CN114881941A - Spine health change prediction method and system based on big data - Google Patents

Abstract

The present invention provides a method and system for predicting changes in spine health based on big data; wherein, the method includes: acquiring big data of spine images at different health stages, forming a training set based on the big data of spine images, and using the training The depth recognition model is trained in the set; obtain the spine image data of the current health stage, input the spine image data into the trained depth recognition model, and the depth recognition model outputs the spine health prediction atlas; according to the preset strategy The spine health prediction atlas is output. The solution of the present invention can accurately predict and obtain the spine health prediction atlas corresponding to the current health stage of the spine, so that doctors and patients can have a deeper understanding of the real situation of spine health, which is conducive to the determination of a more reasonable diagnosis and treatment plan .

Description

A method and system for predicting changes in spine health based on big data

technical field

The present invention relates to the field of medical technology, in particular, to a method and system for predicting changes in spine health based on big data.

Background technique

After entering the pace of life and work in modern society, more and more people have spinal health problems, such as C-type scoliosis, S-type scoliosis and so on. In order to better assist doctors in the diagnosis and treatment of spinal health problems, related technologies have begun to study automatic identification and labeling of spinal images, such as:

Patent Document 1 (CN114187320A) discloses a method for segmenting a spine CT image, comprising: determining an edge between an estimated foreground area and an estimated background area on the enhanced spine CT image, and A directed graph is drawn for each voxel; according to the directed graph, a minimum cut of the enhanced spine CT image is determined to segment the enhanced spine CT image into actual foreground regions including the spine imaged, including The actual background area of the non-spine image; based on the minimum cut, at least the actual foreground area is determined on the enhanced spine CT image.

Patent Document 2 (CN114170114A) discloses a method for enhancing a spine CT image, which includes: performing morphological expansion processing on the spine CT image to generate a morphological expansion image; performing a morphological erosion process on the morphological expansion image to generate a morphological expansion Close the processed image; according to the morphologically closed processed image, an enhanced spine CT image is generated.

Patent document 3 (CN114081471A) discloses a method for measuring the cobb angle of scoliosis based on three-dimensional images and multi-layer perception. The original three-dimensional image is subjected to normalized interpolation processing, and the back image of the human body to be measured is sampled at equal distances and projected on the three-dimensional coronal coordinate system to obtain a three-dimensional image with regular and fixed sampling points; Cross-cut to obtain a series of back cross-sectional profiles, each back cross-sectional profile corresponding to a sequence of sampling points; S4, perform symmetry analysis on the sampling point sequence of each back cross-sectional profile obtained in step S3, and obtain each back cross-sectional profile. The candidate points of the spine midline on the contour of the transverse plane, select the best candidate point as the marked spine point; S5, construct a multi-layer perceptron, and the multi-layer perceptron performs step S3 and step S4 to mark the spine point training, the multi-layer after training The perceptron can accurately find the spine points in the sampling point sequence of the contours of each back transverse plane; S6, input the three-dimensional image of step S2 through the multilayer perceptron and output the spine points on the three-dimensional image; S7, locate the spine points on the three-dimensional coronal image. Perform curve fitting on the spine points on the plane coordinate system to obtain the spine midline, and calculate the normal vector of the spine point located on the three-dimensional coronal plane coordinate system with respect to the spine midline; S8. Use the normal vector obtained in step S7 to construct an n-dimensional real symmetric matrix , respectively calculate the angle between the two normal vectors, and take the two normal vectors that form the largest angle, and the largest angle is the scoliosis cobb angle.

Based on the above-mentioned related patent documents in the prior art, it can be seen that the prior art mainly focuses on the detection and labeling of certain parameters (such as the cobb angle) in the current spine image for spine health problems, but cannot change the spine health. Predictive output is not good for assisting doctors in related diagnosis and treatment, and it is also not conducive to patients' in-depth understanding of their own spinal health problems.

SUMMARY OF THE INVENTION

In order to at least solve the technical problems existing in the above-mentioned background art, the present invention provides a method, system, electronic device and computer storage medium for predicting changes in spine health based on big data.

A first aspect of the present invention provides a method for predicting changes in spine health based on big data, comprising the following steps:

Acquiring spine image big data at different health stages, forming a training set based on the spine image big data, and using the training set to train a depth recognition model;

Obtain the spine image data of the current health stage, input the spine image data into the trained depth recognition model, and the depth recognition model outputs the spine health prediction atlas;

The spine health prediction atlas is output according to a preset strategy.

