CN109685796A - Medical image processing method, device, equipment and storage medium - Google Patents
Medical image processing method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN109685796A CN109685796A CN201811596856.4A CN201811596856A CN109685796A CN 109685796 A CN109685796 A CN 109685796A CN 201811596856 A CN201811596856 A CN 201811596856A CN 109685796 A CN109685796 A CN 109685796A
- Authority
- CN
- China
- Prior art keywords
- calculus
- central area
- network model
- neural network
- region
- 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
- 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]
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The present invention relates to a kind of medical image processing method, device, equipment and storage mediums.Terminal is by obtaining CT scan image, and obtain CT scan image is labeled after obtained calculus roughing region, and then calculus central area is extracted from calculus roughing region, calculus central area is inputted into preset neural network model, exports calculus types results.In the application, the calculus roughing region that terminal obtains after being labeled by acquisition to CT scan image obtains calculus central area, and calculus central area is inputted into preset neural network model, export calculus types results, so that during obtaining calculus types results, it is to avoid by carrying out deep learning acquisition to calculus central area and introduce non-calculus region during obtaining calculus types results, so that the calculus types results obtained are more accurate.
Description
Technical field
The present invention relates to the technical fields of deep learning, more particularly to a kind of medical image processing method, device, set
Standby and storage medium.
Background technique
The ingredient of calculus is there are multiple types, and most of calculus are the calculus of Combination, and Calculus Analysis not only can be with
Determine stone composition, also for formulate the reasonable control measure of calculus and selection litholytic therapy important evidence is provided, to patient diet,
Living habit etc. is instructed, and the recurrence for preventing calculus is facilitated.
In general, being examined using CT scan (Computed Tomography, CT) to stone location
It surveys, obtains stone location image information, and then obtain ingredient, hardness and the brittleness of calculus according to the image information.
However, when being detected using the above method to calculus, due to the pixel of the CT scan image of heterogeneity calculus
There is overlapping between value, and different calculus of identical component, shape, volume, number, in terms of difference in class is all presented
Property big, the small characteristic of class inherited, therefore, by CT scan obtain image data can not accurately judge stone composition.
Summary of the invention
Based on this, it is necessary to aiming at the problem that can not accurately judge stone composition by CT images data, provide one kind
Medical image processing method, device, equipment and storage medium.
In a first aspect, a kind of medical image processing method, which comprises
Obtain CT scan image；
Obtain the calculus roughing region obtained after being labeled to the CT scan image；
Calculus central area is extracted from calculus roughing region；
The calculus central area is inputted into preset neural network model, exports calculus types results.
Above-mentioned medical image processing method, terminal are obtained and are marked to CT scan image by obtaining CT scan image
The calculus roughing region obtained after note, and then calculus central area is extracted from calculus roughing region, calculus central area is defeated
Enter preset neural network model, exports calculus types results.In the present embodiment, terminal carries out CT scan image by obtaining
The calculus roughing region obtained after mark obtains calculus central area, and calculus central area is inputted preset neural network mould
Type exports calculus types results, so that being deep by carrying out to calculus central area during obtaining calculus types results
Degree study obtains, and avoids and introduces non-calculus region during obtaining calculus types results, so that the calculus class obtained
Type result is more accurate.
It is described in one of the embodiments, that calculus central area is extracted from calculus roughing region, comprising:
Extracting scan image pixel in calculus roughing region to be greater than the region of preset threshold range is in the calculus
Heart district domain.
The preset threshold range includes: 76-376HU in one of the embodiments,.
In one of the embodiments, before the calculus central area is inputted preset neural network model, also
Include:
Calculus center is calculated by the calculus central area；
Random offset is carried out to the calculus center by preset offset parameter, obtains multiple calculus centers subregion；
Then, described that the calculus central area is inputted into preset neural network model, export calculus types results, packet
It includes:
The multiple calculus center subregion is inputted into preset neural network model, exports calculus types results.
In one of the embodiments, the method also includes:
Obtain multiple CT scan images；
The multiple calculus roughings region obtained after being labeled to multiple CT scan images is obtained respectively；
Multiple calculus central areas are extracted from multiple calculus roughing regions respectively；
Using the multiple calculus central area as input, by the corresponding multiple institutes in multiple calculus central areas
Calculus type is stated as output, neural network model is trained.
In one of the embodiments, using the multiple calculus central area as input, will be in multiple calculus
The corresponding multiple calculus types in heart district domain are as output, before being trained to neural network model, further includes:
Data enhancing processing is carried out to multiple calculus central areas respectively, obtains the calculus center of multiple enhancings
Domain；
Multiple calculus central areas are inputted into neural network model, the training neural network model obtains default
Neural network model, comprising:
It is using the calculus central area of multiple enhancings as input, multiple calculus central areas are corresponding
Multiple calculus types are trained neural network model, obtain preset neural network model as output.
It is described in one of the embodiments, that data enhancing processing is carried out to multiple calculus central areas respectively, it obtains
Take the calculus central area of multiple enhancings, comprising:
Cutting process and overturning processing are carried out to the calculus region, obtain the calculus central area of the enhancing.
Second aspect, a kind of medical image processing devices, described device include:
Module is obtained, for obtaining CT scan image；
Roughing module, for obtain the CT scan image is labeled after obtained calculus roughing region；
Extraction module, for extracting calculus central area from calculus roughing region；
Output module exports calculus type knot for the calculus central area to be inputted preset neural network model
Fruit.
