7.4, the body of lateral cerebral ventricle index range of female normal is 3.9~7.7.
The ratio of body of lateral cerebral ventricle breadth index outer diameter between skull maximum inner diameter and telocoele top, in general, male
Normal body of lateral cerebral ventricle breadth index range is 3.1~6.7, and the body of lateral cerebral ventricle breadth index range of female normal is 3.5
~6.8.
The ratio of anterior angle index maximum diameter between skull maximum inner diameter and anterior angle, in general, the normal anterior angle of male refer to
Number range is 2.8~8.2, and the anterior angle index range of female normal is 3.0~8.5.
In general, the normal diacele width range of male is 1~6.7, the diacele width model of female normal
Enclose is 0~7.
Encephalatrophy assessment index is inputted encephalatrophy model, identifies encephalatrophy by step S902.
Specifically, encephalatrophy model can be served only for identifying whether there is encephalatrophy, can be used for identifying whether there is encephalatrophy disease
Contracting and encephalatrophy disease compression levels.
When whether encephalatrophy model has encephalatrophy for identification, encephalatrophy model can be Logic Regression Models, pattra leaves
This model etc..
In a kind of possible embodiment, encephalatrophy model is Logic Regression Models, which specifically meets following public affairs
Formula (1):
y1=a1x1+a2x2+a3x3+a4x4+a5x5+a6x6……………(1)
Wherein, y1For brain atrophy value, xiFor encephalatrophy assessment index, i=1,2,3,4,5,6, ajFor weighting coefficient, j
=1,2,3,4,5,6,0≤aj≤1。
As brain atrophy value y1When greater than first threshold, determination has encephalatrophy, as brain atrophy value y1No more than first
When threshold value, determine without encephalatrophy.
In a kind of possible embodiment, encephalatrophy model can be Bayesian model, which specifically meets following
Formula (2):
Wherein, y1For encephalatrophy classification, xiFor encephalatrophy assessment index, i=1,2,3,4,5,6, CkFor class items, C0, C1
For specific classification, being divided into has encephalatrophy and without encephalatrophy, can indicate encephalatrophy with 1,0 indicates no encephalatrophy.
When whether encephalatrophy model has encephalatrophy and encephalatrophy disease compression levels for identification, encephalatrophy model includes encephalatrophy
Determining module and encephalatrophy diversity module determine whether encephalatrophy using encephalatrophy determining module first.There is encephalatrophy disease in determination
When contracting, encephalatrophy disease compression levels are further determined that using encephalatrophy diversity module.Encephalatrophy determining module can for Logic Regression Models,
Bayesian model etc., encephalatrophy diversity module can be Logic Regression Models, Bayesian model etc..
In a kind of possible embodiment, encephalatrophy determining module is Logic Regression Models, which specifically meets
Formula (1) is stated, as brain atrophy value y1When greater than first threshold, determination has encephalatrophy, as brain atrophy value y1No more than
When one threshold value, determine without encephalatrophy.
When determining has encephalatrophy, encephalatrophy disease compression levels are determined using encephalatrophy diversity module, wherein encephalatrophy diversity module
For Logic Regression Models, which specifically meets following formula (3):
y2=b1x1+b2x2+b3x3+b4x4+b5x5+b6x6……………(3)
Wherein, y2For encephalatrophy rank value, xiFor encephalatrophy assessment index, i=1,2,3,4,5,6, bkFor weighting coefficient, k
=1,2,3,4,5,6,0≤bk≤1。
The relationship table between encephalatrophy rank value and encephalatrophy disease compression levels is preset, it is true using encephalatrophy diversity module
After determining encephalatrophy rank value, encephalatrophy disease compression levels can be directly obtained by inquiring the table of comparisons.
In a kind of possible embodiment, encephalatrophy determining module can be Bayesian model, which specifically meets
Above-mentioned formula (2).