Further, the big data of spine images includes first spine images of different health stages;

Then the training set is formed based on the large data of the spine image, including:

Perform first spine parameter extraction processing on each of the first spine images, determine a first attribute of each of the first spine images based on the extracted first spine parameters, and use the first attribute to perform a image annotation;

Fitting processing is performed on each of the first spine images with the same first attribute to obtain a second spine image and a corresponding second spine parameter, and the first spine image of the second spine image is determined according to the second spine parameter. two attributes;

Calculate the first matching degree between the second attribute of each second spine image and the standard attribute stage table, determine the first stage serial number of each second spine image according to the first matching degree, and compare it with each A plurality of the first spine images corresponding to the second spine images are associated with the first stage serial numbers;

Taking the first-stage sequence numbers of the second spine images as a starting point to traverse backwards, and associating a plurality of the first spine images corresponding to the second spine images involved in the traversal with the second-stage sequence numbers;

Several pieces of training data are obtained according to the first spine image and/or the second spine image, and each of the training data is assembled into the training set.

Further, the first spine image is associated with a user attribute, and is associated with a third stage serial number;

Then, associating several of the first spine images corresponding to each of the second spine images with the first stage serial number, including:

Determine at least two of the first spine images based on the third stage sequence number, and calculate a mutation degree according to the first spine parameters of the at least two first spine images;

If the mutation degree is greater than or equal to the first threshold, the corresponding first spine image is deleted, and the remaining several first spine images are associated with the first stage serial number.

Further, after the degree of mutation is greater than or equal to the first threshold, the method further includes:

Calculate whether the mutation degree is less than the second threshold, and if so, then:

Based on the first spine parameters of the first spine image adjacent to the first spine image, a number of first mechanical models are determined to calculate the first spine parameters of the first spine image and the corresponding first spine parameters. The correlation value of the mechanical model;

If the correlation degree value is greater than or equal to the third threshold, it is determined that the mutation degree is greater than or equal to the first threshold, and the corresponding first spine image is deleted; otherwise, it is determined that the mutation degree is less than the first threshold, and the The corresponding first spine image is deleted.

Further, the third threshold can be determined as follows:

A plurality of second mechanical models are determined based on the first spine parameters of the first spine image adjacent to the second front of the first spine image, and the relationship between the second mechanical model and the corresponding first mechanical model is calculated. a first degree of similarity, and the third threshold is determined based on the first degree of similarity;

Wherein, the third threshold is negatively correlated with the first similarity.

Further, the described spine image data is input into the trained depth recognition model, and the depth recognition model outputs the spine health prediction atlas, including:

Calculate the third spine parameter corresponding to the spine image data, input the third spine parameter into the trained depth recognition model, the depth recognition model obtains an initial prediction atlas, and the initial prediction atlas includes Several second spine images and corresponding fourth stage serial numbers;

calculating a second similarity of the first spine parameter corresponding to each of the first spine images associated with the third spine parameter and the second spine image;

If the second similarity is greater than or equal to a fourth threshold, the first spine image associated with the second spine image is used as a target prediction image, and the target corresponding to each second spine image After the predicted image is correspondingly associated with the fourth stage serial number, the spine health prediction atlas is obtained;

The depth recognition model outputs the spine health prediction atlas.

Further, outputting the spine health prediction atlas according to a preset strategy includes:

Calculate the second matching degree of the third attribute corresponding to the spine image data of the current healthy stage and the standard attribute stage table, determine the fifth stage serial number based on the second matching degree, and determine the fifth stage serial number according to the fifth stage serial number. the number of images in the spine health prediction atlas;

The sequence number of the key stage is determined according to the third attribute, and the number of images in the key stage is corrected according to the sequence number of the key stage;

Wherein, the correction degree is positively correlated with the criticality value of the key stage.

A second aspect of the present invention provides a system for predicting changes in spine health based on big data, comprising a first acquisition module, a second acquisition module, a processing module, a storage module, and an output module, the processing module and the first acquisition module module, the second acquisition module, the storage module and the output module are connected; wherein,

The first acquisition module is used to acquire large data of spine images at different health stages and transmit them to the processing module;

The second acquisition module is used to acquire the spine image data of the current health stage and transmit it to the processing module;

the storage module for storing executable computer program codes;

The output module is used to output the spine health prediction atlas according to a preset strategy;

The processing module is configured to execute the method described in any preceding item by calling the executable computer program code in the storage module.

A third aspect of the present invention provides an electronic device, comprising:

memory in which executable program code is stored;

a processor coupled to the memory;

The processor invokes the executable program code stored in the memory to execute the method described in any preceding item.

