KR102639558B1 - Growth analysis prediction apparatus using bone maturity distribution by interest area and method thereof - Google Patents

Abstract

It is a growth analysis prediction device, comprising a memory including instructions, and a processor that executes the instructions to analyze growth characteristics in the input carpal image, where the processor detects main areas where the growth plate is located in the carpal image, and detects the main areas where the growth plate is located in the carpal image. Bone maturity is analyzed for each area, a bone maturity distribution map for major areas is created, and based on the bone maturity distribution map, one or more of the following are analyzed: growth delay, precocious puberty, bone growth rate, and growth prediction.

Description

Growth analysis prediction device and method using bone maturation distribution by area of interest {GROWTH ANALYSIS PREDICTION APPARATUS USING BONE MATURITY DISTRIBUTION BY INTEREST AREA AND METHOD THEREOF}

A growth analysis prediction device and method using bone maturation distribution for each region of interest is provided.

Among the various methods developed to measure bone age, the most commonly used methods are the Greulich-Plye (GP) method and the Tanner-Whitehouse 3 (TW3) method. The GP method measures bone age by taking a picture of the left hand and wrist and comparing it with the GP atlas standard image of the Bone Growth Atlas. It is simple and can be used quickly in clinical practice. However, the GP method is a simple comparison of multiple images, and the measurement is subjective and greatly influenced by the reader's experience. In addition, because most of the bone age intervals in the guidebook are divided into 1-year increments, there are limitations in measuring bone age semiquantitatively and precisely.

Meanwhile, the TW3 method is a method that evaluates bone age by measuring the bone maturity grade for each major area within the carpal bone that shows significant changes depending on bone age, scoring it, and then summing it all up. Since it is a method that reflects and quantifies the degree of bone maturity in each major area where changes in bone maturity are evident, it is evaluated to be more objective and accurate with less deviation than the GP method. When applying deep learning techniques, the accuracy of classification can be improved by focusing on and analyzing areas of high importance for skeletal maturity classification, such as TW3.

However, the TW3 method has the disadvantage of taking a long time because it requires analysis for each area. In addition, in classifying each major area into only 9 grades from A to I, many cases where the grade classification is ambiguous may occur, and even if classified into the same grade, there are cases where there is a large difference in maturity level. In this way, there may be deviations depending on the evaluator's clinical experience in determining the maturity level for each region.

The problem to be solved by the present invention is to provide a growth analysis prediction device and method using bone maturation distribution for each region of interest that analyzes and predicts growth based on the distribution of bone maturity degrees analyzed for each major region of interest precisely based on artificial intelligence. It is for.

In addition to the above tasks, it can be used to achieve other tasks not specifically mentioned.

A growth analysis and prediction device according to an embodiment of the present invention includes a memory including instructions, and a processor that executes the instructions to analyze growth characteristics in the input carpal image, and the processor is configured to determine the main location of the growth plate in the carpal image. Detect regions, analyze bone maturity in each detected major region, generate a bone maturity distribution map for each major region, and based on the bone maturity distribution map, determine whether there is growth delay, whether there is precocious puberty, bone growth rate, and growth prediction values. Analyze growth characteristic values including one or more.

The processor selects two or more of the following major regions: the radius, ulna, distal phalange, middle phalange, proximal phalange, metacarpal, and carpal. Areas can be set and main areas detected in the carpal image through a learned main area detection model.

The processor can derive bone maturity through analysis of the growth plate for each major region through region analysis models that have been trained to derive bone maturity for each major region.

The processor creates a bone maturity distribution value by connecting the bone maturity levels of major areas, and inputs one or more data from the subject's height, age, and weight for the bone maturity distribution value and carpal image into a fully trained integrated analysis model for growth. Characteristic values can be derived.

The processor collects multiple carpal images, labels key regions highly related to the growth plate in each carpal image, constructs learning data by labeling bone maturity for key regions, and uses the learning data to label key regions highly related to the growth plate. Train a main region detection model to derive regions, train a region analysis model to derive bone maturity labeled in key regions, and calculate bone maturity distribution values for key regions and actual height, weight, and age for carpal images. An integrated analysis model can be trained to derive growth characteristic values based on basic information about the growth characteristics.