When determining has encephalatrophy, encephalatrophy disease compression levels are determined using encephalatrophy diversity module, wherein encephalatrophy diversity module
For Bayesian model, which specifically meets following formula (4):
Wherein, y2For encephalatrophy disease compression levels, xiFor encephalatrophy assessment index, i=1,2,3,4,5,6, DkFor class items, D0, D1,
D2For specific classification, it is divided into slight encephalatrophy, moderate encephalatrophy and severe encephalatrophy, can indicates slight encephalatrophy with 0,
1 indicates moderate encephalatrophy, and 2 indicate severe encephalatrophy.
In the embodiment of the present invention, the key frame in brain phantom sequence is detected first, then detects the key in key frame
Point determines first kind graded index based on key point, the second graded index is determined by being split to key frame, for difference
The characteristics of graded index, is using different detection modes, to improve the detection accuracy of graded index.Referred to using first kind classification
Mark and the second class graded index identification encephalatrophy and determining encephalatrophy disease compression levels also improve identification encephalatrophy and determine encephalatrophy disease
The precision of division grade.Secondly, using neural network model automatic identification encephalatrophy and encephalatrophy disease compression levels are determined, compared to artificial
For hand dipping and calculating, manually rely on small and high-efficient.
Based on the same technical idea, the embodiment of the invention provides a kind of devices for identifying encephalatrophy, as shown in Figure 10,
The device can execute the process of the method for identification encephalatrophy, which includes:
Key frame detection module 1001, for detecting the key frame in brain phantom sequence;
Critical point detection module 1002, for detecting the key point in the key frame, and according in the key frame
Key point determines first kind graded index;
Image segmentation module 1003 determines the second class graded index for being split to the key frame;
Identification module 1004, for identifying encephalatrophy disease according to the first kind graded index and the second class graded index
Contracting.
Optionally, the first kind graded index includes between anterior angle between maximum diameter, anterior angle between most path, choroid plexus of lateral ventricle
Outer diameter between diameter and telocoele top;
The key frame detection module 1001 includes:
First detection module, for detecting the anterior angle key point in the key frame, telocoele key point;
First determining module, for being determined between the anterior angle between maximum diameter and the anterior angle most according to the anterior angle key point
Path;
Second determining module, for determining diameter and the side between the choroid plexus of lateral ventricle according to the telocoele key point
Outer diameter between ventricles of the brain top.
Optionally, the second class graded index includes the most wide diameter of three ventricles of the brain;
Described image divides module 1003
Second detection module, for determining first area according to the key point in the key frame;
Third detection module determines second area for carrying out binary conversion treatment to the first area;
First segmentation module determines three ventricles of the brain areas for being split using image segmentation algorithm to the second area
Domain;
Third determining module, for determining the most wide diameter of three ventricles of the brain according to three ventricle region.
Optionally, the second class graded index includes skull maximum outside diameter and skull maximum inner diameter;
Described image divides module 1003
Second segmentation module determines the first boundary for being split according to the corresponding CT value of skull to the key frame;
Third divides module, for being split using image segmentation algorithm to first boundary, determines skull bone boundary;
4th determining module, for determining that the skull maximum outside diameter and the skull are maximum according to the skull bone boundary
Internal diameter.
Optionally, the identification module 1004 includes:
5th determining module, for determining encephalatrophy according to the first kind graded index and the second class graded index
Assessment index;
6th determining module identifies encephalatrophy for the encephalatrophy assessment index to be inputted encephalatrophy model.
Optionally, the critical point detection module 1002 and the key frame detection module 1001 are convolutional neural networks.
Based on the same technical idea, the embodiment of the invention provides a kind of computer equipments, as shown in figure 11, including extremely
Lack a processor 1101, and the memory 1102 connecting at least one processor, does not limit place in the embodiment of the present invention
The specific connection medium between device 1101 and memory 1102 is managed, is passed through between processor 1101 and memory 1102 in Figure 11 total
For line connection.Bus can be divided into address bus, data/address bus, control bus etc..
