CN110400292A - Shift outcome evaluation method, computer equipment and storage medium - Google Patents
Shift outcome evaluation method, computer equipment and storage medium Download PDFInfo
- Publication number
- CN110400292A CN110400292A CN201910597572.5A CN201910597572A CN110400292A CN 110400292 A CN110400292 A CN 110400292A CN 201910597572 A CN201910597572 A CN 201910597572A CN 110400292 A CN110400292 A CN 110400292A
- Authority
- CN
- China
- Prior art keywords
- object construction
- feature
- medical image
- image group
- image
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30096—Tumor; Lesion
Abstract
This application involves a kind of transfer outcome evaluation method, computer equipment and storage mediums, by the medical image for obtaining multiple mode of object construction, then the comprehensive characteristics value of object construction is extracted from each mode medical image, based on the comprehensive characteristics value to object construction whether transfer is assessed, in this way, the multiple modalities of object construction and the situation of change of comprehensive characteristics value are considered simultaneously, according to the assessment transfer of the Multiple factors of object construction as a result, greatly improving the precision of assessment result.
Description
Technical field
This application involves medical detection technologies, more particularly to a kind of transfer outcome evaluation method, computer equipment
And storage medium.
Background technique
Thyroid cancer is the most common thyroid malignancy, it has also become in the world disease incidence fastest-rising tumour it
One.Wherein, lower with grade of malignancy, the preferable papillary carcinoma of prognostic is most commonly seen.When usually major part papillary carcinoma is made a definite diagnosis
With metastasic cervical lymph nodes, lymphatic metastasis is the most important risks and assumptions of local recurrence, and harm is more compared with primary tumor
Seriously, therefore, the preoperative accurate judgement whether shifted to lymph node has weight to the selection of art formula and the prediction of tumor recurrence
The meaning wanted.
At present both at home and abroad to the preoperative evaluation of lymphatic metastasis, by some special imaging features to certain groups of lymph nodes
Judged whether transfer, such as: pass through ultrasonic (Ultrasound, US) or computed tomography (Computed
Tomography, CT) judge the sign that whether there is lymph node in medical image, then by the lymph for these signs that can most coincide
Knot is determined as lymphatic metastasis.
But there is a problem of that precision is lower to the judgment method of lymphatic metastasis at present.
Summary of the invention
Based on this, it is necessary to which to the judgment method of lymphatic metastasis, there are the lower technologies of precision to ask for above-mentioned at present
Topic provides a kind of transfer outcome evaluation method, computer equipment and storage medium.
In a first aspect, the embodiment of the present application provides a kind of transfer outcome evaluation method, this method comprises:
Obtain the medical image of object construction；Wherein, medical image includes the medical image and target of at least two mode
Segmentation mask of the structure in each mode medical image；
The comprehensive characteristics value of object construction is extracted from the medical image of object construction；Wherein, comprehensive characteristics value includes mesh
Mark the image group feature of structure；
The transfer result of object construction is assessed according to the comprehensive characteristics value of object construction.
Second aspect, the embodiment of the present application provide a kind of transfer outcome evaluation device, and described device includes:
Module is obtained, for obtaining the medical image of object construction；Wherein, medical image includes the doctor of at least two mode
Learn the segmentation mask of image and object construction in each mode medical image；
Extraction module, for extracting the comprehensive characteristics value of object construction from the medical image of object construction；Wherein, comprehensive
Characteristic value includes the image group feature of object construction；
Evaluation module, for assessing the transfer result of object construction according to the comprehensive characteristics value of object construction.
The third aspect, the embodiment of the present application provide a kind of computer equipment, including memory and processor, memory storage
There is computer program, processor realizes the step for any one method that above-mentioned first aspect embodiment provides when executing computer program
Suddenly.
Fourth aspect, the embodiment of the present application provide a kind of computer readable storage medium, are stored thereon with computer program,
The step of any one method that above-mentioned first aspect embodiment provides is realized when computer program is executed by processor.
A kind of transfer outcome evaluation method, computer equipment and storage medium provided by the embodiments of the present application, pass through acquisition
Then the medical image of multiple mode of object construction extracts the comprehensive characteristics value of object construction from each mode medical image,
Based on the comprehensive characteristics value to object construction whether transfer is assessed, in this way, considering the multiple modalities of object construction simultaneously
With the situation of change of comprehensive characteristics value, transfer is assessed according to the Multiple factors of object construction as a result, greatly improving assessment knot
The precision of fruit.
Detailed description of the invention
Fig. 1 is a kind of applied environment figure for transfer outcome evaluation that one embodiment provides；
Fig. 2 is a kind of flow diagram for transfer outcome evaluation method that one embodiment provides；
Fig. 3 is a kind of flow diagram for transfer outcome evaluation method that one embodiment provides；
Fig. 4 is a kind of transfer outcome evaluation method complete diagram that one embodiment provides；
Fig. 5 is a kind of flow diagram for transfer outcome evaluation method that one embodiment provides；
Fig. 6 is a kind of flow diagram for transfer outcome evaluation method that one embodiment provides；
Fig. 7 is a kind of flow diagram for transfer outcome evaluation method that one embodiment provides；
Fig. 8 is a kind of structural block diagram for transfer outcome evaluation device that one embodiment provides；
Fig. 9 is a kind of structural block diagram for transfer outcome evaluation device that one embodiment provides；
Figure 10 is a kind of structural block diagram for transfer outcome evaluation device that one embodiment provides；
Figure 11 is a kind of structural block diagram for transfer outcome evaluation device that one embodiment provides；
Figure 12 is a kind of structural block diagram for transfer outcome evaluation device that one embodiment provides.
