Scoliosis screening method, device, storage medium and product based on smartphone

Abstract

The present invention discloses a scoliosis screening method, equipment, storage medium and product based on a smartphone, which belongs to the field of scoliosis screening. Among them, a scoliosis screening device based on a smartphone includes a mobile phone holder module, a handle body module and a roller module. A smartphone with an inertial measurement unit function is placed in the mobile phone holder module, which has the advantages of being handheld and portable. A scoliosis screening method based on a smartphone collects acceleration and angular velocity sequence data of the human spine trajectory, and performs pre-processing such as outlier detection, smoothing and normalization on it, and then uses an outlier detection algorithm based on an isolation forest based on weighted features to identify abnormal samples, and finally achieves the purpose of scoliosis screening. The present invention can detect scoliosis problems quickly and without radiation, reduce the workload of operators, improve measurement accuracy, and is suitable for large-scale census promotion.

Description

Technical Field

[0001] The present invention relates to the field of scoliosis screening, and in particular to a scoliosis screening method, device, storage medium and product based on a smartphone. Background Art

[0002] In recent years, scoliosis has gradually attracted attention. Currently, commonly used scoliosis detection methods include:

[0003] Direct diagnosis: An experienced physician assesses the shape of the patient's back. The patient is asked to stand with their back exposed, arms hanging naturally. The physician assesses whether the shoulders are level and whether one shoulder blade is prominent. The physician also pinches the lower spine protuberance between the middle and index fingers to observe whether it is straight. Abnormality in any of these parameters indicates that the skeletal angle has deviated from the normal range.

[0004] X-ray measurement: The doctor manually measures the Cobb angle to assess the severity of scoliosis based on the shape of the spine shown on an X-ray taken of the patient.

[0005] Scoliosis measurement with a scoliosis ruler: The doctor places a scoliosis ruler on the patient's back, parallel to the spine. Typically, one end of the ruler is placed over the cervical spine, and the other over the pelvic area. The doctor observes the indicator on the ruler and notes its position. The position of the indicator indicates the degree of scoliosis, with larger numbers indicating more severe scoliosis.

[0006] While traditional physical testing methods are simple and easy to use, they rely heavily on the doctor's experience and can lead to misdiagnosis. While using X-ray images for diagnosis improves accuracy, they can expose the patient to radiation damage and are expensive, making them unsuitable for large-scale screening. Summary of the Invention

[0007] The purpose of the present invention is to solve the problems of high professionalism, poor versatility and high cost in the initial screening process of scoliosis, and to propose a scoliosis screening method and device based on a smartphone.

[0008] The purpose of the present invention can be achieved through the following technical solutions:

[0009] A smartphone-based scoliosis screening device includes a smartphone holder module, a handle module, and a roller module. A smartphone equipped with an inertial measurement unit (IMU) is placed in the holder module to obtain the smartphone's acceleration and angular velocity sequence along the subject's spine. The device is handheld and portable.

[0010] A scoliosis screening method based on a smartphone comprises the following steps:

[0011] Step S1: Data Collection. The subject stands naturally, and the roller module of the scoliosis screening device is placed against the subject's back. The roller slides along the spine from the subject's cervical spine (the starting point of the measurement) to the subject's coccyx (the end point of the measurement) at a constant speed. The trajectory data of the route is collected, including the acceleration and angular velocity sequences along the X-axis, Y-axis, and Z-axis directions. , Indicates the of whom ,in , , , , , Representing the The acceleration and angular velocity of the smartphone along the X-axis, Y-axis, and Z-axis in this state. The total number of times the tester slides from the back of the neck to the back of the pelvis, Represents the spine trajectory data of m subjects.

[0012] Step S2: Data preprocessing. For the initial spine sample sequence , perform outlier detection, data smoothing, and data normalization in turn to obtain the standard input data set ,in After preprocessing, The human spine trajectory data of the test subjects, After preprocessing Sampling results, L is the length of the unified sequence;

[0013] Step S3: Construct an isolation forest scoliosis abnormality detection model based on weighted features.