In embodiments of the present invention, memory 1102 is stored with the instruction that can be executed by least one processor 1101, until
The instruction that a few processor 1101 is stored by executing memory 1102 can execute in the method above-mentioned for identifying encephalatrophy
Included step.
Wherein, processor 1101 is the control centre of computer equipment, can use various interfaces and connection calculates
The various pieces of machine equipment are stored in memory by running or executing the instruction being stored in memory 1102 and calling
Data in 1102, to identify encephalatrophy.Optionally, processor 1101 may include one or more processing units, processor
1101 can integrate application processor and modem processor, wherein the main processing operation system of application processor, user interface
With application program etc., modem processor mainly handles wireless communication.It is understood that above-mentioned modem processor
It can not be integrated into processor 1101.In some embodiments, processor 1101 and memory 1102 can be in same chips
Upper realization, in some embodiments, they can also be realized respectively on independent chip.
Processor 1101 can be general processor, such as central processing unit (CPU), digital signal processor, dedicated collection
At circuit (Application Specific Integrated Circuit, ASIC), field programmable gate array or other
Perhaps transistor logic, discrete hardware components may be implemented or execute the present invention in fact for programmable logic device, discrete gate
Apply each method, step disclosed in example and logic diagram.General processor can be microprocessor or any conventional processing
Device etc..The step of method in conjunction with disclosed in the embodiment of the present invention, can be embodied directly in hardware processor and execute completion, or
With in processor hardware and software module combination execute completion.
Memory 1102 is used as a kind of non-volatile computer readable storage medium storing program for executing, can be used for storing non-volatile software journey
Sequence, non-volatile computer executable program and module.Memory 1102 may include the storage medium of at least one type,
It such as may include flash memory, hard disk, multimedia card, card-type memory, random access storage device (Random Access
Memory, RAM), static random-access memory (Static Random Access Memory, SRAM), may be programmed read-only deposit
Reservoir (Programmable Read Only Memory, PROM), read-only memory (Read Only Memory, ROM), band
Electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory,
EEPROM), magnetic storage, disk, CD etc..Memory 1102 can be used for carrying or storing have instruction or data
The desired program code of structure type and can by any other medium of computer access, but not limited to this.The present invention is real
Applying the memory 1102 in example can also be circuit or other devices that arbitrarily can be realized store function, for storing program
Instruction and/or data.
Based on the same technical idea, it the embodiment of the invention provides a kind of computer readable storage medium, is stored with
The computer program that can be executed by computer equipment, when described program is run on a computing device, so that the computer
Equipment executes the step of method of identification encephalatrophy.
It should be understood by those skilled in the art that, the embodiment of the present invention can provide as method or computer program product.
Therefore, complete hardware embodiment, complete software embodiment or embodiment combining software and hardware aspects can be used in the present invention
Form.It is deposited moreover, the present invention can be used to can be used in the computer that one or more wherein includes computer usable program code
The shape for the computer program product implemented on storage media (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.)
Formula.
The present invention be referring to according to the method for the embodiment of the present invention, the process of equipment (system) and computer program product
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
Although preferred embodiments of the present invention have been described, it is created once a person skilled in the art knows basic
Property concept, then additional changes and modifications may be made to these embodiments.So it includes excellent that the following claims are intended to be interpreted as
It selects embodiment and falls into all change and modification of the scope of the invention.
Obviously, various changes and modifications can be made to the invention without departing from essence of the invention by those skilled in the art
Mind and range.In this way, if these modifications and changes of the present invention belongs to the range of the claims in the present invention and its equivalent technologies
Within, then the present invention is also intended to include these modifications and variations.
Claims (10)
1. a kind of method for identifying encephalatrophy characterized by comprising
Key frame in brain phantom sequence is determined using key frame detection module;
Key point in the key frame is detected using critical point detection module, and is determined according to the key point in the key frame
First kind graded index;
The key frame is split using image segmentation module, determines the second class graded index;
Encephalatrophy is identified according to the first kind graded index and the second class graded index.