Specific embodiment
It is with reference to the accompanying drawings and embodiments, right in order to which the objects, technical solutions and advantages of the application are more clearly understood
The application is further elaborated.It should be appreciated that specific embodiment described herein is only used to explain the application, not
For limiting the application.
A kind of transfer outcome evaluation method provided by the present application, can be applied in application environment as shown in Figure 1, the meter
Calculating machine equipment can be server, and internal structure chart can be as shown in Figure 1.The computer equipment includes being connected by system bus
Processor, memory, network interface and the database connect.Wherein, the processor of the computer equipment is calculated and is controlled for providing
Ability processed.The memory of the computer equipment includes non-volatile memory medium, built-in storage.The non-volatile memory medium is deposited
Contain operating system, computer program and database.The built-in storage is operating system and meter in non-volatile memory medium
The operation of calculation machine program provides environment.The database of the computer equipment is used for the data of memory transfer outcome evaluation method.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 transfer outcome evaluation method.
Thyroid cancer is the most common thyroid malignancy, accounts for about the 1% of whole body malignant tumour, and become the world
One of upper fastest-rising tumour of disease incidence.In general, thyroid cancer includes papillary carcinoma, undifferentiated carcinoma, folliculus dress cancer and marrow sample
Four kinds of histological types of cancer, wherein it is lower with grade of malignancy, the preferable papillary carcinoma of prognostic is most commonly seen, account for about 80%-90%.
However, with metastasic cervical lymph nodes, especially center stack lymphatic metastasis when the papillary carcinoma of 30%-90% is made a definite diagnosis.Leaching
The most important risks and assumptions that transfer is local recurrence are fawned on, endanger even more serious compared with primary tumor, or even are needed twice and two
The secondary above operation is cut off, and the quality of life of patient has been seriously affected.Preventative lymph node dissection can reduce tumour
Possibility, while also increasing the incidence of the severe complications such as parathyroid gland and recurrent nerve injury, it is therefore, preoperative to lymph
The accurate judgement whether knot shifts has great importance to the selection of art formula and the prediction of tumor recurrence.
At present both at home and abroad to the preoperative evaluation of metastasic cervical lymph nodes, image compares with pathology/cytology and remains in group-
Group is horizontal, i.e., judges whether by some special imaging features to certain groups of lymphatic metastasis, such as ultrasound
(Ultrasound, US) discovery is focal or diffused strong echo, downright bad capsule changes, most path/maximum diameter > 0.5, around or
Mix blood flow patterns, computed tomography (Computed Tomography, CT) find Microcalcification, downright bad capsule become, uniformly or
Uneven high enhancing, most path/maximum diameter > 0.5, the lymph node interpretation for these signs that can most coincide is transfer, therefore in iconography
Lymphatic metastasis be not can one-to-one correspondence pathologically, although preoperative fine needle puncture, which positions, can be achieved image and pathology knot-knot
Level control, however will increase clinical position amount, and bring certain damage to patient, therefore its application is limited, how to realize image
It is the huge challenge that clinical and image department doctor faces with pathology/cytology knot-knot level control.Therefore, US and CT is clinical
The common method of guiding puncture biopsy, wherein the most commonly used with the former in terms of thyroid nodule and superficial lymph knot, US guidance
The accuracy rate of lower fine needle aspiration biopsy (Fine Needle Aspiration, FNA) etiologic diagnosis is more than 94%, but when US encounters
When background echo is mixed and disorderly, lesion echo difference is small or poor image quality, FNA success rate is limited, in addition, also normal in neck
It is influenced by gas in tracheae, oesophagus and the cover of breastbone and maxillofacial bone, after limiting FNA to center stack, upper mediastinum and pharynx
The application of deep lymph node；Although CT has, soft tissue resolution is low, patient and operator's raying exposure etc. in bootup process
Deficiency, but possess good spatial resolution, so that the shortcomings that lower FNA of CT-US fusion navigation can not only overcome US, application
Its Real-time Motion Image is shown, and can play the advantage that CT clearly shows lesion and neighbouring important anatomy structure, to realize leaching
Fawn on that image and pathology are/cytological to tie-to bear water flat compare.But the judgment method of current lymphatic metastasis there are precision compared with
Low, therefore, the embodiment of the present application provides a kind of transfer outcome evaluation method, apparatus, computer equipment and storage medium, it is intended to solve
Certainly technical problem lower there are precision to the judgment method of lymphatic metastasis at present.It below will be by embodiment and in conjunction with attached drawing
Specifically it is described in detail to how the technical solution of the technical solution of the application and the application solves above-mentioned technical problem.
These specific embodiments can be combined with each other below, may be in some embodiments for the same or similar concept or process
In repeat no more.It should be noted that a kind of transfer outcome evaluation method provided by the present application, the executing subject of Fig. 2-Fig. 7 are
Computer equipment, wherein its executing subject can also be transfer outcome evaluation device, and wherein the device can be by software, hard
The mode of part or software and hardware combining is implemented as some or all of of transfer outcome evaluation.
To keep the purposes, technical schemes and advantages of the embodiment of the present application clearer, below in conjunction with the embodiment of the present application
In attached drawing, the technical scheme in the embodiment of the application is clearly and completely described, it is clear that described embodiment is
Some embodiments of the present application, instead of all the embodiments.