A fourth aspect of the present invention provides a computer storage medium, where a computer program is stored thereon, and the computer program is executed by a processor to execute the method described in any one of the above.

The solution of the present invention is to obtain large data of spine images in different health stages, build a training set based on the large data of spine images, and use the training set to train a depth recognition model; The spine image data is input into the trained depth recognition model, and the depth recognition model outputs a spine health prediction atlas; the spine health prediction atlas is output according to a preset strategy. The solution of the present invention can accurately predict and obtain the spine health prediction atlas corresponding to the current health stage of the spine, so that doctors and patients can more deeply understand the real situation of spine health, which is conducive to the determination of a more reasonable diagnosis and treatment plan .

Description of drawings

In order to illustrate the technical solutions of the embodiments of the present invention more clearly, the following briefly introduces the accompanying drawings used in the embodiments. It should be understood that the following drawings only show some embodiments of the present invention, and therefore do not It should be regarded as a limitation of the scope, and for those of ordinary skill in the art, other related drawings can also be obtained according to these drawings without any creative effort.

1 is a schematic flowchart of a method for predicting changes in spine health based on big data disclosed in an embodiment of the present invention;

2 is a schematic diagram of obtaining a virtual and equivalent second spine image based on the first spine image fitting disclosed in an embodiment of the present invention;

3 is a schematic diagram of a conversion relationship between a first-stage serial number and a second-stage serial number disclosed in an embodiment of the present invention;

4 is a schematic diagram of an output mode of a spine health prediction atlas disclosed in an embodiment of the present invention;

5 is a schematic structural diagram of a system for predicting changes in spine health based on big data disclosed in an embodiment of the present invention;

FIG. 6 is a schematic structural diagram of an electronic device disclosed in an embodiment of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

The words "first, second, third, etc." in the description and claims, or similar terms such as module A, module B, module C, etc., are only used to distinguish similar objects, and do not represent a specific ordering of objects, which can be understood Where permitted, the specific order or sequence may be interchanged to enable the embodiments of the invention described herein to be practiced in sequences other than those illustrated or described herein.

In the following description, the reference numerals representing steps, such as S110, S120, etc., do not necessarily mean that this step will be performed, and the order of the preceding and following steps may be interchanged or performed simultaneously if permitted.

The term "comprising" used in the description and claims should not be interpreted as being limited to what is listed thereafter; it does not exclude other elements or steps. Accordingly, it should be interpreted as specifying the presence of said features, integers, steps or components mentioned, but not excluding the presence or addition of one or more other features, integers, steps or components and groups thereof. Therefore, the expression "apparatus comprising means A and B" should not be limited to apparatuses consisting of parts A and B only.

Reference in this specification to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Thus, the appearances of the terms "in one embodiment" or "in an embodiment" in various places in this specification are not necessarily all referring to the same embodiment, but can refer to the same embodiment. Furthermore, the particular features, structures or characteristics can be combined in any suitable manner, as would be apparent to one of ordinary skill in the art from this disclosure, in one or more embodiments.

Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. If there is any inconsistency, the meaning described in this specification or the meaning derived from the content described in this specification shall prevail. Also, the terminology used herein is for the purpose of describing the embodiments of the present invention only, and is not intended to limit the present invention.

Example 1

Please refer to FIG. 1. FIG. 1 is a schematic flowchart of a method for predicting changes in spine health based on big data disclosed in an embodiment of the present invention. As shown in FIG. 1 , a method for predicting changes in spine health based on big data according to an embodiment of the present invention includes the following steps:

Acquiring spine image big data at different health stages, forming a training set based on the spine image big data, and using the training set to train a depth recognition model;

Obtain the spine image data of the current health stage, input the spine image data into the trained depth recognition model, and the depth recognition model outputs the spine health prediction atlas;

The spine health prediction atlas is output according to a preset strategy.

In this step, as described in the background art, in practice, both the doctor and the patient have needs for the follow-up development of the current spine health, while the prior art can only detect images (such as X-ray images) of the current spine. The current spinal health status is also obtained by the detection. Obviously, the existing technology cannot accurately predict the development and changes of the spinal health status, resulting in the inability to assist doctors to obtain better diagnosis and treatment results.

In view of this, the present invention collects spine images at different health stages to obtain large data of spine images, and then trains a depth recognition model based on a training set obtained from them, so that the depth recognition model is established and can be used to accurately describe changes in spine health. Function relationship, in practical application, input the captured spine image data of the current health stage into the trained depth recognition model, and then the corresponding spine health prediction atlas can be obtained, and these atlases can be output to enable doctors, The patient has a deeper understanding of the true spinal health status, which is conducive to the accurate determination of the follow-up treatment plan.