2. the method as described in claim 1, which is characterized in that the first kind graded index includes maximum diameter between anterior angle, preceding
Outer diameter between diameter and telocoele top between most path, choroid plexus of lateral ventricle between angle;
The key point detected using critical point detection module in the key frame, and according to the key point in the key frame
Determine first kind graded index, comprising:
Anterior angle key point, the telocoele key point in the key frame are detected using critical point detection module;
Most path between maximum diameter and the anterior angle is determined between the anterior angle according to the anterior angle key point;
Outer diameter between diameter and the telocoele top is determined between the choroid plexus of lateral ventricle according to the telocoele key point.
3. the method as described in claim 1, which is characterized in that the second class graded index includes the most wide diameter of three ventricles of the brain;
It is described that the key frame is split using image segmentation module, determine the second class graded index, comprising:
First area is determined according to the key point in the key frame;
Binary conversion treatment is carried out to the first area, determines second area;
The second area is split using image segmentation algorithm, determines three ventricle regions;
The most wide diameter of three ventricles of the brain is determined according to three ventricle region.
4. the method as described in claim 1, which is characterized in that the second class graded index include skull maximum outside diameter and
Skull maximum inner diameter;
It is described that the key frame is split using image segmentation module, determine the second class graded index, comprising:
The key frame is split according to skull corresponding CT value, determines the first boundary;
First boundary is split using image segmentation algorithm, determines skull bone boundary;
The skull maximum outside diameter and the skull maximum inner diameter are determined according to the skull bone boundary.
5. the method as described in Claims 1-4 is any, which is characterized in that described according to the first kind graded index and institute
State the second class graded index identification encephalatrophy, comprising:
Encephalatrophy assessment index is determined according to the first kind graded index and the second class graded index;
The encephalatrophy assessment index is inputted into encephalatrophy model, identifies encephalatrophy.
6. the method as described in claim 1, which is characterized in that the critical point detection module and the key frame detection module
For convolutional neural networks.
7. a kind of device for identifying encephalatrophy characterized by comprising
Key frame detection module, for detecting the key frame in brain phantom sequence;
Critical point detection module, for detecting the key point in the key frame, and it is true according to the key point in the key frame
Determine first kind graded index;
Image segmentation module determines the second class graded index for being split to the key frame;
Identification module, for identifying encephalatrophy according to the first kind graded index and the second class graded index.
8. device as claimed in claim 7, which is characterized in that the first kind graded index includes maximum diameter between anterior angle, preceding
Outer diameter between diameter and telocoele top between most path, choroid plexus of lateral ventricle between angle;
The critical point detection module includes:
First detection module, for detecting the anterior angle key point in the key frame, telocoele key point;
First determining module, it is minimum between maximum diameter and the anterior angle for being determined between the anterior angle according to the anterior angle key point
Diameter;
Second determining module, for determining diameter and the telocoele between the choroid plexus of lateral ventricle according to the telocoele key point
Outer diameter between top.
9. a kind of computer equipment including memory, processor and stores the meter that can be run on a memory and on a processor
Calculation machine program, which is characterized in that the processor is realized described in claim 1~6 any claim when executing described program
The step of method.