[0014] Step S4: Data Analysis: The input data is tested using a scoliosis abnormality detection model based on weighted feature isolation forest. If it is determined to be an abnormal sample, it means that the subject has a risk of scoliosis; otherwise, it is a normal sample.

[0015] As a further solution of the present invention: in step S1, the mobile phone IMU module is called during the data acquisition process to record and save the acceleration and angular velocity sequence of the smartphone from the measurement start point to the measurement end point.

[0016] As a further solution of the present invention: in step S2, the original acceleration and angular velocity sequence There are noise and anomalies in the , and the outliers are detected by the mean square error method. The calculation steps are as follows:

[0017] First, the sequence Calculate mean square error , and calculate each sequence sample in turn and If the distance exceeds , then the sample is considered an outlier, and let If the distance does not exceed , then the sample is considered a normal point and no processing is done;

[0018] Then the sliding average method is used to smooth the sequence data. The specific smoothing formula is: ,in is the smoothed The sampling results, is the sampling window radius, For The original value of .

[0019] In addition, due to the sequence data generated by different subjects The lengths of the sequences are different. The present invention uses the average sampling method to unify the data dimensions and unify the lengths of different sequences to L.

[0020] As a further solution of the present invention: in step S3, the scoliosis abnormality detection model based on the weighted feature isolation forest is specifically:

[0021] Step S31: In the data set Random selection The sample points are put into the root node of the isolation tree;

[0022] Step S32: Get all features Q and set weights , weight is the probability of each feature being selected, in this invention It obeys the normal distribution and can be expressed as , and have , randomly assign a dimension feature , randomly generate a cutting point in the current node data , The value range of the current node feature Between the maximum and minimum values ​​in ;

[0023] Step S33: Generate a hyperplane with the cutting point, divide the current node data space into two subspaces, and Upper less than The data is placed in the left child node of the current node, which is greater than or equal to The data is placed in the right child node of the current node;

[0024] Step S34: Recursively call steps S32 and S33 in the child node, and continuously construct new child nodes until there is only one data in the child node or the limited height is reached. ;

[0025] Step S35: Loop through steps S31 to S34 until a Isolated trees, training is complete.

[0026] As a further solution of the present invention: in step S4, for the newly added test samples, the formula for calculating the outlier score of each newly added test data is: , the larger the score, the more abnormal the sample. When traversing isolated trees for newly added test data, the average path length in each tree is obtained. ,in .

[0027] Beneficial effects of the present invention:

[0028] 1. The present invention provides a smartphone-based scoliosis screening method and device that can quickly detect scoliosis without radiation, reducing operator workload, improving measurement accuracy, and is suitable for large-scale surveys and promotion.

[0029] 2. The data preprocessing method provided by the present invention can significantly improve the accuracy and stability of the model. By handling anomalies, it can avoid model deviation and accelerate the model training process.

[0030] 3. The weighted feature-based isolation forest scoliosis detection method provided by the present invention enhances the model's perception of important features by assigning higher weights to important spinal features, improves the detection rate, reduces the interference of irrelevant features on model decision-making, and improves the robustness of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The present invention will be further described below with reference to the accompanying drawings.

[0032] Figure 1 Schematic diagram of the structure of the smartphone-based scoliosis screening device of the present invention (device smartphone);

[0033] Figure 2 This is a schematic diagram of the structure of the smartphone-based scoliosis screening device of the present invention (without the smartphone);

[0034] Figure 3 is a front view of the smartphone-based scoliosis screening device of the present invention (without the smartphone);

[0035] Figure 4 is a top view of the smartphone-based scoliosis screening device of the present invention (without the smartphone installed);

[0036] Figure 5 is a side view of the smartphone-based scoliosis screening device of the present invention (without the smartphone);

[0037] Figure 6 This is a detection flow chart of the scoliosis screening device based on a smartphone of the present invention.