Among them, the deep recognition model can be built by a variety of intelligent prediction algorithms existing in the prior art, such as Deep Belief Network (DBN), Recurrent Neural Network (RNN), long/short-term memory (Long/Short Term Memory (LSTM)), Gated Recurrent Unit (GRU), Auto Encoder (AE), Hopfield Network (HN), Deep Belief network (Deep Belief Network (DBN)), etc., which is not limited in the present invention.

It should be noted that the solution of the present invention can be implemented in multiple ways, mainly including a field terminal and a server. The field terminal can be a terminal with wireless connection function and network connection function, such as user equipment such as smart phone, tablet computer, desktop computer, notebook computer, super personal computer and wearable device; and the server can be an independent physical server , or it can provide cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), big data and Cloud servers for basic cloud computing services such as artificial intelligence platforms.

Further, the big data of spine images includes first spine images of different health stages;

Then the training set is formed based on the large data of the spine image, including:

Perform first spine parameter extraction processing on each of the first spine images, determine a first attribute of each of the first spine images based on the extracted first spine parameters, and use the first attribute to perform a image annotation;

Fitting processing is performed on each of the first spine images with the same first attribute to obtain a second spine image and a corresponding second spine parameter, and the first spine image of the second spine image is determined according to the second spine parameter. two attributes;

Calculate the first matching degree between the second attribute of each second spine image and the standard attribute stage table, determine the first stage serial number of each second spine image according to the first matching degree, and compare it with each A plurality of the first spine images corresponding to the second spine images are associated with the first stage serial numbers;

Taking the first-stage sequence numbers of the second spine images as a starting point to traverse backwards, and associating a plurality of the first spine images corresponding to the second spine images involved in the traversal with the second-stage sequence numbers;

Several pieces of training data are obtained according to the first spine image and/or the second spine image, and each of the training data is assembled into the training set.

In this step, the first attribute of each first spine image is determined according to the spine parameters of the first spine image, and the first attribute is used to describe the health status of the spine, including but not limited to whether the scoliosis is curved or not, the number of scoliosis segments, the scoliosis direction, the side curvature Bend out the cobb angle, etc.; refer to Figure 2, and then perform fitting processing on each first spine image with the same first attribute (for example, 1.1-1.3 shown in Figure 2), and then obtain a virtual, etc. A valid second spine image (for example, 2.1 shown in Figure 2), similar to the previous one, and then determine its second attribute based on the second spine parameters of the second spine image; meanwhile, the standard can be preset based on artificial experience Attribute stage table, the standard attribute stage table can describe the stage attributes of spine health, and then through the matching calculation, the serial number of the health stage in which each equivalent second spine image is located can be determined; The spine image is traversed backwards from the starting point, that is, the whole data after a certain health stage is regarded as a training data. In this way, several training data starting from different health stages can be obtained, and then a training set can be formed. , that is, the first spine image, the second spine image, or both can be used simultaneously; correspondingly, the depth recognition model obtained by training also establishes a plurality of the aforementioned functional relationships. Wherein, in addition to the corresponding first spine image and/or second spine image, the training data may also include first spine parameters and first attributes corresponding to the first spine image, and second spine parameters corresponding to the second spine image and the second attribute, set in this way, the deep recognition model can avoid repeated parameter extraction.

It can be seen from the above analysis that when determining the training set, the present invention determines the real health stage corresponding to each first spine image through two-step processing. This setting can solve the existence of the first spine parameter of each first spine image. The problem of accurate identification of the health stage when there is a slight difference; and, since the input current health status of the patient's spine is quite different, the present invention also takes the second spine images of different health stages as a starting point to traverse backward to obtain the corresponding training. The data, set in this way, can obtain the functional relationship corresponding to different health stages, making the predicted spine health prediction atlas more accurate.

It should be noted that the standard attribute stage table is an attribute comparison table determined from the healthy spine as the starting point and the extreme severe spine as the end point. Referring to Figure 3, the first stage serial number is in one-to-one correspondence with the standard attribute stage table. The sequence number of the second stage is for the renumbering of the traversal result. For example, when the sequence number of the first stage is 5, but it is located in the 4th position of the traversal result, the sequence number of the second stage is 4.