10. a kind of computer readable storage medium, which is characterized in that it is stored with the computer journey that can be executed by computer equipment
Sequence, when described program is run on a computing device, so that computer equipment perform claim requirement 1~6 is any described
The step of method.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910736591.1A CN110517766B (en) | 2019-08-09 | 2019-08-09 | Method and device for identifying brain atrophy |
PCT/CN2019/130863 WO2021027240A1 (en) | 2019-08-09 | 2019-12-31 | Brain atrophy identification method, and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910736591.1A CN110517766B (en) | 2019-08-09 | 2019-08-09 | Method and device for identifying brain atrophy |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110517766A true CN110517766A (en) | 2019-11-29 |
CN110517766B CN110517766B (en) | 2020-10-16 |
Family
ID=68625476
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910736591.1A Active CN110517766B (en) | 2019-08-09 | 2019-08-09 | Method and device for identifying brain atrophy |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN110517766B (en) |
WO (1) | WO2021027240A1 (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111462055A (en) * | 2020-03-19 | 2020-07-28 | 沈阳先进医疗设备技术孵化中心有限公司 | Skull detection method and device |
CN111862014A (en) * | 2020-07-08 | 2020-10-30 | 深圳市第二人民医院(深圳市转化医学研究院) | ALVI automatic measurement method and device based on left and right ventricle segmentation |
WO2021027240A1 (en) * | 2019-08-09 | 2021-02-18 | 上海依智医疗技术有限公司 | Brain atrophy identification method, and device |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050014821A1 (en) * | 2002-07-24 | 2005-01-20 | President And Fellows Of Harvard College | Compounds and methods for treating neurodegenerative disorders |
US20080192049A1 (en) * | 2007-02-13 | 2008-08-14 | Thomas Schiwietz | Image deformation using physical models |
US9558396B2 (en) * | 2013-10-22 | 2017-01-31 | Samsung Electronics Co., Ltd. | Apparatuses and methods for face tracking based on calculated occlusion probabilities |
CN107103612A (en) * | 2017-03-28 | 2017-08-29 | 深圳博脑医疗科技有限公司 | Automate the quantitative calculation method of subregion brain atrophy |
CN109389002A (en) * | 2017-08-02 | 2019-02-26 | 阿里巴巴集团控股有限公司 | Biopsy method and device |
CN109509211A (en) * | 2018-09-28 | 2019-03-22 | 北京大学 | Positioning simultaneously and the feature point extraction and matching process and system built in diagram technology |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10064594B2 (en) * | 2015-08-06 | 2018-09-04 | Case Western Reserve University | Characterizing disease and treatment response with quantitative vessel tortuosity radiomics |
CN105844617A (en) * | 2016-03-17 | 2016-08-10 | 电子科技大学 | Brain parenchyma segmentation realization based on improved threshold segmentation algorithm |
CN109214451A (en) * | 2018-08-28 | 2019-01-15 | 北京安德医智科技有限公司 | A kind of classification method and equipment of brain exception |
CN109389585B (en) * | 2018-09-20 | 2021-11-02 | 东南大学 | Brain tissue extraction method based on full convolution neural network |
CN109509177B (en) * | 2018-10-22 | 2021-02-23 | 杭州依图医疗技术有限公司 | Method and device for recognizing brain image |
CN110517766B (en) * | 2019-08-09 | 2020-10-16 | 上海依智医疗技术有限公司 | Method and device for identifying brain atrophy |
-
2019
- 2019-08-09 CN CN201910736591.1A patent/CN110517766B/en active Active
- 2019-12-31 WO PCT/CN2019/130863 patent/WO2021027240A1/en active Application Filing
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050014821A1 (en) * | 2002-07-24 | 2005-01-20 | President And Fellows Of Harvard College | Compounds and methods for treating neurodegenerative disorders |
US20080192049A1 (en) * | 2007-02-13 | 2008-08-14 | Thomas Schiwietz | Image deformation using physical models |
US9558396B2 (en) * | 2013-10-22 | 2017-01-31 | Samsung Electronics Co., Ltd. | Apparatuses and methods for face tracking based on calculated occlusion probabilities |
CN107103612A (en) * | 2017-03-28 | 2017-08-29 | 深圳博脑医疗科技有限公司 | Automate the quantitative calculation method of subregion brain atrophy |