The third aspect, a kind of computer equipment, including memory and processor, the memory are stored with computer journey
Sequence, the processor perform the steps of when executing the computer program
Obtain CT scan image；
Obtain the calculus roughing region obtained after being labeled to the CT scan image；
Calculus central area is extracted from calculus roughing region；
The calculus central area is inputted into preset neural network model, exports calculus types results.
Fourth aspect, a kind of computer readable storage medium are stored thereon with computer program, the computer program quilt
Processor performs the steps of when executing
Obtain the calculus roughing region obtained after the acquisition of CT scan image is labeled the CT scan image；
Calculus central area is extracted from calculus roughing region；
The calculus central area is inputted into preset neural network model, exports calculus types results.
Above-mentioned medical image processing method, device, equipment and storage medium, terminal are obtained by obtaining CT scan image
The calculus roughing region obtained after being labeled to CT scan image is taken, and then extracts calculus center from calculus roughing region
Calculus central area is inputted preset neural network model, exports calculus types results by domain.In the present embodiment, terminal passes through
It obtains obtained calculus roughing region after being labeled to CT scan image and obtains calculus central area, and by calculus central area
Preset neural network model is inputted, calculus types results are exported, so that being to pass through during obtaining calculus types results
Deep learning acquisition is carried out to calculus central area, avoids and introduces non-calculus area during obtaining calculus types results
Domain, so that the calculus types results obtained are more accurate.
Detailed description of the invention
Fig. 1 is the flow diagram of one embodiment traditional Chinese medicine image processing method；
Fig. 2 is the flow diagram of another embodiment traditional Chinese medicine image processing method；
Fig. 2 a is the schematic diagram of calculus center subregion in one embodiment；
Fig. 3 is the flow diagram of another embodiment traditional Chinese medicine image processing method；
Fig. 4 is the flow diagram of another embodiment traditional Chinese medicine image processing method；
Fig. 5 is the structural schematic diagram of the medical image processing devices provided in one embodiment；
Fig. 6 is the structural schematic diagram of the medical image processing devices provided in another embodiment；
Fig. 7 is the structural schematic diagram of the medical image processing devices provided in another embodiment；
Fig. 8 is the structural schematic diagram of the medical image processing devices provided in another embodiment；
Fig. 9 is the internal structure chart of computer equipment in one embodiment.
Specific embodiment
The ingredient of calculus is there are multiple types, and most of calculus are the calculus of Combination, and Calculus Analysis not only can be with
Determine stone composition, also for formulate the reasonable control measure of calculus and selection litholytic therapy important evidence is provided, to patient diet,
Living habit etc. is instructed, and the recurrence for preventing calculus is facilitated.In general, using CT scan
(Computed Tomography, CT) detects stone location, obtains stone location image information, and then according to the shadow
As the ingredient of information acquisition calculus, hardness and brittleness.However, when being detected using the above method to calculus, due to it is different at
There are overlapping, and the different calculus of identical component between the pixel value of the CT scan image of point calculus, shape, volume, number,
It is big that position etc. is all presented otherness in class, the small characteristic of class inherited, therefore, the image data obtained by CT scan
It can not accurately judge stone composition.Medical image processing method, device, equipment and storage medium provided by the invention, it is intended to
Improve the accuracy for obtaining calculus types results.
Medical image processing method provided in this embodiment, can be adapted in the terminal of Medical Image Processing, medicine figure
As the terminal of processing can have number for smart phone, tablet computer, laptop, desktop computer or personal digital assistant etc.
According to the electronic equipment of processing function, the present embodiment to the concrete form of Medical Image Processing terminal without limitation.
It should be noted that the method for Medical Image Processing provided in an embodiment of the present invention, executing subject can be doctor
The device of image procossing is learned, which can be implemented as medical image by way of software, hardware or software and hardware combining
Processing terminal it is some or all of.
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art
Every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
Fig. 1 is the flow diagram of one embodiment traditional Chinese medicine image processing method, and what is involved is by mentioning for the present embodiment
Calculus central area is taken, and obtains the detailed process of calculus types results according to calculus central area, as shown in Figure 1, this method
The following steps are included:
S101, CT scan image is obtained.
Wherein, CT scan image can be through scanner scanning tissue, and different tissues is passing through scanning
Instrument scanning obtains the different image of pixel value, as CT scan image.It can be local scan image, be also possible to complete
The scan image of body, the embodiment of the present application are without limitation.For example, abdominal CT scan image may include liver's CT scan figure
Picture, gall-bladder CT scan image, pancreas CT scan image and spleen CT scan image.
When specifically obtaining CT scan image, non-reinforcing list source CT, Micro-CT scanning, dual energy CT or double source dual intensity can be passed through
CT etc. is measured, tissue is scanned, the CT scan image of acquisition.
The calculus roughing region that S102, acquisition obtain after being labeled to CT scan image.
Wherein, calculus roughing region may include the CT scan image for getting rid of background area, in the base of above-described embodiment
It, can be with since organ has a degree of overlapping in vivo, in CT scan image when obtaining CT scan image on plinth
Including other organs in addition to organ where calculus.For example, may include kidney scan image to kidney portion CT scan image and defeated
Urinary catheter scan image, wherein ureter scan image is the background area in kidney portion CT scan image, calculus roughing area, kidney portion
Domain can be the CT scan image in removal ureter region.
At the calculus roughing region obtained after specific acquisition is labeled CT scan image, it can be and obtain user couple
Background area is labeled in CT scan image, CT scan image after being marked；It is also possible to user to CT scan figure
Non- background area is labeled as in, the CT scan image after being marked；The embodiment of the present application is without limitation.With kidney portion
For CT scan image, user can be labeled ureter region, alternatively, can also be labeled to kidney portion region, be obtained
The CT scan image that must be marked.