In one embodiment, Fig. 2 provides a kind of transfer outcome evaluation method, and what is involved is computers to set for the present embodiment
The standby comprehensive characteristics value that object construction is extracted from the medical image of object construction, and commented according to the comprehensive characteristics value of object construction
Estimate the detailed process of the transfer result of object construction, as shown in Fig. 2, this method comprises:
S101 obtains the medical image of object construction；Wherein, medical image include at least two mode medical image and
Segmentation mask of the object construction in each mode medical image.
In the present embodiment, the medical image of object construction indicates the medical image of transfer result to be assessed, traditional Chinese medicine shadow
As such as can be conventional CT image, MRI image, PET-MRI image, the present embodiment is not limited this.Its traditional Chinese medicine shadow
As including the segmentation mask of the medical image and object construction of both modalities which in each mode medical image, i.e., obtained in this step
The medical image of object construction be target in the medical image and each mode medical image of at least two mode of object construction
The segmentation mask of structure, wherein the segmentation mask of object construction can be in each mode medical image, by each of object construction
By preparatory trained parted pattern, perhaps detection model splits object construction or target mode medical image
After structure detection comes out, segmentation mask of the object construction in each mode medical image is obtained.For example, the medicine of the object construction
Image can be the CT images of cervical lymph node.It should be noted that the object construction medical image of the application step is target
The medical image of at least two mode of structure, the embodiment of the present application are to be illustrated with both modalities which, but actually answering
In, the object construction medical image of available more kinds of mode, the present embodiment is not specifically limited in this embodiment.
Wherein, in practical applications, the mode that computer equipment obtains the medical image of object construction receives user's input
Image acquisition instruction, and according to image acquisition instruction obtain object construction medical image, wherein wrapped in image acquisition instruction
It can be with the information of hard objectives structural medical image containing object construction mark, medical image type or medical image quantity etc..
Specifically, computer equipment can be in the mode for the medical image for obtaining object construction through other dedicated scan equipment pair
Object construction carries out real time scan acquisition, is also possible to obtain the medical image of pre-stored object construction from memory, also
It can be and download from the Internet, the present embodiment does not limit this.
S102 extracts the comprehensive characteristics value of object construction from the medical image of object construction；Wherein, comprehensive characteristics value packet
Include the image group feature of object construction.
Based on the medical image of the object construction obtained in above-mentioned S101 step, medicine of the computer equipment from object construction
The comprehensive characteristics value of object construction is extracted in image, wherein indicate can be with the concentrated expression object construction characteristic for comprehensive characteristics value
Characteristic value, the image group feature including object construction, optionally, which includes first-order statistical properties, shape
The long matrix of feature, gray level co-occurrence matrixes, gray level, gray scale and in matrix, neighborhood grey scale difference matrix and gray scale independent matrix
At least one.Illustratively, in practical applications, the mode that computer equipment extracts object construction comprehensive characteristics value, which can be, to be passed through
Preset algorithm or the model for extracting characteristic value trained in advance extract, and the present embodiment proposes comprehensive characteristics value
The mode taken without limitation, for example, it may be pyradiomics method carries out characteristic value, characteristic analysis method extracts feature etc..
S103 assesses the transfer result of object construction according to the comprehensive characteristics value of object construction.
In this step, based on the object construction comprehensive characteristics value extracted in above-mentioned S102 step, computer equipment is according to mesh
The transfer of structure composition characteristic value assessment object construction is marked as a result, for example, lymph node by extracting from lymph node CT images
Comprehensive characteristics value, assess whether the lymph node shifts.
Optionally, computer equipment assesses the mode of the transfer result of object construction according to the comprehensive characteristics value of object construction
It can be, the comprehensive characteristics value of object construction is input in transfer classifier, the transfer result of object construction is obtained.Wherein should
Classifier is trained in advance for distinguishing the model whether object construction has transfer phenomena.
A kind of transfer outcome evaluation method provided in this embodiment, the medicine shadow of multiple mode by obtaining object construction
Then picture extracts the comprehensive characteristics value of object construction from each mode medical image, based on the comprehensive characteristics value to object construction
Whether transfer is assessed, in this way, consider the multiple modalities of object construction and the situation of change of comprehensive characteristics value simultaneously, according to
The Multiple factors assessment transfer of object construction is as a result, greatly improve the precision of assessment result.
For the mode of the comprehensive characteristics value of object construction is extracted in above-described embodiment from the medical image of object construction,
The embodiment of the present application provides the mode of two kinds of extraction comprehensive characteristics.Then on the basis of above embodiments, first way such as Fig. 3
Shown, above-mentioned S102 step includes:
S201, from unenhanced medical image extract object construction the first voxel average value, from enhancing medical image in mention
Take object construction the second voxel average value, from enhancing medical image in extract object construction image group feature.
The present embodiment is the case where above-mentioned medical image is unenhanced medical image and enhancing medical image, as shown in figure 4, meter
Calculate machine equipment is extracted from the unenhanced medical image of object construction object construction the first voxel average value, from enhance medical image
It is middle extract object construction the second voxel average value, and from enhancing medical image in extract object construction image group feature.
Optionally, the above-mentioned S201 step includes: the segmentation mask according to object construction in unenhanced medical image, from flat
Sweep the first voxel average value that object construction is extracted in medical image；It is covered according to segmentation of the object construction in enhancing medical image
Mould extracts the second voxel average value of object construction from enhancing medical image.