Further, the first spine image is associated with a user attribute, and is associated with a third stage serial number;

Then, associating several of the first spine images corresponding to each of the second spine images with the first stage serial number, including:

Determine at least two of the first spine images based on the third stage sequence number, and calculate a mutation degree according to the first spine parameters of the at least two first spine images;

If the mutation degree is greater than or equal to the first threshold, the corresponding first spine image is deleted, and the remaining several first spine images are associated with the first stage serial number.

In this step, the collected first spine images can be associated with each user and marked with serial numbers, and the spine health of a single user should be gradually developed. Therefore, the present invention calculates the distance between adjacent second spine images by calculating If the mutation degree is higher, those second spine images with too high mutation are deleted to make the training set more accurate.

An example is as follows:

There are 20 first spine images of user A collected, and the degree of mutation between adjacent first spine images is calculated one by one according to the first spine parameters (such as the cobb angle), for example, the 9th and 10th first spine images are found If the mutation between the images is too large, it means that the 10th first spine image is abnormal. For example, the first spine image of user B is incorrectly associated with user A, or the first spine parameter of user A is abnormal due to other than traffic. "mutation". At this time, in order for the abnormally "mutated" first spine image to affect the establishment of the aforementioned functional relationship, it should be deleted.

It should be noted that the degree of mutation may be determined based on the difference, change trend, similarity, and the like between the first spine parameters of adjacent first spine images. For the similarity among them, for example, the 9th image is C-type scoliosis, and the 10th image is S-type scoliosis. At this time, the similarity is low, and the corresponding mutation degree is high. The similarity calculation involved here can be obtained based on cosine similarity, Pearson correlation coefficient, Tanimoto coefficient, Hamming distance, etc., which is not limited in the present invention.

Further, after the degree of mutation is greater than or equal to the first threshold, the method further includes:

Calculate whether the mutation degree is less than the second threshold, and if so, then:

Based on the first spine parameters of the first spine image adjacent to the first spine image, a number of first mechanical models are determined to calculate the first spine parameters of the first spine image and the corresponding first spine parameters. The correlation degree value of the mechanical model;

If the correlation degree value is greater than or equal to the third threshold, it is determined that the mutation degree is greater than or equal to the first threshold, and the corresponding first spine image is deleted; otherwise, it is determined that the mutation degree is less than the first threshold, and the The corresponding first spine image is deleted.

In this improvement step, although the overall change in the health of the spine occurs gradually, there is also a rapid deterioration in some stages, and when the sampling interval is long, the difference between the adjacent first spine images It is also more likely to have large differences in health status between countries. In view of this, the present invention further analyzes several first mechanical models (mainly for the part where scoliosis has already occurred) based on the first spine parameters in the adjacent images before the “mutated” first spine image, and then analyzes the The correlation value between the first spine parameter in the “mutated” first spine image and the mechanical model (wherein, the mechanical model is the cause, the first spine parameter corresponding to the “mutation” is the result, and the correlation value is the cause and the result When the correlation degree is higher than the third threshold, it can be determined that the "mutation" is a local deterioration of the disease development, which is normal; otherwise, it can be determined that the "mutation" may be caused by other factors, not are normal pathological changes.

It should be noted that the second threshold is used to describe the intermediate situation where the mutation degree is greater than or equal to the first threshold, but has not exceeded the obvious limit.

Further, the third threshold can be determined as follows:

A plurality of second mechanical models are determined based on the first spine parameters of the first spine image adjacent to the second front of the first spine image, and the relationship between the second mechanical model and the corresponding first mechanical model is calculated. a first degree of similarity, and the third threshold is determined based on the first degree of similarity;

Wherein, the third threshold is negatively correlated with the first similarity.

In this step, the present invention further calculates the first similarity of the corresponding mechanical models between the two images before the “mutated” first spine image, and when the first similarity is lower, that is, the difference between the two mechanical models is larger. , indicating that the “mutation”, that is, the more obvious the precursor of local deterioration, the smaller the third threshold is set at this time, that is, the judgment sensitivity of the “mutation” of the “mutated” first spine image is improved, and the correlation degree value is further improved. Calculate accuracy.

Further, when using the training set to train the depth recognition model, the following loss function is used:

In the formula, L represents the loss function of the deep recognition model during the training process, M represents the current number of training iterations, F _i represents the first equivalent feature matrix output by the deep recognition model after receiving the i-th round of training data, f _i Represents the second equivalent feature matrix of the i-th round of training data received by the depth recognition model, D(F _i , f _i ) represents the distance between F _i and f _i (such as Euclidean distance, cosine distance, etc.); F _j represents After receiving the i-th round of training data, the output feature matrix corresponding to j≠y _i , where j≠y _i represents that the output feature matrix exceeds the set range of the i-th round of training, correspondingly, D(F _j , f _i ) represents the distance between F _j and f _i (eg Euclidean distance, cosine distance, etc.); m is the adjustment base. Among them, the equivalent matrix is obtained based on the feature matrix fitting of multiple input data and output data.