CN109389002A (en) * | 2017-08-02 | 2019-02-26 | 阿里巴巴集团控股有限公司 | Biopsy method and device |
CN109509211A (en) * | 2018-09-28 | 2019-03-22 | 北京大学 | Positioning simultaneously and the feature point extraction and matching process and system built in diagram technology |
Non-Patent Citations (3)
Title |
---|
ANNUAL REVIEW OF BIOMEDICAL ENGINEERING: "Deep Learning in Medical Image Analysis", 《ANNUAL REVIEW OF BIOMEDICAL ENGINEERING》 * |
杨震卢虹冰: "3DsMax软件用于人体软组织器官形变模拟的研究", 《医疗卫生装备》 * |
陈新: "脑萎缩率定量测量的研究", 《中国医疗器诫杂志》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2021027240A1 (en) * | 2019-08-09 | 2021-02-18 | 上海依智医疗技术有限公司 | Brain atrophy identification method, and device |
CN111462055A (en) * | 2020-03-19 | 2020-07-28 | 沈阳先进医疗设备技术孵化中心有限公司 | Skull detection method and device |
CN111462055B (en) * | 2020-03-19 | 2024-03-08 | 东软医疗***股份有限公司 | Skull detection method and device |
CN111862014A (en) * | 2020-07-08 | 2020-10-30 | 深圳市第二人民医院(深圳市转化医学研究院) | ALVI automatic measurement method and device based on left and right ventricle segmentation |
Also Published As
Publication number | Publication date |
---|---|
CN110517766B (en) | 2020-10-16 |
WO2021027240A1 (en) | 2021-02-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111798462B (en) | Automatic delineation method of nasopharyngeal carcinoma radiotherapy target area based on CT image | |
CN105574859B (en) | A kind of liver neoplasm dividing method and device based on CT images | |
CN109598727B (en) | CT image lung parenchyma three-dimensional semantic segmentation method based on deep neural network | |
CN110706246B (en) | Blood vessel image segmentation method and device, electronic equipment and storage medium | |
CN105957063B (en) | CT image liver segmentation method and system based on multiple dimensioned weighting similarity measure | |
Nambakhsh et al. | Left ventricle segmentation in MRI via convex relaxed distribution matching | |
CN109035252A (en) | A kind of super-pixel method towards medical image segmentation | |
Kavitha et al. | An efficient approach for brain tumour detection based on modified region growing and neural network in MRI images | |
CN110517766A (en) | Identify the method and device of encephalatrophy | |
CN109389062B (en) | Method for extracting lake water and land dividing line by using high-resolution satellite-borne SAR image | |
Rodriguez et al. | Optical fish trajectory measurement in fishways through computer vision and artificial neural networks | |
CN109087703A (en) | Abdominal cavity CT image peritonaeum metastatic marker method based on depth convolutional neural networks | |
CN109685804B (en) | Multi-channel head magnetic resonance imaging tissue segmentation method | |
US20210248747A1 (en) | Organs at risk auto-contouring system and methods | |
CN108765427A (en) | A kind of prostate image partition method | |
Ontiveros et al. | Computer tomography 3D edge detection comparative for metrology applications | |
CN113989551A (en) | Alzheimer disease classification method based on improved ResNet network | |
Kitrungrotsakul et al. | Interactive deep refinement network for medical image segmentation | |
CN111798463B (en) | Method for automatically segmenting multiple organs in head and neck CT image | |
CN114862799B (en) | Full-automatic brain volume segmentation method for FLAIR-MRI sequence | |
CN115797378A (en) | Prostate contour segmentation method based on geometric intersection ratio loss | |
Samudrala et al. | Semantic Segmentation in Medical Image Based on Hybrid Dlinknet and Unet | |
CN108460383A (en) | Saliency refined method based on neural network and image segmentation | |
CN114529818A (en) | Cultivated land plot extraction method and system | |
Rodriguez et al. | Blood vessel segmentation via neural network in histological images |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20220322 Address after: 100080 zone a, 21 / F, block a, No. 8, Haidian Street, Haidian District, Beijing Patentee after: BEIJING SHENRUI BOLIAN TECHNOLOGY Co.,Ltd. Patentee after: Hangzhou Shenrui Bolian Technology Co., Ltd Address before: Units 06 and 07, 23rd Floor, 523 Loushanguan Road, Changning District, Shanghai, 2003 Patentee before: SHANGHAI YIZHI MEDICAL TECHNOLOGY Co.,Ltd. |