Wherein, object construction is pin-pointed to the segmentation mask of object construction when extracting object construction voxel average value
Then the voxel average value of object construction is extracted in region from the target area of positioning, facilitate so accurately from flat
It sweeps in medical image and enhancing medical image and extracts voxel average value.Illustratively, computer equipment extracts the first voxel average value
It can be with the mode of the second voxel average value and first obtain each point voxel value from object construction region, then calculate the target knot
The voxel average value of structure region can be obtained the first voxel average value and the second voxel average value, can also use certainly
Other modes, such as inputted in trained algorithm model in advance after the object construction region segmentation is gone out, pass through the algorithm mould
Type extracts the first voxel average value of object construction and the second voxel average value, the present embodiment do not limit this.Based on extraction
After first voxel average value and the second voxel average value, the image group spy of object construction is extracted from enhancing medical image
Sign, wherein the image group feature that computer equipment extracts object construction from enhancing medical image can be unites comprising single order
Count feature, shape feature, gray level co-occurrence matrixes, the long matrix of gray level, gray scale and with matrix, neighborhood grey scale difference matrix and gray scale
At least one in independent matrix, wherein first-order statistical properties reflect the intensity profile of lesion, comprising: energy, entropy, the degree of bias, peak
Degree, variance, mean absolute deviation etc.；Shape feature has reacted the global shape of lesion, comprising: the unbalanced degree of compactedness, spherical shape,
Sphericity, volume, surface area etc.；Textural characteristics describe the spatial distribution of the gray-scale intensity of lesion, by gray level co-occurrence matrixes feature,
Gray scale run-length matrix feature composition, such as 103 category features can be extracted in total, in practical applications, computer equipment extracts shadow
Mode as organizing feature can be extracted using common pyradiomics method, can also be mentioned using method for feature analysis
It takes, the present embodiment does not limit this.
S202 screens goal description feature from image group feature；Wherein goal description character representation and object construction
The degree of correlation between actual characteristic is greater than the feature of the first preset threshold.
Image group feature based on the object construction that above-mentioned S201 step is extracted from enhancing medical image, computer are set
It is standby that goal description feature is screened from image group feature, wherein between goal description character representation and object construction actual characteristic
The degree of correlation be greater than the first preset threshold feature, that is, goal description character representation and object construction actual characteristic it is closest, close
The strongest feature of connection property, filtering out closest, the strongest feature of relevance from image group feature in this way can greatly enhance
Effect of the feature as transfer assessment result.
Optionally, as shown in figure 5, screening a kind of achievable mode packet of goal description feature from image group feature
It includes:
Image group label and image group feature are compared, determine image group feature and image group by S301
The degree of correlation between label；Image group label is the label obtained previously according to object construction pathological diagnosis result.
Wherein, the image group label that image group tag representation is constructed previously according to object construction pathological diagnosis result,
Illustrate the actual feature of object construction, based on the image group label constructed in advance, computer equipment is by image group
Label is compared with image group feature, is determined related between each feature and image group label in image group feature
The corresponding degree of correlation of each feature in image group feature can be obtained in degree.
The feature that the degree of correlation in image group feature is greater than the first preset threshold is determined as goal description feature by S302.
Based on the corresponding degree of correlation of each feature obtained in above-mentioned S301 step, the degree of correlation is greater than the first preset threshold
Feature is determined as goal description feature.The present embodiment in practical applications, can pass through lasso (Least absolute
Extraction method is estimated in shrinkage and selection operator, a kind of compression) method filters out and image group label phase
Highest 10 features of relationship are as goal description feature.
The difference of first voxel average value and the second voxel average value, goal description feature are determined as object construction by S203
Comprehensive characteristics value.
In this step, by difference, the Yi Jishang of the first voxel average value extracted in above-mentioned steps and the second voxel average value
State the comprehensive characteristics value that the goal description feature filtered out is determined as object construction.Illustratively, in conjunction in above-mentioned S302 step
Citing, filter out with highest 10 features of image group label phase relation as goal description feature after, by enhanced CT with
The difference of the voxel average value (the first voxel average value and the second voxel average value) of plain CT, finally should as the 11st feature
Comprehensive characteristics value of 11 features as object construction.
A kind of transfer outcome evaluation method provided in this embodiment, due to by the enhanced CT of object construction and plain CT simultaneously
Considered, and extract different characteristic from plain CT and enhanced CT respectively, finally filter out comprehensive characteristics value, in this way, will
The foundation of the situation of change judgement of the voxel value of enhanced CT and plain CT, and extracted and reality from the image group feature of enhanced CT
Feature association stronger feature in border determines comprehensive characteristics value from multi-faceted angle, ensure that the comprehensive characteristics of acquisition significantly
The comprehensive and accuracy of value.
Equally, the comprehensive characteristics of object construction are extracted on the basis based on Fig. 2 embodiment from the medical image of object construction
The second way of value is as shown in fig. 6, then above-mentioned S102 step includes:
S401 extracts the image group feature of object construction in each mode medical image respectively, obtains the first image group
Feature and the second image group feature.
The case where the present embodiment is the medical image any two mode of object construction, extracts each mode medical image respectively
The image group feature of middle object construction obtains the first image group feature and the second image group feature, with enhanced CT peace
For sweeping CT, that is, need to extract image group feature from enhanced CT and plain CT.
S402, by each feature difference between the first image group feature and the second image group feature, with image group
Feature tag compares, and determines the degree of correlation between each feature difference and image group feature tag.
Based on the first image group feature and the second image group feature extracted in above-mentioned S401 step, obtained in this step
It takes the difference in this two group images group feature between each characteristic value, and each feature difference and image group feature tag is carried out pair
Than obtaining the degree of correlation between each feature difference and image group feature tag.