In this step, the loss function directly affects the training result of the depth recognition model. The present invention designs the above loss function, wherein the training data is input to the depth recognition model for training according to the rounds, and after each round of training, the The first equivalent feature matrix within the set range and each feature matrix beyond the set range output in this round (wherein, the depth recognition model has multiple outputs in each training round), and then use the above formula The distance between the failed data and the successful data in the training results of this round can be obtained. When the distance is close enough, that is, the failed data is closely surrounding the successful data center, it means that the training meets the requirements and the training can be ended.

Further, the described spine image data is input into the trained depth recognition model, and the depth recognition model outputs the spine health prediction atlas, including:

Calculate the third spine parameter corresponding to the spine image data, input the third spine parameter into the trained depth recognition model, the depth recognition model obtains an initial prediction atlas, and the initial prediction atlas includes Several second spine images and corresponding fourth stage serial numbers;

calculating a second similarity of the first spine parameter corresponding to each of the first spine images associated with the third spine parameter and the second spine image;

If the second similarity is greater than or equal to a fourth threshold, the first spine image associated with the second spine image is used as a target prediction image, and the target corresponding to each second spine image After the predicted image is correspondingly associated with the fourth stage serial number, the spine health prediction atlas is obtained;

The depth recognition model outputs the spine health prediction atlas.

In this step, the depth recognition model of the present invention first screens out the corresponding second spine images and their corresponding serial numbers based on the second spine parameters corresponding to the spine image data and the functional relationship obtained by training, and then based on the second similarity The degree calculation selects the first spine image associated with the second spine image, wherein the filtered first spine image is used as the final output target prediction image. The invention sets the depth recognition model into two parts, and the two parts work in an orderly manner, which can improve the calculation efficiency of the depth recognition model. There may be multiple first spine images satisfying the second similarity greater than or equal to the fourth threshold, and the output mode in this case will be described later.

Further, the second similarity is calculated by the following formula:

In the formula, s(data _i ) represents the similarity between the second spine parameter corresponding to the spine image data and the first spine parameter of the i-th first spine image associated with the second spine image. degree; d(data _i ) represents the similarity value between the second spine parameter and the first spine parameter calculated by using the benchmark similarity algorithm; d _min , d _max represent the calculation using different similarity algorithms The minimum value and the maximum value among the similarity values between the second spine parameter and the first spine parameter; k represents the number of the first spine image associated with the second spine image, that is, the first spine image. The number of spine parameters.

Among them, the present invention comprehensively uses a variety of traditional similarity algorithms to obtain a more accurate second similarity calculation result, for example, the reference similarity algorithm can be the Euclidean distance algorithm, and other similarity algorithms can be cosine similarity, Pearson correlation coefficient, Tanimoto coefficient, Hamming distance, etc., so different similarity algorithms are used to obtain the corresponding similarity value, and then a more accurate second similarity value is obtained based on the above similarity fusion formula.

Further, outputting the spine health prediction atlas according to a preset strategy includes:

Calculate the second matching degree of the third attribute corresponding to the spine image data of the current healthy stage and the standard attribute stage table, determine the fifth stage serial number based on the second matching degree, and determine the fifth stage serial number according to the fifth stage serial number. the number of images in the spine health prediction atlas;

The sequence number of the key stage is determined according to the third attribute, and the number of images in the key stage is corrected according to the sequence number of the key stage;

Wherein, the correction degree is positively correlated with the criticality value of the key stage.

In this step, the spines of different health stages have different concerns, so the present invention first determines the fifth stage serial number corresponding to the spine image data of the current health stage in a similar manner to the above, that is, to determine the currently input spine According to which health stage the image data belongs to, the appropriate number of images in the spine health prediction atlas can be determined, and the number of images in the key stage can also be properly corrected (for example, increasing the number of images), and the greater the criticality value of the key stage. higher, the higher the number of images to set the key stage.