[0038] Explanation of the accompanying drawings: 1. mobile phone holder module; 2. handle main body module; 3. roller module; 4. telescopic splint. DETAILED DESCRIPTION

[0039] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0040] Example 1

[0041] See also Figures 1 to 5 As shown, the present invention is a handheld portable scoliosis detection device, comprising a mobile phone holder module 1, a handle body module 2 and a roller module 3. The roller module 3 is arranged below the mobile phone holder module 1. The roller module 3 includes four rollers to fit the human spine. The handle body module 2 is placed behind the roller in the forward direction.

[0042] The handle body module 2 is convenient for the user to hold;

[0043] The mobile phone holder module 1 is used to hold a smart phone and includes a telescopic clamping plate 4 with a telescopic function to accommodate mobile phones of different sizes.

[0044] The roller module 3 is used to fit the surface of the human body to facilitate the acquisition of acceleration and angular velocity information of the smartphone.

[0045] Example 2

[0046] Figure 6 This is a flow chart of the scoliosis screening device based on a smartphone of the present invention, which mainly includes the following steps:

[0047] Step S1: Device installation: Place the smartphone with the IMU module on the smartphone holder module 1 of the scoliosis screening device, and use the telescopic splint 4 to clamp the smartphone.

[0048] Step S2: Data Collection. The subject stands naturally, and the roller module 3 is placed against the subject's back. The roller module slides uniformly along the spine from the subject's cervical spine (the measurement starting point) to the subject's coccyx (the measurement ending point). The mobile phone IMU module is called to collect and record the trajectory data (acceleration and angular velocity sequence) of the smartphone's route from the measurement starting point to the measurement ending point, including acceleration and angular velocity sequences along the X-axis, Y-axis, and Z-axis directions. ,in ,in , , , , , Representing the The acceleration and angular velocity of the smartphone along the X-axis, Y-axis, and Z-axis in this state. The total number of states collected is from the back of the neck to the back of the pelvis of the subject. The total number of states collected varies depending on the sample. In this embodiment, .

[0049] Step S3: Data preprocessing. For the initial spine sample sequence , which contains noise and anomalies. The outliers are detected by the mean square error method. The calculation steps are as follows:

[0050] First, the sequence Calculate mean square error , and calculate each sequence sample in turn and If the distance exceeds , then the sample is considered an outlier, and let To correct the outlier, if the distance does not exceed , then the sample is considered a normal point and no processing is done;

[0051] Then the sliding average method is used to smooth the sequence data. The specific smoothing formula is: ,in is the smoothed The sampling results, is the sampling window radius, For The original value of .

[0052] Sequence data generated by different subjects The lengths of the sequences are different. The present invention uses the average sampling method to unify the data dimensions and unify the lengths of different sequences to L. In this method, L=100.

[0053] Perform outlier detection, data smoothing, and data normalization in turn. Obtain standard input data .

[0054] Step S4: Data Analysis: A scoliosis abnormality detection model based on weighted feature isolation forest is trained and used to detect the input data. If it is determined to be an abnormal sample, it means that the subject is at risk of scoliosis; otherwise, it is a normal sample.

[0055] As a further solution of the present invention: In step S4, the scoliosis abnormality detection model based on the weighted feature isolation forest is specifically as follows: according to step S1 and step S2, data of 400 subjects without scoliosis and 100 subjects with scoliosis are collected in advance to construct a data set. , among which The trajectory data of a subject can be expressed as , The total number of times the tester slides from the back of the neck to the back of the pelvis, ,in , , , , , Representing the The acceleration and angular velocity of the smartphone along the X-axis, Y-axis, and Z-axis in the state. And according to step S3, the data set is processed for outlier detection, data smoothing, and data normalization to obtain a new data set , based on the new dataset To build an anomaly detection model based on weighted feature isolation forest, the steps are as follows:

[0056] a. In the dataset Random selection The sample points are put into the root node of the isolation tree. In this embodiment, ;

[0057] b. Obtain all features Q. According to the characteristics of the human spine, the head and tail data generally cannot show scoliosis. Therefore, it is assumed that the scoliosis data follows a normal distribution and the weights are set. , weight The probability of each feature being selected can be expressed as , and have , the spine features are generally manifested in the middle of the spine, and the features at both ends are relatively common, so the use of the isolation forest method based on weighted features can further give the middle features a higher weight, reduce the interference of weak features at both ends on the model, improve the model detection ability, and enhance the model robustness. Randomly assign a dimension feature , randomly generate a cutting point in the current node data , The value range of the current node feature Between the maximum and minimum values ​​in ;

[0058] c. Generate a hyperplane with the cutting point, and divide the current node data space into two subspaces. Upper less than The data is placed in the left child node of the current node, which is greater than or equal to The data is placed in the right child node of the current node;

[0059] d. Recursively call steps b and c in the child node, and continuously construct new child nodes until there is only one data in the child node or the limit height is reached ;

[0060] e. Loop from a to d until it generates Isolated trees, training is complete.

[0061] As a further solution of the present invention: in step S4, for the newly added test samples, the formula for calculating the outlier score of each newly added test data is: , the larger the score, the more abnormal the sample. When traversing isolated trees for newly added test data, the average path length in each tree is obtained. ,in . .

[0062] Example 3

[0063] An embodiment of the present invention further provides a terminal device, which includes: a memory, a processor, and a computer program / instruction stored in the memory, and the processor executes the computer program / instruction to implement the above-mentioned smartphone-based scoliosis screening method.

[0064] Although not shown, the terminal device includes a processor that can perform various appropriate operations and processes based on programs and / or data stored in a read-only memory (ROM) or loaded from a storage unit into a random access memory (RAM). The processor can be a multi-core processor or can include multiple processors. In some embodiments, the processor can include a general-purpose main processor and one or more specialized coprocessors, such as a central processing unit, a graphics processing unit (GPU), a neural network processor (NPU), a digital signal processor (DSP), and so on. The RAM also stores various programs and data required for the operation of the terminal device. The processor, ROM, and RAM are connected to each other via a bus. An input / output (I / O) interface is also connected to the bus.

[0065] The processor and memory are used together to execute the program / instructions stored in the memory. When the program / instructions are executed by the computer, the methods, steps or functions described in the above embodiments can be implemented.

[0066] Although not shown, an embodiment of the present invention further provides a computer-readable storage medium having a computer program / instruction stored thereon. When the computer program / instruction is executed by a processor, the scoliosis screening method described above is implemented.

[0067] Storage media in embodiments of the present invention include permanent and non-permanent, removable and non-removable items that can be used to store information using any method or technology. Examples of storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0068] Computer-readable storage media include both permanent and non-permanent, removable and non-removable media that can be implemented using any method or technology to store information. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media such as modulated data signals and carrier waves.

[0069] Although not shown, an embodiment of the present invention further provides a computer program product, comprising: a computer program / instruction, which implements the scoliosis screening method described above when executed by a processor.

[0070] The above disclosure is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field can easily think of changes or modifications within the technical scope disclosed in the present invention, and they should all be covered by the scope of protection of the present invention.