Further, it when terminal obtains calculus roughing region according to the CT scan image after mark, can be according to mark
Background area is removed, calculus roughing region is obtained.Continue by taking kidney portion CT scan image as an example, if the CT scan image of mark is pair
The CT scan image that ureter region (background area) is labeled, then terminal can delete the ureter region of mark, obtain
Calculus roughing region.
S103, calculus central area is extracted from calculus roughing region.
Wherein, on the basis of the above embodiments, calculus central area can be region of the calculus in CT scan image,
It can be a part of CT scan image.Calculus central area can be the CT scan image-region according to calculus size adjusting,
It is also possible to that the CT scan image-region of size is fixedly installed, the embodiment of the present application is without limitation.In calculus central area
It may include the corresponding CT scan image of 1 calculus, also may include the corresponding CT scan image of multiple calculus, the application is implemented
Example is without limitation.
When specifically extracting calculus central area from calculus roughing region, calculus can be extracted by partitioning algorithm
Central area.For example, on the basis of the above embodiments, it is determined that, should by partitioning algorithm segmentation behind the roughing region of calculus
Calculus roughing region can be through the partitioning algorithm based on threshold value, the partitioning algorithm based on region, the segmentation based on edge
Any one of algorithm and the partitioning algorithm based on specific theory divide the calculus roughing region, obtain multiple regions, it is assumed that
For left kidney region and right kidney region；Therefrom extracting according to the left kidney region where calculus is calculus central area.
S104, calculus central area is inputted into preset neural network model, exports calculus types results.
Wherein, preset neural network model can be by carrying out deep learning to the calculus central area of input, obtain
Obtain the neural network algorithm of calculus types results.From the constituent of calculus, calculus may include calcinm oxalate calculus, calcium phosphate
Calculus, uric acid (lithate) calculus, struvite calculus, cystine stone and purine calculus.Common calculus, constituent
It can be what above-mentioned Multiple components composition was mixed to form, the shape of calculus, volume, number, there is also biggish differences between position
Not, calculus types results may include this Shen of information such as the constituent of above-mentioned calculus, the shape of calculus, volume, number, position
Please embodiment it is without limitation.
Calculus central area is specifically being inputted into preset neural network model, during exporting calculus types results,
It, which can be, inputs preset neural network model for a calculus central area, exports its corresponding calculus types results；?
It can be and one group of calculus central area is inputted into preset neural network model, it is corresponding to export each calculus central area respectively
Calculus types results；The embodiment of the present application compares with no restrictions.When preset neural network model exports calculus central area
It, can also be by the mapping relations between the calculus central area and calculus type, as training after corresponding calculus types results
Sample, the training neural network model.When exporting calculus types results, neural network model can directly export calculus class
Type is also possible to export the mapping relations between calculus central area and calculus type, as calculus types results.
Above-mentioned medical image processing method, terminal are obtained and are marked to CT scan image by obtaining CT scan image
The calculus roughing region obtained after note, and then calculus central area is extracted from calculus roughing region, calculus central area is defeated
Enter preset neural network model, exports calculus types results.In the present embodiment, terminal carries out CT scan image by obtaining
The calculus roughing region obtained after mark obtains calculus central area, and calculus central area is inputted preset neural network mould
Type exports calculus types results, so that being deep by carrying out to calculus central area during obtaining calculus types results
Degree study obtains, and avoids and introduces non-calculus region during obtaining calculus types results, so that the calculus class obtained
Type result is more accurate.
It is possible to further determine calculus central area by scan image pixel in calculus roughing region, optionally,
S103 " calculus central area is extracted from calculus roughing region " a kind of possible implementation method, comprising: extract calculus roughing area
It is calculus central area that scan image pixel, which is greater than the region of preset threshold range, in domain.
Specifically, since calculus is different from the density of organ or tissue, calculus and organ or tissue are in scan image
In pixel it is different.On the basis of the above embodiments, the scan image for obtaining known calculus, is being scanned according to known calculus
Pixel in image determines preset threshold range, and optionally, preset threshold range is 76-376 Heng Shi unit (Hounsfield
Unit, HU).The region that terminal is greater than preset threshold range by extracting pixel in calculus roughing region, as calculus center
Domain.
Above-described embodiment is described in detail by extracting calculus central area, and obtains calculus class according to calculus central area
The detailed process of type result.It further, can also be right before calculus central area is inputted preset neural network model
Calculus central area carries out migration processing, obtains multiple calculus centers subregion.It is described in detail below by Fig. 2.
Fig. 2 is the flow diagram of another embodiment traditional Chinese medicine image processing method, and what is involved is to knot for the present embodiment
Stone central area carries out migration processing, obtains multiple calculus centers subregion, and the input of multiple calculus central areas is preset
Neural network model exports the detailed process of calculus types results, as shown in Fig. 2, this method is further comprising the steps of:
S201, calculus center is calculated by calculus central area.
Wherein, calculus can be three-dimension object in irregular shape, and calculus center can be the three-dimensional in irregular shape
The geometric center of object.When calculating calculus center in particular by calculus central area, on the basis of the above embodiments,
After obtaining calculus central area, due between the shape of calculus, volume, number, position there is also biggish difference,
Calculus center can be obtained according to the concrete condition of calculus, it, then can be with for example, calculus central area may include 2 calculus
Respectively, further, the region that scan image pixel in calculus central area is greater than preset threshold range is obtained, 2 calculus are obtained
Corresponding scan image region calculates separately the geometric areas in the corresponding scan image region of each calculus, obtains calculus center.