Among them, when the current monitoring stage is in the early stage (that is, the spine is slightly curved), it is necessary to output a relatively sparse and holistic spine health prediction atlas that can describe the changes in spine health, so that the patient can at least know The need for scoliosis treatment. When the current monitoring stage is in the middle stage, especially when the later stage of development is difficult for surgical treatment, a more detailed and larger number of spine health prediction atlases should be output. At the same time, the middle stage will also correspond to In a critical stage of treatment (eg, the aforementioned optimal surgical treatment period, or the last surgical treatment period), the number of images in the spine health prediction atlas should also be appropriately increased in this critical stage. Of course, the above is only for illustration, and is not intended to limit the protection scope of the present invention. Other situations can be set correspondingly, and the present invention will not be repeated here. In addition, for key stages, the relationship between different current monitoring stages and the corresponding key stages can be established based on human experience, and can also be predicted by building a deep recognition model.

It should be noted that the spine health prediction atlas can be output in the form of conventional static images or GIF animations or videos, and specifically can be output based on the time axis. Of course, the present invention does not exclude other output forms, which will not be repeated here. . In addition, as shown in FIG. 4 , when there may be multiple first spine images satisfying the second similarity greater than or equal to the fourth threshold, when longitudinal output is performed based on the time axis, multiple images are output simultaneously or separately at the corresponding position horizontally. of the first spine image.

Embodiment 2

Please refer to FIG. 5 , which is a schematic structural diagram of a system for predicting changes in spine health based on big data disclosed in an embodiment of the present invention. As shown in FIG. 5 , a system for predicting changes in spine health based on big data according to an embodiment of the present invention includes a first acquisition module (101), a second acquisition module (102), a processing module (103), and a storage module (104) ), an output module (105), the processing module (101) and the first acquisition module (101), the second acquisition module (102), the storage module (104) and the output module (105) ) connection; where,

The first acquisition module (101) is used to acquire large data of spine images at different health stages, and transmit it to the processing module (103);

The second acquisition module (102) is used to acquire the spine image data of the current healthy stage, and transmit it to the processing module (103);

the storage module (104) for storing executable computer program codes;

The output module (103) is used to output the spine health prediction atlas according to a preset strategy;

The processing module (103) is configured to execute the method according to the first embodiment by calling the executable computer program code in the storage module (104).

For the specific function of a system for predicting changes in spinal health based on big data in this embodiment, refer to the above-mentioned first embodiment. Since the system in this embodiment adopts all the technical solutions of the above-mentioned embodiments, it has at least the technical solutions of the above-mentioned embodiments. All the beneficial effects brought by the scheme will not be repeated here.

Embodiment 3

Please refer to FIG. 6. FIG. 6 is an electronic device disclosed in an embodiment of the present invention, including: a memory storing executable program codes; a processor coupled to the memory; The executable program code executes the method described in the first embodiment.

Embodiment 4

The embodiment of the present invention also discloses a computer storage medium, where a computer program is stored on the storage medium, and the computer program is executed by the processor to execute the method described in the first embodiment.

The computer storage medium in the embodiments of the present invention may adopt any combination of one or more computer-readable mediums. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or any combination of the above. More specific examples (a non-exhaustive list) of computer readable storage media include: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), Erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disk read only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

A computer-readable signal medium may include a propagated data signal in baseband or as part of a carrier wave, with computer-readable program code embodied thereon. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device .

Program code embodied on a computer readable medium may be transmitted using any suitable medium including, but not limited to, wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

Computer program code for carrying out operations of the present invention may be written in one or more programming languages, including object-oriented programming languages—such as Java, Smalltalk, C++, but also conventional Procedural programming language - such as the "C" language or similar programming language. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a local area network (LAN) or wide area network (WAN), or may be connected to an external computer (eg, through the Internet using an Internet service provider) connect).

Note that the above are only the preferred embodiments of the present invention and the applied technical principles. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the protection scope of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and can also include more other equivalent embodiments without departing from the concept of the present invention, all of which belong to protection scope of the present invention.

Claims (10)

1. A spine health change prediction method based on big data is characterized by comprising the following steps:

acquiring spine image big data of different health stages, establishing a training set based on the spine image big data, and training a depth recognition model by using the training set;

acquiring spine image data of a current health stage, inputting the spine image data into the trained depth recognition model, and outputting a spine health prediction atlas by the depth recognition model;

and outputting the spine health prediction atlas according to a preset strategy.