S403 will be greater than the target in the feature difference and the second image group feature of the second preset threshold in the degree of correlation
Expressive Features are determined as the comprehensive characteristics value of object construction；The second shadow of goal description character representation in second image group feature
Feature as the degree of correlation in group feature between object construction actual characteristic greater than preset critical, wherein the second default threshold
The meaning that value, above-mentioned first preset threshold and preset critical indicate is the same, and three is preset, and three's value
Can be the same, can also be different, specific value can according to the actual situation depending on, such as 80%, 90% etc., this implementation
Example does not limit this.
Based on the degree of correlation between the obtained each characteristic value difference of above-mentioned S402 step and image group feature tag, by phase
Guan Du is greater than the feature difference of the second preset threshold and is determined as mesh from the goal description feature in the second image group feature
Mark the comprehensive characteristics value of structure, wherein extracted shown in mode and above-mentioned Fig. 3 of goal description feature in the second image group feature
First way in extract goal description feature mode it is consistent, naturally it is also possible to other modes are used, as long as from the second shadow
It is learned as group and determines that the degree of correlation between object construction actual characteristic is greater than the feature of preset critical in feature.It needs
Illustrate, the stronger mode of information representation in each mode medical image of the second image group character representation in the present embodiment
Medical image, i.e., if each mode medical image includes plain CT and enhanced CT, the second image group feature is from enhanced CT
Middle extraction.
In addition, the embodiment of the present application provides a kind of transfer outcome evaluation for the training process of above-mentioned transfer classifier
Method, the then as shown in fig. 7, training process of above-mentioned transfer classifier includes:
S501, the corresponding each structure of the sample medical image of the sample medical image, each structure that obtain multiple structures turn
Move result.
In the present embodiment, the sample medical image of the sample medical image and each structure that obtain multiple structures is corresponding
As a result, i.e. standard results, the data obtained in this step are the training datas as transfer classifier for the transfer of each structure, because
This each data needs data that are a large amount of, can representing a variety of situations.
S502, according to the transfer of the sample medical image of multiple structures and each structure as a result, to initial transfer classifier into
Row training obtains transfer classifier.
Based on the training data obtained in above-mentioned S501 step, by the sample medical image of multiple structure and each structure
Transfer is as a result, be trained initial transfer classifier, until training to obtain transfer classifier, in practical applications,
Since the tubercle mark with pathological diagnosis result is more few, that is, the transfer result for wanting to obtain a large amount of structures is relatively difficult, because
Classifier whether this this programme can be using Gaussian process training tubercle transfer, Gaussian process is sentenced based on kernel function and probability
Other Naive Bayes machine learning model can be effectively applied to and solve the problems, such as to return, classify, relative to other kernel functions point
For class device, the advantage of Gaussian process classifier is that output is probability rather than the value of determination using probabilistic model, and high
This classifier is | || non-parameter model, i.e., researcher does not need to manually select the parameter of Gaussian classifier, and Gaussian process classifier exists
When running Gaussian process model, parameter can automatically obtain in the solution procedure of algorithm, therefore, using Gaussian process classifier
Can prediction according to existing sample to result, it is more effective for a small amount of sample data in this way.
Transfer outcome evaluation method provided in this embodiment obtains the training data of training transfer classifier, and root in advance
Transfer classifier is trained according to training data, in this way, the purposive training data for meeting practical scene that obtains is to transfer
Classifier is trained, and can to obtain transfer classifier and is more in line with usage scenario, and the result of judgement is also more accurate, and
It in order to avoid the less situation of training data, is trained using Gaussian process, further enhances the essence of transfer classifier
True property.
It should be understood that although each step in the flow chart of Fig. 2-7 is successively shown according to the instruction of arrow,
These steps are not that the inevitable sequence according to arrow instruction successively executes.Unless expressly stating otherwise herein, these steps
Execution there is no stringent sequences to limit, these steps can execute in other order.Moreover, at least one in Fig. 2-7
Part steps may include that perhaps these sub-steps of multiple stages or stage are not necessarily in synchronization to multiple sub-steps
Completion is executed, but can be executed at different times, the execution sequence in these sub-steps or stage is also not necessarily successively
It carries out, but can be at least part of the sub-step or stage of other steps or other steps in turn or alternately
It executes.
In one embodiment, as shown in figure 8, providing a kind of transfer outcome evaluation device, which includes: acquisition mould
Block 10, extraction module 11 and evaluation module 12, wherein
Module 10 is obtained, for obtaining the medical image of object construction；Wherein, medical image includes at least two mode
The segmentation mask of medical image and object construction in each mode medical image；
Extraction module 11, for extracting the comprehensive characteristics value of object construction from the medical image of object construction；Wherein, comprehensive
Close the image group feature that characteristic value includes object construction；
Evaluation module 12, for assessing the transfer result of object construction according to the comprehensive characteristics value of object construction.
A kind of transfer outcome evaluation device provided by the above embodiment, implementing principle and technical effect and the above method are real
It is similar to apply example, details are not described herein.
In one embodiment, as shown in figure 9, providing a kind of transfer outcome evaluation device, said extracted module 11 is wrapped
It includes: the first extraction unit 111, screening unit 112 and the first comprehensive characteristics unit 113, wherein
First extraction unit 111, for extracting the first voxel average value of object construction from unenhanced medical image, from increasing
The image group extracted the second voxel average value of object construction in strong medical image, extract from enhancing medical image object construction
Learn feature；
Screening unit 112, for screening goal description feature from image group feature；Wherein goal description character representation
The degree of correlation between object construction actual characteristic is greater than the feature of the first preset threshold；
Comprehensive characteristics unit 113 is used for the difference of the first voxel average value and the second voxel average value, goal description feature
It is determined as the comprehensive characteristics value of object construction.