S202, random offset is carried out to calculus center by preset offset parameter, obtains multiple calculus centers subregion.
Wherein, preset offset parameter can be the size to calculus center mobile parameter and calculus center subregion
Size, calculus center subregion can be a part in the CT scan image comprising calculus center, can be each calculus
Respectively correspond that be also possible in a calculus area-of-interest in the region of CT scan image corresponding in the area of CT scan image
Domain, the embodiment of the present application are without limitation.For example, when in calculus central area there are when multiple calculus, calculus center sub-district
Domain can be the corresponding CT scan image of each calculus；When, there are when a calculus, which includes more in calculus central area
Various constituents, calculus center subregion can be a kind of corresponding CT scan image of constituent of the calculus.It is further false
If the constituent of a calculus is that calcium oxalate and calcium phosphate form calculus, then can be to the corresponding calculus center of the calculus
Domain, by obtaining two calculus center subregions, being respectively divided into calcinm oxalate calculus center subregion to calculus off-centring
With calcium phosphate calculus center subregion.
When carrying out random offset to calculus center especially by preset offset parameter, multiple calculus centers sub-district is obtained
During domain, random offset can be carried out by the three-dimensional coordinate to calculus center, then cut calculus central area and tied
Stone center subregion.For example, in a calculus central area, as shown in Figure 2 a, by preset offset parameter in calculus
The three-dimensional coordinate of the heart 210 carries out random offset, and offset parameter can be [10,10,2], obtains calcinm oxalate calculus center sub-district
Domain 220 and calcium phosphate calculus center subregion 230.
S203, multiple calculus centers subregion is inputted into preset neural network model, exports calculus types results.
Specifically, on the basis of the above embodiments, multiple calculus centers subregion is obtained, it can will be in multiple calculus
The preset neural network model of the input of center region one by one exports the corresponding calculus class of each calculus center subregion respectively
Type result.It is also possible to establishing multiple calculus centers subregion into a set, which is inputted into preset neural network mould
Type exports calculus types results set, includes the corresponding calculus types results of each calculus subregion in calculus types results set,
With corresponding relationship between calculus subregion and calculus types results.The embodiment of the present application is without limitation.
Above-mentioned medical image processing method, terminal calculate calculus center by calculus central area, and by preset inclined
Shifting parameter carries out random offset to calculus center, obtains multiple calculus centers subregion, and then by multiple calculus centers subregion
Preset neural network model is inputted, calculus types results are exported.In the present embodiment, terminal inputs in advance to calculus central area
If neural network model when, the calculus center subregion that more segments is obtained by calculus central area, is reduced preset
The background area of neural network model input, further improves calculus types results accuracy.
On the basis of the above embodiments, terminal can also be using multiple calculus central areas as input, by multiple calculus
The corresponding multiple calculus types in central area are trained neural network model as output.Come below by Fig. 3 detailed
It describes in detail bright.
Fig. 3 is the flow diagram of another embodiment traditional Chinese medicine image processing method, as shown in figure 3, this method is also wrapped
Include following steps:
S301, multiple CT scan images are obtained.
Specifically, multiple CT scan images are obtained, wherein multiple CT scan images can be history CT scan image,
It can be and obtain multiple CT scan images in real time, can also be fetching portion history CT scan image and part CT scan figure in real time
Picture, the embodiment of the present application are without limitation.
S302, the multiple calculus roughings region obtained after being labeled to multiple CT scan images is obtained respectively.
S303, multiple calculus central areas are extracted from multiple calculus roughings region respectively.
S304, using multiple calculus central areas as input, by the corresponding multiple calculus in multiple calculus central areas
Type is trained neural network model as output.
Specifically, on the basis of the above embodiments, can be using multiple calculus central areas as input, it can also be with
Multiple calculus central areas are grouped, obtain one or more calculus central area set as input, the application is implemented
Example only limits this.During being specifically trained to neural network model, it can be and preset one group of training ginseng
Number, inputs multiple calculus central areas, by the model of pre-set one group of training parameter, obtains corresponding calculus type,
The calculus type actual calculus type corresponding with corresponding calculus center grade is compared, according to the comparing result,
Adjusting training parameter, until corresponding with calculus central area actual by the calculus types results that neural network model obtains
Calculus types results meet preset requirement, as target training parameter, according to the target training parameter, determine preset mind
Through network model.
Above-mentioned medical image processing method, terminal obtains multiple CT scan images, and is obtained respectively to multiple CT scan figures
As the multiple calculus roughings region obtained after being labeled, multiple calculus centers are extracted from multiple calculus roughings region respectively
Domain, and then using multiple calculus central areas as input, the corresponding multiple calculus types in multiple calculus central areas are made
For output, neural network model is trained.In the present embodiment, terminal is exporting calculus by preset neural network model
Before types results, by using multiple calculus central areas as input, multiple calculus central areas are corresponding multiple
Calculus type is trained neural network model, obtains preset neural network model, improve preset mind as output
Accuracy through network model, and then improve the accuracy for obtaining calculus types results.
Further, using multiple calculus central areas as input, multiple calculus central areas are corresponding more
A calculus type is as output, before being trained to neural network model, terminal can also to multiple calculus central areas into
Row data enhancing processing, obtains the calculus central area of multiple enhancings.
Fig. 4 is the flow diagram of another embodiment traditional Chinese medicine image processing method, and what is involved is pass through for the present embodiment
The detailed process of the preset neural network model of calculus central area training of multiple enhancings, as shown in figure 4, S304 " will be multiple
Calculus central area is as input, using the corresponding multiple calculus types in multiple calculus central areas as output, to nerve
Network model is trained " a kind of possible implementation method, comprising the following steps:
S401, data enhancing processing is carried out to multiple calculus central areas respectively, obtains the calculus center of multiple enhancings
Domain.