2. The big-data-based spine health change prediction method according to claim 1, wherein: the spine image big data comprises first spine images of different health stages;

then the building of a training set based on the spine image big data comprises:

performing first spine parameter extraction processing on each first spine image, determining a first attribute of each first spine image based on the extracted first spine parameters, and labeling each first spine image by using the first attribute;

fitting each first spine image with the same first attribute to obtain a second spine image and a corresponding second spine parameter, and determining a second attribute of the second spine image according to the second spine parameter;

calculating a first matching degree of the second attribute of each second spine image and a standard attribute stage table, determining a first stage sequence number of each second spine image according to the first matching degree, and associating a plurality of first spine images corresponding to each second spine image with the first stage sequence number;

traversing backwards by taking the first-stage sequence number of each second spine image as a starting point, and associating a plurality of first spine images related to traversal and corresponding to the second spine images with the second-stage sequence numbers;

and obtaining a plurality of training data according to the first spine image and/or the second spine image, and establishing each training data into the training set.

3. The big-data-based spine health change prediction method according to claim 2, wherein: the first spine image is associated with a user attribute and is associated with a third-stage sequence number;

said associating a number of said first spine images corresponding to each said second spine image with said first phase sequence number comprises:

determining at least two first spine images based on the third-stage sequence number, and calculating a degree of mutation according to the first spine parameters of the at least two first spine images;

and if the mutation degree is larger than or equal to a first threshold value, deleting the corresponding first spine image, and associating the remaining first spine images with the first stage sequence number.

4. The big-data-based spine health change prediction method according to claim 3, wherein: after the degree of mutation is greater than or equal to the first threshold, the method further comprises:

calculating whether the degree of mutation is smaller than a second threshold value, if so:

determining a number of first mechanical models based on the first spine parameters of the first spine image previously adjacent to the first spine image and calculating a correlation value of the first spine parameters of the first spine image and the corresponding first mechanical models;

if the relevance value is larger than or equal to a third threshold value, judging that the degree of mutation is larger than or equal to a first threshold value, and deleting the corresponding first spine image; otherwise, the mutation degree is judged to be smaller than a first threshold value, and the corresponding first spine image is not deleted.

5. The big-data-based spine health change prediction method according to claim 4, wherein: the third threshold is determined by:

determining a number of second dynamical models based on the first spine parameter of a second anterior adjacent of the first spine image, calculating a first similarity of the second dynamical models to the corresponding first dynamical models, determining the third threshold based on the first similarity;

wherein the third threshold is inversely related to the first similarity.

6. A big data based spinal health change prediction method as claimed in any of claims 1-5, wherein: the inputting the spine image data into the trained depth recognition model, the depth recognition model outputting a spine health prediction atlas, comprising:

calculating a third spine parameter corresponding to the spine image data, and inputting the third spine parameter into the trained depth recognition model, wherein the depth recognition model obtains an initial prediction atlas, and the initial prediction atlas comprises a plurality of second spine images and corresponding fourth-stage sequence numbers;

calculating a second similarity of the third spine parameter to the first spine parameter corresponding to each of the first spine images associated with the second spine image;

if the second similarity is larger than or equal to a fourth threshold, taking the first spine image associated with the second spine images as a target prediction image, and correspondingly associating the target prediction image corresponding to each second spine image with the fourth stage sequence number to obtain the spine health prediction image set;

the depth recognition model outputs the spine health prediction atlas.

7. The big-data-based spine health change prediction method according to claim 6, wherein: the outputting the spine health prediction atlas according to a preset strategy comprises:

calculating a second matching degree of a third attribute corresponding to the spine image data of the current health stage and a standard attribute stage table, determining a fifth stage sequence number based on the second matching degree, and determining the number of images in the spine health prediction image set according to the fifth stage sequence number;

determining a sequence number of a key stage according to the third attribute, and correcting the number of the images of the key stage according to the sequence number of the key stage;

wherein the correction degree is positively correlated with the criticality value of the critical phase.

8. A spine health change prediction system based on big data comprises a first acquisition module, a second acquisition module, a processing module, a storage module and an output module, wherein the processing module is connected with the first acquisition module, the second acquisition module, the storage module and the output module; wherein,

the first acquisition module is used for acquiring the spine image big data in different health stages and transmitting the spine image big data to the processing module;

the second acquisition module is used for acquiring the spine image data of the current health stage and transmitting the spine image data to the processing module;

the storage module is used for storing executable computer program codes;

the output module is used for outputting the spine health prediction atlas according to a preset strategy;

the method is characterized in that: the processing module for executing the method according to any one of claims 1-7 by calling the executable computer program code in the storage module.

9. An electronic device, comprising:

a memory storing executable program code;

a processor coupled with the memory;

the method is characterized in that: the processor calls the executable program code stored in the memory to perform the method of any of claims 1-7.

10. A computer storage medium having a computer program stored thereon, characterized in that: the computer program, when executed by a processor, performs the method of any one of claims 1-7.

Images