A kind of transfer outcome evaluation device provided by the above embodiment, implementing principle and technical effect and the above method are real
It is similar to apply example, details are not described herein.
In one embodiment, as shown in Figure 10, a kind of transfer outcome evaluation device, above-mentioned screening unit 112 are provided
It include: degree of correlation subelement 1121 and target signature subelement 1122, wherein
Degree of correlation subelement 1121 determines image group for comparing image group label and image group feature
Learn the degree of correlation between feature and image group label；Image group label is to obtain previously according to object construction pathological diagnosis result
The label obtained；
Target signature subelement 1122, the feature for the degree of correlation in image group feature to be greater than the first preset threshold are true
It is set to goal description feature.
A kind of transfer outcome evaluation device provided by the above embodiment, implementing principle and technical effect and the above method are real
It is similar to apply example, details are not described herein.
In one embodiment, said extracted unit 111 is specifically used for according to object construction in unenhanced medical image
Segmentation mask extracts the first voxel average value of object construction from unenhanced medical image；According to object construction in enhancing medicine
Segmentation mask in image extracts the second voxel average value of object construction from enhancing medical image.
A kind of transfer outcome evaluation device provided by the above embodiment, implementing principle and technical effect and the above method are real
It is similar to apply example, details are not described herein.
In one embodiment, as shown in figure 11, a kind of transfer outcome evaluation device is provided, said extracted module 11 is wrapped
It includes: the second extraction unit 114, correlation unit 115 and the second comprehensive characteristics unit 116, wherein
Second extraction unit 114 is obtained for extracting the image group feature of object construction in each mode medical image respectively
To the first image group feature and the second image group feature；
Correlation unit 115, for each feature between the first image group feature and the second image group feature is poor
Value, compares with image group feature tag, determines the degree of correlation between each feature difference and image group feature tag；
Second comprehensive characteristics unit 116, for the feature difference and the second shadow of the second preset threshold will to be greater than in the degree of correlation
As group learns the comprehensive characteristics value that the goal description feature in feature is determined as object construction；Target in second image group feature
Expressive Features indicate that the degree of correlation in the second image group feature between object construction actual characteristic is greater than preset critical
Feature.
A kind of transfer outcome evaluation device provided by the above embodiment, implementing principle and technical effect and the above method are real
It is similar to apply example, details are not described herein.
Above-mentioned evaluation module 12 is specifically used for for the comprehensive characteristics value of object construction being input in one of the embodiments,
It shifts in classifier, obtains the transfer result of object construction.
A kind of transfer outcome evaluation device provided by the above embodiment, implementing principle and technical effect and the above method are real
It is similar to apply example, details are not described herein.
In one embodiment, as shown in figure 12, a kind of transfer outcome evaluation device is provided, which includes: sample
Obtain module 13 and training module 14, wherein
Sample acquisition module 13, for obtaining the sample medical image of multiple structures, the sample medical image pair of each structure
The transfer result for each structure answered；
Training module 14, for according to the sample medical image of multiple structures and the transfer of each structure as a result, turning to initial
It moves classifier to be trained, obtains transfer classifier.
Above-mentioned image group feature includes the first-order statistical properties of object construction, shape spy in one of the embodiments,
The long matrix of sign, gray level co-occurrence matrixes, gray level, gray scale and in matrix, neighborhood grey scale difference matrix and gray scale independent matrix extremely
It is one few.
A kind of transfer outcome evaluation device provided by the above embodiment, implementing principle and technical effect and the above method are real
It is similar to apply example, details are not described herein.
Specific about transfer outcome evaluation device limits the limit that may refer to above for transfer outcome evaluation method
Fixed, details are not described herein.Modules in above-mentioned transfer outcome evaluation device can fully or partially through software, hardware and its
Combination is to realize.Above-mentioned each module can be embedded in the form of hardware or independently of in the processor in computer equipment, can also be with
It is 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, internal structure
Figure can be as shown in Figure 1.The computer equipment includes processor, the memory, network interface, display connected by system bus
Screen and input unit.Wherein, the processor of the computer equipment is for providing calculating and control ability.The computer equipment is deposited
Reservoir includes non-volatile memory medium, built-in storage.The non-volatile memory medium is stored with operating system and computer journey
Sequence.The built-in storage provides environment for the operation of operating system and computer program in non-volatile memory medium.The calculating
The network interface of machine equipment is used to communicate with external terminal by network connection.When the computer program is executed by processor with
Realize a kind of transfer outcome evaluation method.The display screen of the computer equipment can be liquid crystal display or electric ink is shown
Screen, the input unit of the computer equipment can be the touch layer covered on display screen, be also possible on computer equipment shell
Key, trace ball or the Trackpad of setting can also be external keyboard, Trackpad or mouse etc..
It will be understood by those skilled in the art that structure shown in Fig. 1, 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 computer equipment, including memory and processor are provided, is stored in memory
Computer program, the processor perform the steps of when executing computer program
Obtain the medical image of object construction；Wherein, medical image includes the medical image and target of at least two mode
Segmentation mask of the structure in each mode medical image；
The comprehensive characteristics value of object construction is extracted from the medical image of object construction；Wherein, comprehensive characteristics value includes mesh
Mark the image group feature of structure；
The transfer result of object construction is assessed according to the comprehensive characteristics value of object construction.
A kind of computer equipment provided by the above embodiment, implementing principle and technical effect and above method embodiment class
Seemingly, details are not described herein.