Specifically, the calculus central area of enhancing, which can be, carries out processing acquisition to partial region in CT scan image,
For example, CT scan image is mostly three-dimensional tangent plane picture, wherein calculus is corresponding is specifically carrying out data increasing to calculus central area
When strength is managed, it can be and random offset is carried out by preset offset parameter to calculus central area, obtain multiple calculus centers
Subregion, the calculus central area as enhanced；It is also possible to carry out overturning processing to calculus central area, obtains multiple overturnings
Calculus center subregion, the calculus central area as enhanced, the embodiment of the present application is without limitation.Optionally, to knot
Stone region carries out cutting process and overturning processing, obtains the calculus central area of enhancing.
S402, using the calculus central area of multiple enhancings as input, multiple calculus central areas are corresponding more
A calculus type is trained neural network model, obtains preset neural network model as output.
Specifically, using the calculus central area of multiple enhancings as input, multiple calculus central areas are corresponding
Multiple calculus types, to the process that neural network model is trained, can be as output and preset one group of training parameter,
The calculus central area for inputting enhancing obtains corresponding calculus type, by this by the model of pre-set one group of training parameter
Calculus type actual calculus type corresponding with the calculus central area of enhancing compares, according to the comparing result, adjustment
Training parameter, until the calculus types results reality corresponding with the calculus central area of enhancing obtained by neural network model
Calculus types results, meet preset requirement, as target training parameter, according to the target training parameter, obtain preset
Neural network model.
Above-mentioned medical image processing method, terminal carry out data enhancing processing to multiple calculus central areas respectively, obtain
The calculus central area of multiple enhancings, and using the calculus central area of multiple enhancings as input, by multiple calculus central areas
Corresponding multiple calculus types are trained neural network model, obtain preset neural network model as output.
In the present embodiment, terminal is when being trained neural network model by multiple calculus central areas, to calculus central area
Data enhancing processing is carried out, the calculus central area of multiple enhancings is obtained, so that the input of training neural network model is enhancing
Calculus central area further increase and obtain calculus type so that preset neural network model is more accurate
As a result accuracy.
Although should be understood that each step in the flow chart of Fig. 1-4 according to the instruction of arrow, is successively shown,
It is these steps is not that the inevitable sequence according to arrow instruction successively executes.Unless expressly stating otherwise herein, these steps
There is no stringent sequences to limit for rapid execution, these steps can execute in other order.Moreover, in Fig. 1-4 at least
A part of step may include that perhaps these sub-steps of multiple stages or stage are not necessarily in same a period of time to multiple sub-steps
Quarter executes completion, but can execute at different times, the execution in these sub-steps or stage be sequentially also not necessarily according to
Secondary progress, but in turn or can replace at least part of the sub-step or stage of other steps or other steps
Ground executes.
Fig. 5 is the structural schematic diagram of the medical image processing devices provided in one embodiment, as shown in figure 5, the medicine
Image processing apparatus includes: to obtain module 10, roughing module 20, extraction module 30 and output module 40, in which:
Module 10 is obtained, for obtaining CT scan image；
Roughing module 20, for obtain the CT scan image is labeled after obtained calculus roughing region；
Extraction module 30, for extracting calculus central area from calculus roughing region；
Output module 40 exports calculus type for the calculus central area to be inputted preset neural network model
As a result.
In one embodiment, it is big to be specifically used for extracting scan image pixel in calculus roughing region for extraction module 30
In the region of preset threshold range be the calculus central area.
In one embodiment, the preset threshold range includes: 76-376HU.
Medical image processing devices provided in an embodiment of the present invention can execute above method embodiment, realization principle
Similar with technical effect, details are not described herein.
Fig. 6 is the structural schematic diagram of the medical image processing devices provided in another embodiment, embodiment shown in Fig. 5
On the basis of, as shown in fig. 6, medical image processing devices further include: computing module 50 and offset module 60, wherein
Computing module 50, for calculating calculus center by the calculus central area；
Offset module 60 obtains multiple for carrying out random offset to the calculus center by preset offset parameter
Calculus center subregion；
Then, output module 40 are also used to the multiple calculus center subregion inputting preset neural network model, defeated
Calculus types results out.
Medical image processing devices provided in an embodiment of the present invention can execute above method embodiment, realization principle
Similar with technical effect, details are not described herein.
Fig. 7 is the structural schematic diagram of the medical image processing devices provided in another embodiment, shown in Fig. 5 or Fig. 6
On the basis of embodiment, as shown in fig. 7, medical image processing devices further include: training module 70, wherein
Module 10 is obtained, is also used to obtain multiple CT scan images；
Roughing module 20 is also used to obtain the multiple calculus obtained after being labeled to multiple CT scan images respectively
Roughing region；
Extraction module 30 is also used to extract multiple calculus central areas from multiple calculus roughing regions respectively；
Training module 70 is used for using the multiple calculus central area as input, by multiple calculus central areas
Corresponding multiple calculus types are trained neural network model as output.
Medical image processing devices provided in an embodiment of the present invention can execute above method embodiment, realization principle
Similar with technical effect, details are not described herein.
Fig. 8 is the structural schematic diagram of the medical image processing devices provided in another embodiment, in any one of Fig. 5-7 institute
On the basis of showing embodiment, as shown in figure 8, medical image processing devices further include: enhancing module 80, wherein
Enhance module 80, for carrying out data enhancing processing to multiple calculus central areas respectively, obtains multiple increasings
Strong calculus central area；
Then,
Training module 70 is also used to using the calculus central area of multiple enhancings as input, by multiple calculus
The corresponding multiple calculus types in central area are trained neural network model, obtain preset as output
Neural network model.