Claims

1. A scoliosis screening method based on a smartphone, applied to a scoliosis screening device based on a smartphone, characterized in that: The screening device comprises: a mobile phone holder module (1), a handle body module (2) and a roller module (3); The mobile phone holder module (1) is used to place a smart phone with an IMU function, and includes a clamping plate with a telescopic function to accommodate mobile phones of different sizes; The handle body module (2) is placed at the rear side of the roller in the forward direction, so as to facilitate the user to hold the measurement operation; The roller module (3) is located below the mobile phone holder module (1) and includes four rollers to fit the human spine, so that the smart phone with the IMU function can obtain acceleration and angular velocity information; The screening method comprises: Step S1: Using the IMU state data D = [Y1, Y2, ..., Y i ,...,Y m ], m represents the number of subjects to be tested, Y i =[X1,X2,...,X j ,...,X n ] represents the human spine trajectory data of the i-th subject, n is the data length, X j =[ax j ,ay j ,az j ,ωx j ,ωy j ,ωz j ], where ax j ,ay j , az j ,ωx j ,ωy j ,ωz j Respectively represent the acceleration and angular velocity along the X-axis, Y-axis, and Z-axis in the jth state; Step S2: Preprocess the human spine trajectory data in step S1 to obtain a standard input data set D * =[Y1',Y2',...,Y i ',...,Y m '], where Y i '=[X1',X2',...,X j ',...,X L '] represents the human spine trajectory data of the i-th subject after preprocessing, X j ' is the jth sampling result after preprocessing, and L is the length of different sequences after unification; Step S3: construct an isolated forest scoliosis abnormality detection model based on weighted features, assign different weights according to the characteristics of the spine data, and identify more accurate spine morphology; Step S4: Train the weighted feature-based isolated forest scoliosis anomaly detection model in step S3, and judge human scoliosis by calculating the outlier score of each newly added data to be tested. If the outlier score is greater than the preset condition, it is judged as an abnormal sample, which means that the person to be tested has a risk of scoliosis. Otherwise, the person to be tested has no risk of scoliosis.

2. The scoliosis screening method according to claim 1, characterized in that: The step S2 pre-processes the human spine trajectory data, including the following steps: Step S21: Calculate the mean square error σ of the sequence Y, and calculate the mean square error σ of each sequence sample X in turn. j If the distance exceeds 3σ, the sample is considered an outlier and X j =(X j-1 +X j+1 ) / 2, if the distance does not exceed 3σ, the sample is considered a normal point and no processing is performed; Step S22: Use the sliding average method to smooth the sequence data. The specific smoothing formula is: Where X i ' is the i-th sampling result after smoothing, m is the sampling window radius, X j is the original value at j; Step S23: The lengths of the sequence data Y generated by different subjects are different. The present invention uses an average sampling method to unify the data dimension and unify the different sequence lengths to L length.

3. The scoliosis screening method according to claim 2, characterized in that: In step S3, constructing an isolated forest scoliosis abnormality detection model based on weighted features includes the following steps: Step S31: In the data set D * Random selection The sample points are put into the root node of the isolation tree; Step S32: Obtain all feature sequences and set weights w according to the characteristics of the human spine i , weight w i is the probability of each feature being selected, w i It obeys normal distribution and is expressed as And Where N is the number of features, a dimension feature q is randomly specified, and a cutting point p is randomly generated in the current node data. The value range of p is between the maximum and minimum values ​​of the current node feature q; Step S33: Generate a hyperplane with the cutting point, divide the current node data space into two subspaces, place the data less than p on q in the left child node of the current node, and place the data greater than or equal to p in the right child node of the current node; Step S34: recursively call steps S32 and S33 in the child node, and continuously construct new child nodes until there is only one data in the child node or the specified height has been reached. Step S35: loop step S31 to step S34 until t isolated trees are generated and the training is finished.

4. A scoliosis screening method according to claim 3, characterized in that: The formula for calculating the outlier score of each newly added data to be tested in step S4 is: Among them, E(h(x)) is the average path length of each tree when the newly added test data traverses the isolated tree. Among them, H(i)=ln(i)+0.57721.

5. A terminal device comprising a memory, a processor and a computer program / instruction stored in the memory, characterized in that: The processor executes the computer program / instructions to implement the scoliosis screening method according to any one of claims 1 to 4.

6. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the scoliosis screening method according to any one of claims 1 to 4 is implemented.

7. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the scoliosis screening method according to any one of claims 1 to 4 is implemented.

Images