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 the medical image of object construction；Wherein, medical image includes the medical image and target of at least two mode
Segmentation mask of the structure in each mode medical image；
The comprehensive characteristics value of object construction is extracted from the medical image of object construction；Wherein, comprehensive characteristics value includes mesh
Mark the image group feature of structure；
The transfer result of object construction is assessed according to the comprehensive characteristics value of object construction.
A kind of computer readable storage medium provided by the above embodiment, implementing principle and technical effect and the above method
Embodiment is similar, and 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 above embodiments can be combined arbitrarily, for simplicity of description, not to above-described embodiment
In each technical characteristic it is all possible combination be all described, as long as however, the combination of these technical characteristics be not present lance
Shield all should be considered as described in this specification.
The several embodiments of the application above described embodiment only expresses, 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 concept of this application, various modifications and improvements can be made, these belong to the protection of the application
Range.Therefore, the scope of protection shall be subject to the appended claims for the application patent.
Claims (10)
1. a kind of transfer outcome evaluation method, which is characterized in that the described method includes:
Obtain the medical image of object construction；The medical image includes the medical image and the target knot of at least two mode
Segmentation mask of the structure in each mode medical image；
The comprehensive characteristics value of the object construction is extracted from the medical image of the object construction；The comprehensive characteristics value includes
The image group feature of the object construction；
The transfer result of the object construction is assessed according to the comprehensive characteristics value of the object construction.
2. the method according to claim 1, wherein the medical image includes unenhanced medical image and enhancing doctor
Learn image；
The comprehensive characteristics value that the object construction is extracted from the medical image of the object construction, comprising:
From the first voxel average value of the object construction is extracted in the unenhanced medical image, from the enhancing medical image
It extracts the second voxel average value of the object construction, extract the image group of the object construction from the enhancing medical image
Learn feature；
Goal description feature is screened from the image group feature；The goal description character representation and the object construction are real
The degree of correlation between the feature of border is greater than the feature of the first preset threshold；
The difference of the first voxel average value and the second voxel average value, the goal description feature are determined as the mesh
Mark the comprehensive characteristics value of structure.
3. according to the method described in claim 2, it is characterized in that, described screen goal description from the image group feature
Feature, comprising:
Image group label and the image group feature are compared, determine the image group feature and the image group
Learn the degree of correlation between label；The image group label is the label obtained previously according to object construction pathological diagnosis result；
It is special that the feature that the degree of correlation in the image group feature is greater than first preset threshold is determined as the goal description
Sign.
4. according to the method in claim 2 or 3, which is characterized in that it is described from the unenhanced medical image extract described in
First voxel average value of object construction, the second voxel that the object construction is extracted from the enhancing medical image are average
Value, comprising:
According to segmentation mask of the object construction in the unenhanced medical image, institute is extracted from the unenhanced medical image
State the first voxel average value of object construction；According to segmentation mask of the object construction in enhancing medical image, from enhancing
The second voxel average value of the object construction is extracted in medical image.
5. the method according to claim 1, wherein described extract institute from the medical image of the object construction
State the comprehensive characteristics value of object construction, comprising:
The image group feature for extracting object construction in each mode medical image respectively, obtains the first image group feature and second
Image group feature；
It is special with image group by each feature difference between the first image group feature and the second image group feature
Sign label compares, and determines the degree of correlation between each feature difference and image group feature tag；
It is retouched by the feature difference for being greater than the second preset threshold in the degree of correlation, with the target in the second image group feature
State the comprehensive characteristics value that feature is determined as the object construction；Goal description character representation in the second image group feature
The degree of correlation in the second image group feature between the object construction actual characteristic is greater than the feature of preset critical.
6. the method according to claim 1, wherein described assess according to the comprehensive characteristics value of the object construction
The transfer result of the object construction, comprising:
The comprehensive characteristics value of the object construction is input in transfer classifier, the transfer result of the object construction is obtained.
7. the method according to claim 1, wherein the training process of the transfer classifier includes:
Obtain the transfer knot of the sample medical image of multiple structures, the corresponding each structure of sample medical image of each structure
Fruit；
According to the transfer of the sample medical image of the multiple structure and each structure as a result, being carried out to initial transfer classifier
Training, obtains the transfer classifier.
8. method according to claim 1-7, which is characterized in that the image group feature includes the target
The long matrix of the first-order statistical properties of structure, shape feature, gray level co-occurrence matrixes, gray level, gray scale and with matrix, neighborhood gray scale difference
At least one in sub-matrix and gray scale independent matrix.
9. a kind of computer equipment, including memory and processor, the memory are stored with computer program, feature exists
In the step of processor realizes any one of claims 1 to 8 the method when executing the computer program.