In one embodiment, enhancing module 80 is specifically used for carrying out at cutting process and overturning the calculus region
Reason, obtains the calculus central area of the enhancing.
Medical image processing devices provided in an embodiment of the present invention can execute above method embodiment, realization principle
Similar with technical effect, details are not described herein.
A kind of specific restriction about medical image processing devices may refer to above to medical image processing method
It limits, details are not described herein.Modules in above-mentioned medical image processing devices can fully or partially through software, hardware and
A combination thereof is realized.Above-mentioned each module can be embedded in the form of hardware or independently of in the processor in computer equipment, can also
Be stored in the memory in computer equipment in a software form, in order to which processor calls the above modules of execution corresponding
Operation.
In one embodiment, a kind of computer equipment is provided, which can be terminal device, inside
Structure chart can be as shown in Figure 9.The computer equipment include by system bus connect processor, memory, network interface,
Display screen and input unit.Wherein, the processor of the computer equipment is for providing calculating and control ability.The computer equipment
Memory include non-volatile memory medium, built-in storage.The non-volatile memory medium is stored with operating system and calculating
Machine program.The built-in storage provides environment for the operation of operating system and computer program in non-volatile memory medium.It should
The network interface of computer equipment is used to communicate with external terminal by network connection.The computer program is executed by processor
When to realize a kind of medical image processing method.The display screen of the computer equipment can be liquid crystal display or electric ink
Display screen, the input unit of the computer equipment can be the touch layer covered on display screen, be also possible to outside computer equipment
Key, trace ball or the Trackpad being arranged on shell can also be external keyboard, Trackpad or mouse etc..
It will be understood by those skilled in the art that structure shown in Fig. 9, only part relevant to application scheme is tied
The block diagram of structure does not constitute the restriction for the computer equipment being applied thereon to application scheme, specific computer equipment
It may include perhaps combining certain components or with different component layouts than more or fewer components as shown in the figure.
In one embodiment, a kind of terminal device, including memory and processor are provided, the memory is stored with
Computer program, the processor perform the steps of when executing the computer program
Obtain CT scan image；
Obtain the calculus roughing region obtained after being labeled to the CT scan image；
Calculus central area is extracted from calculus roughing region；
The calculus central area is inputted into preset neural network model, exports calculus types results.
In one embodiment, it is also performed the steps of when processor executes computer program and extracts the calculus roughing
It is the calculus central area that scan image pixel, which is greater than the region of preset threshold range, in region.
In one embodiment, the preset threshold range includes: 76-376HU.
In one embodiment, it also performs the steps of when processor executes computer program through the calculus center
Region calculates calculus center；Random offset is carried out to the calculus center by preset offset parameter, is obtained in multiple calculus
Center region；The multiple calculus center subregion is inputted into preset neural network model, exports calculus types results.
In one embodiment, acquisition multiple CT are also performed the steps of when processor executes computer program to sweep
Trace designs picture；The multiple calculus roughings region obtained after being labeled to multiple CT scan images is obtained respectively；Respectively from more
Multiple calculus central areas are extracted in a calculus roughing region；It, will be more using the multiple calculus central area as input
The corresponding multiple calculus types in a calculus central area are trained neural network model as output.
In one embodiment, it also performs the steps of respectively when processor executes computer program to multiple knots
Stone central area carries out data enhancing processing, obtains the calculus central area of multiple enhancings；It will be in the calculus of multiple enhancings
Heart district domain as input, will the corresponding multiple calculus types in the multiple calculus central areas as exporting, to mind
It is trained through network model, obtains preset neural network model.
In one embodiment, processor execute computer program when also perform the steps of to the calculus region into
Row cutting process and overturning processing, obtain the calculus central area of the enhancing.
Terminal device provided in this embodiment, implementing principle and technical effect are similar with above method embodiment, herein
It repeats no more.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
Machine program performs the steps of when being executed by processor
Obtain CT scan image；
Obtain the calculus roughing region obtained after being labeled to the CT scan image；
Calculus central area is extracted from calculus roughing region；
The calculus central area is inputted into preset neural network model, exports calculus types results.
In one embodiment, it is performed the steps of when computer program is executed by processor and extracts the calculus roughing
It is the calculus central area that scan image pixel, which is greater than the region of preset threshold range, in region.
In one embodiment, the preset threshold range includes: 76-376HU.
In one embodiment, it is performed the steps of when computer program is executed by processor through the calculus center
Region calculates calculus center；Random offset is carried out to the calculus center by preset offset parameter, is obtained in multiple calculus
Center region；The multiple calculus center subregion is inputted into preset neural network model, exports calculus types results.
In one embodiment, acquisition multiple CT are performed the steps of when computer program is executed by processor to sweep
Trace designs picture；The multiple calculus roughings region obtained after being labeled to multiple CT scan images is obtained respectively；Respectively from more
Multiple calculus central areas are extracted in a calculus roughing region；It, will be more using the multiple calculus central area as input
The corresponding multiple calculus types in a calculus central area are trained neural network model as output.
In one embodiment, it is performed the steps of when computer program is executed by processor respectively to multiple knots
Stone central area carries out data enhancing processing, obtains the calculus central area of multiple enhancings；It will be in the calculus of multiple enhancings
Heart district domain as input, will the corresponding multiple calculus types in the multiple calculus central areas as exporting, to mind
It is trained through network model, obtains preset neural network model.
In one embodiment, performed the steps of when computer program is executed by processor to the calculus region into
Row cutting process and overturning processing, obtain the calculus central area of the enhancing.
Computer readable storage medium provided in this embodiment, implementing principle and technical effect and above method embodiment
Similar, details are not described herein.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the computer program can be stored in a non-volatile computer
In read/write memory medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein,
To any reference of memory, storage, database or other media used in each embodiment provided herein,
Including non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include
Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms,
Such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhancing
Type SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM
(RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
Each technical characteristic of embodiment described above can be combined arbitrarily, for simplicity of description, not to above-mentioned reality
It applies all possible combination of each technical characteristic in example to be all described, as long as however, the combination of these technical characteristics is not deposited
In contradiction, all should be considered as described in this specification.
The embodiments described above only express several embodiments of the present invention, and the description thereof is more specific and detailed, but simultaneously
It cannot therefore be construed as limiting the scope of the patent.It should be pointed out that coming for those of ordinary skill in the art
It says, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to protection of the invention
Range.Therefore, the scope of protection of the patent of the invention shall be subject to the appended claims.
Claims (10)
1. a kind of medical image processing method, which is characterized in that the described method includes:
Obtain CT scan image；
Obtain the calculus roughing region obtained after being labeled to the CT scan image；
Calculus central area is extracted from calculus roughing region；
The calculus central area is inputted into preset neural network model, exports calculus types results.
2. method according to claim 1, which is characterized in that described to extract calculus center from calculus roughing region
Domain, comprising:
Extracting scan image pixel in calculus roughing region to be greater than the region of preset threshold range is the calculus center
Domain.
3. method according to claim 3, which is characterized in that the preset threshold range includes: 76-376HU.
4. method according to any one of the claim 1 to 3, which is characterized in that default inputting the calculus central area
Neural network model before, further includes:
Calculus center is calculated by the calculus central area；
Random offset is carried out to the calculus center by preset offset parameter, obtains multiple calculus centers subregion；
Then, described that the calculus central area is inputted into preset neural network model, export calculus types results, comprising:
The multiple calculus center subregion is inputted into preset neural network model, exports calculus types results.
5. method according to any one of the claim 1 to 3, which is characterized in that the method also includes:
Obtain multiple CT scan images；
The multiple calculus roughings region obtained after being labeled to multiple CT scan images is obtained respectively；
Multiple calculus central areas are extracted from multiple calculus roughing regions respectively；
Using the multiple calculus central area as input, by the corresponding multiple knots in multiple calculus central areas
Stone type is trained neural network model as output.
6. method according to claim 5, which is characterized in that using the multiple calculus central area as input, will be more
The corresponding multiple calculus types in a calculus central area are trained it to neural network model as output
Before, further includes:
Data enhancing processing is carried out to multiple calculus central areas respectively, obtains the calculus central area of multiple enhancings；
Multiple calculus central areas are inputted into neural network model, the training neural network model obtains preset mind
Through network model, comprising:
It is using the calculus central area of multiple enhancings as input, multiple calculus central areas are corresponding multiple
The calculus type is trained neural network model, obtains preset neural network model as output.
7. method according to claim 6, which is characterized in that described to carry out data to multiple calculus central areas respectively
Enhancing processing, obtains the calculus central area of multiple enhancings, comprising:
Cutting process and overturning processing are carried out to the calculus region, obtain the calculus central area of the enhancing.
8. a kind of medical image processing devices, which is characterized in that described device includes:
Module is obtained, for obtaining CT scan image；
Roughing module, for obtain the CT scan image is labeled after obtained calculus roughing region；
Extraction module, for extracting calculus central area from calculus roughing region；
Output module exports calculus types results for the calculus central area to be inputted preset neural network model.
9. a kind of computer equipment, including memory and processor, the memory are stored with computer program, feature exists
In when the processor executes the computer program the step of any one of realization claim 1-7 the method.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program
The step of method of any of claims 1-7 is realized when being executed by processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811596856.4A CN109685796B (en) | 2018-12-26 | 2018-12-26 | Medical image processing method, apparatus, device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811596856.4A CN109685796B (en) | 2018-12-26 | 2018-12-26 | Medical image processing method, apparatus, device and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109685796A true CN109685796A (en) | 2019-04-26 |
| CN109685796B CN109685796B (en) | 2021-05-18 |
Family
ID=66188483
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811596856.4A Active CN109685796B (en) | 2018-12-26 | 2018-12-26 | Medical image processing method, apparatus, device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109685796B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111105472A (en) * | 2019-11-26 | 2020-05-05 | 上海联影智能医疗科技有限公司 | Attenuation correction method and device for PET image, computer device and storage medium |