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 described in any item of the claim 1 to 8 is realized when being executed by processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910597572.5A CN110400292B (en) | 2019-07-04 | 2019-07-04 | Transfer result evaluation method, computer device, and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910597572.5A CN110400292B (en) | 2019-07-04 | 2019-07-04 | Transfer result evaluation method, computer device, and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110400292A true CN110400292A (en) | 2019-11-01 |
| CN110400292B CN110400292B (en) | 2021-08-13 |
Family
ID=68323757
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910597572.5A Active CN110400292B (en) | 2019-07-04 | 2019-07-04 | Transfer result evaluation method, computer device, and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110400292B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111583217A (en) * | 2020-04-30 | 2020-08-25 | 深圳开立生物医疗科技股份有限公司 | Tumor ablation curative effect prediction method, device, equipment and computer medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8472684B1 (en) * | 2010-06-09 | 2013-06-25 | Icad, Inc. | Systems and methods for generating fused medical images from multi-parametric, magnetic resonance image data |
| CN103345638A (en) * | 2013-06-24 | 2013-10-09 | 清华大学深圳研究生院 | Cavity focus computer-assisted detecting method based on medical image |
| WO2015010745A1 (en) * | 2013-07-26 | 2015-01-29 | Brainlab Ag | Multi-modal segmentation of image data |
| CN107248162A (en) * | 2017-05-18 | 2017-10-13 | 杭州全景医学影像诊断有限公司 | The method of preparation method and acute cerebral ischemia the image segmentation of acute cerebral ischemia Image Segmentation Model |
| CN108986106A (en) * | 2017-12-15 | 2018-12-11 | 浙江中医药大学 | Retinal vessel automatic division method towards glaucoma clinical diagnosis |
| 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 |
| CN109273084A (en) * | 2018-11-06 | 2019-01-25 | 中山大学附属第医院 | The method and system of feature modeling are learned based on multi-modal ultrasound group |
-
2019
- 2019-07-04 CN CN201910597572.5A patent/CN110400292B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8472684B1 (en) * | 2010-06-09 | 2013-06-25 | Icad, Inc. | Systems and methods for generating fused medical images from multi-parametric, magnetic resonance image data |
| CN103345638A (en) * | 2013-06-24 | 2013-10-09 | 清华大学深圳研究生院 | Cavity focus computer-assisted detecting method based on medical image |
| WO2015010745A1 (en) * | 2013-07-26 | 2015-01-29 | Brainlab Ag | Multi-modal segmentation of image data |
| CN107248162A (en) * | 2017-05-18 | 2017-10-13 | 杭州全景医学影像诊断有限公司 | The method of preparation method and acute cerebral ischemia the image segmentation of acute cerebral ischemia Image Segmentation Model |
| CN108986106A (en) * | 2017-12-15 | 2018-12-11 | 浙江中医药大学 | Retinal vessel automatic division method towards glaucoma clinical diagnosis |
| 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 |
| CN109273084A (en) * | 2018-11-06 | 2019-01-25 | 中山大学附属第医院 | The method and system of feature modeling are learned based on multi-modal ultrasound group |
Non-Patent Citations (1)
| Title |
|---|
| 张玉琴 等: "多模态磁共振成像对33例胆管癌腹腔转移的评估价值", 《中华消化杂志》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111583217A (en) * | 2020-04-30 | 2020-08-25 | 深圳开立生物医疗科技股份有限公司 | Tumor ablation curative effect prediction method, device, equipment and computer medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110400292B (en) | 2021-08-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11610308B2 (en) | Localization and classification of abnormalities in medical images | |
| JP7425145B2 (en) | Systems and methods for comprehensive multi-assay tissue analysis | |
| Silva-Rodríguez et al. | Going deeper through the Gleason scoring scale: An automatic end-to-end system for histology prostate grading and cribriform pattern detection | |
| Madabhushi et al. | Image analysis and machine learning in digital pathology: Challenges and opportunities | |
| JP5814504B2 (en) | Medical image automatic segmentation system, apparatus and processor using statistical model | |
| CN109993726A (en) | Detection method, device, equipment and the storage medium of medical image | |
| JP7168237B2 (en) | Systems and methods for examining regions of interest in hematoxylin and eosin (H&E) stained tissue images to quantify intratumoral cellular spatial heterogeneity in multiplexed/highly multiplexed fluorescence tissue images | |
| Cohen et al. | Memory based active contour algorithm using pixel-level classified images for colon crypt segmentation | |
| Singanamalli et al. | Identifying in vivo DCE MRI markers associated with microvessel architecture and gleason grades of prostate cancer | |
| US11276490B2 (en) | Method and apparatus for classification of lesion based on learning data applying one or more augmentation methods in lesion information augmented patch of medical image | |
| Ni et al. | Segmentation of ultrasound image sequences by combing a novel deep siamese network with a deformable contour model | |
| CN110633651A (en) | Abnormal cell automatic identification method based on image splicing | |
| Chan et al. | Eigentumors for prediction of treatment failure in patients with early-stage breast cancer using dynamic contrast-enhanced MRI: a feasibility study | |
| CN112263217B (en) | Improved convolutional neural network-based non-melanoma skin cancer pathological image lesion area detection method | |
| Mercan et al. | Deep feature representations for variable-sized regions of interest in breast histopathology | |
| Zhang et al. | Novel indices for left-ventricular dyssynchrony characterization based on highly automated segmentation from real-time 3-D echocardiography | |
| CN110400292A (en) | Shift outcome evaluation method, computer equipment and storage medium | |
| CN116051553B (en) | Method and device for marking inside three-dimensional medical model | |
| Ezzat et al. | Automatic liver segmentation in computed tomography scans using deep semantic segmentation | |
| Sharma | Medical image analysis of gastric cancer in digital histopathology: methods, applications and challenges | |
| Li et al. | A Modified Fuzzy Markov Random Field Incorporating Multiple Features for Liver Tumor Segmentation | |
| JP2008000496A (en) | Medical image display device and its program | |
| Gann | Enhanced Visualization of Magnetic Resonance Brain Tumor Segmnetation Volumes from Deep Automated Models | |
| Zhang et al. | Deep learning and radiomics-based approach to meningioma grading: exploring the potential value of peritumoral edema regions | |
| Jonnalagadda | Incorporating Human Visual Properties into Neural Network Models |
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 |