| CN112656483A (en) * | 2020-12-21 | 2021-04-16 | 中南大学湘雅医院 | Visual portable choledochoscope lithotomy forceps |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101044988A (en) * | 2006-03-31 | 2007-10-03 | 西门子公司 | Method and device for automatic distinguishing kidney stone types by computer tomography contrast |
| US20120143803A1 (en) * | 2010-12-06 | 2012-06-07 | Furuno Electric Co., Ltd. | Bottom sediment determination device, ultrasonic finder, and method and program for setting parameters |
| EP2889838A1 (en) * | 2013-12-27 | 2015-07-01 | Nuctech Company Limited | Fluoroscopic inspection system and method for automatic classification and recognition of cargoes |
| CN106250712A (en) * | 2016-09-28 | 2016-12-21 | 湖南老码信息科技有限责任公司 | A kind of ureteral calculus Forecasting Methodology based on increment type neural network model and prognoses system |
| CN107368670A (en) * | 2017-06-07 | 2017-11-21 | 万香波 | Stomach cancer pathology diagnostic support system and method based on big data deep learning |
| CN107480677A (en) * | 2017-08-07 | 2017-12-15 | 北京深睿博联科技有限责任公司 | The method and device of area-of-interest in a kind of identification three-dimensional CT image |
| CN107492097A (en) * | 2017-08-07 | 2017-12-19 | 北京深睿博联科技有限责任公司 | A kind of method and device for identifying MRI image area-of-interest |
| CN108364006A (en) * | 2018-01-17 | 2018-08-03 | 超凡影像科技股份有限公司 | Medical Images Classification device and its construction method based on multi-mode deep learning |
-
2018
- 2018-12-26 CN CN201811596856.4A patent/CN109685796B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101044988A (en) * | 2006-03-31 | 2007-10-03 | 西门子公司 | Method and device for automatic distinguishing kidney stone types by computer tomography contrast |
| US20120143803A1 (en) * | 2010-12-06 | 2012-06-07 | Furuno Electric Co., Ltd. | Bottom sediment determination device, ultrasonic finder, and method and program for setting parameters |
| EP2889838A1 (en) * | 2013-12-27 | 2015-07-01 | Nuctech Company Limited | Fluoroscopic inspection system and method for automatic classification and recognition of cargoes |
| CN106250712A (en) * | 2016-09-28 | 2016-12-21 | 湖南老码信息科技有限责任公司 | A kind of ureteral calculus Forecasting Methodology based on increment type neural network model and prognoses system |
| CN107368670A (en) * | 2017-06-07 | 2017-11-21 | 万香波 | Stomach cancer pathology diagnostic support system and method based on big data deep learning |
| CN107480677A (en) * | 2017-08-07 | 2017-12-15 | 北京深睿博联科技有限责任公司 | The method and device of area-of-interest in a kind of identification three-dimensional CT image |
| CN107492097A (en) * | 2017-08-07 | 2017-12-19 | 北京深睿博联科技有限责任公司 | A kind of method and device for identifying MRI image area-of-interest |
| CN108364006A (en) * | 2018-01-17 | 2018-08-03 | 超凡影像科技股份有限公司 | Medical Images Classification device and its construction method based on multi-mode deep learning |
Non-Patent Citations (2)
| Title |
|---|
| CORINA BOTOCA ET AL: "Cellular neural networks assisted automatic detection of elements in microscopic medical images. A preliminary study", 《2014 11TH INTERNATIONAL SYMPOSIUM ON ELECTRONICS AND TELECOMMUNICATIONS (ISETC)》 * |
| 张鹏等: "应用人工神经网络预测ESWL单次治疗肾结石的成功率", 《临床泌尿外科杂志》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111105472A (en) * | 2019-11-26 | 2020-05-05 | 上海联影智能医疗科技有限公司 | Attenuation correction method and device for PET image, computer device and storage medium |
| CN111105472B (en) * | 2019-11-26 | 2024-03-22 | 上海联影智能医疗科技有限公司 | Attenuation correction method, apparatus, computer device and storage medium for PET image |
| CN112656483A (en) * | 2020-12-21 | 2021-04-16 | 中南大学湘雅医院 | Visual portable choledochoscope lithotomy forceps |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109685796B (en) | 2021-05-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110148192B (en) | Medical image imaging method, device, computer equipment and storage medium | |
| JP5814504B2 (en) | Medical image automatic segmentation system, apparatus and processor using statistical model | |
| CN110111305B (en) | Processing scheme generation method, device, equipment and storage medium | |
| CN110211076B (en) | Image stitching method, image stitching equipment and readable storage medium | |
| CN111047591A (en) | Focal volume measuring method, system, terminal and storage medium based on deep learning | |
| CN110415792B (en) | Image detection method, image detection device, computer equipment and storage medium | |
| CN111161216A (en) | Intravascular ultrasound image processing method, device, equipment and storage medium based on deep learning | |
| CN109712163B (en) | Coronary artery extraction method, device, image processing workstation and readable storage medium | |
| CN111325759A (en) | Blood vessel segmentation method, device, computer equipment and readable storage medium | |
| CN110363774A (en) | Image partition method, device, computer equipment and storage medium | |
| CN111080584A (en) | Quality control method for medical image, computer device and readable storage medium | |
| CN109754396A (en) | Method for registering, device, computer equipment and the storage medium of image | |
| CN107885476A (en) | A kind of medical image display methods and device | |
| CN117173050A (en) | Image processing method, device, computer equipment and storage medium | |
| CN113538471B (en) | Plaque segmentation method, plaque segmentation device, computer equipment and storage medium | |
| CN109685796A (en) | Medical image processing method, device, equipment and storage medium | |
| CN109378068A (en) | A kind of method for automatically evaluating and system of Therapeutic Effects of Nasopharyngeal | |
| CN111028212A (en) | Key point detection method and device, computer equipment and storage medium | |
| CN110473226B (en) | Training method of image processing network, computer device and readable storage medium | |
| CN111128348B (en) | Medical image processing method, medical image processing device, storage medium and computer equipment | |
| US9123163B2 (en) | Medical image display apparatus, method and program | |
| US20130072782A1 (en) | System and method for automatic magnetic resonance volume composition and normalization | |
| CN110974416B (en) | Puncture parameter determination method, device, system, computer equipment and storage medium | |
| KR20170128975A (en) | Vessel segmentation device and vessel segmentation method thereof | |
| CN114375179A (en) | Ultrasonic image analysis method, ultrasonic imaging system, and computer storage medium |
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 |