A method of operating a growth analysis prediction device according to an embodiment of the present invention, collecting multiple carpal images, labeling main regions for each carpal image, labeling bone maturity for each key region, and labeling bone maturity for each key region. A step of constructing learning data by labeling growth characteristic values for each maturity distribution value. Using the learning data, a main region detection model is trained to derive key regions from the carpal image, and the region is analyzed to derive bone maturity labeled in the main region. Learning the model and learning an integrated analysis model to derive labeled growth characteristic values from the bone maturity distribution values, and receiving the carpal image of the subject to be predicted, a main region detection model, a region analysis model, and an integrated analysis model It includes the step of providing growth characteristic values for the subject derived by sequentially applying .

In the step of building learning data, image quality improvement or equalization work can be performed on the carpal image based on the input format of the main region detection model, region analysis model, and integrated analysis model.

The learning step is to derive one or more growth characteristic values among growth retardation, precocious puberty, and bone growth rate based on basic information about actual height, weight, and age for bone maturation distribution values for key regions and carpal images. An integrated analysis model can be trained to do this.

The step provided may be provided by analyzing growth prediction values, including growth potential or predicted height when growth is completed, based on one or more data among growth delay, precocious puberty, and bone growth rate.

The provided stage can be provided by converting one or more data from the following data, bone maturation distribution by major area, growth delay, precocious puberty, bone growth rate, bone age chronological change, and predicted height, into a table, graph, or graphic format. .

According to one embodiment of the present invention, the level of bone maturity in each major region can be analyzed very precisely in the major region detected from the carpal image through an artificial intelligence model.

According to one embodiment of the present invention, it is possible to accurately predict the growth potential of a subject by analyzing the distribution of bone maturation and clearly provide data on the basis for the prediction.

1 is a structural diagram showing a growth analysis prediction device according to an embodiment of the present invention.
Figure 2 is an exemplary diagram showing an AI model according to an embodiment of the present invention.
Figure 3 is a flowchart showing a method for precise analysis of arthritis severity according to an embodiment of the present invention.
Figure 4 is an exemplary diagram showing main areas of interest according to an embodiment of the present invention.
Figure 5 is an exemplary diagram showing various bone age distributions for each region of interest according to an embodiment of the present invention.
Figure 6 is a hardware configuration diagram of a computing device according to an embodiment of the present invention.

With reference to the attached drawings, embodiments of the present invention will be described in detail so that those skilled in the art can easily implement the present invention. The present invention may be implemented in many different forms and is not limited to the embodiments described herein. In order to clearly explain the present invention in the drawings, parts not related to the description are omitted, and the same reference numerals are used for identical or similar components throughout the specification. Additionally, in the case of well-known and well-known technologies, detailed descriptions thereof are omitted.

Throughout the specification, when a part is said to “include” a certain element, this means that it may further include other elements rather than excluding other elements, unless specifically stated to the contrary.

The devices described in the present invention are composed of hardware including at least one processor, a memory device, a communication device, etc., and a program that is executed in conjunction with the hardware is stored in a designated location. The hardware has a configuration and performance capable of executing the method of the present invention. The program includes instructions that implement the operating method of the present invention described with reference to the drawings, and executes the present invention by combining it with hardware such as a processor and memory device.

In this specification, “transmission or provision” may include not only direct transmission or provision, but also indirect transmission or provision through another device or using a circuitous route.

In this specification, expressions described as singular may be interpreted as singular or plural, unless explicit expressions such as “one” or “single” are used.

In this specification, the same reference numbers refer to the same elements regardless of the drawings, and “and/or” includes each and all combinations of one or more of the mentioned elements.

In this specification, terms including ordinal numbers, such as first, second, etc., may be used to describe various components, but the components are not limited by the terms. The above terms are used only for the purpose of distinguishing one component from another. For example, a first component may be referred to as a second component, and similarly, the second component may be referred to as a first component without departing from the scope of the present disclosure.

In the flowcharts described herein with reference to the drawings, the order of operations may be changed, several operations may be merged, certain operations may be divided, and certain operations may not be performed.

1 is a structural diagram showing a growth analysis prediction device according to an embodiment of the present invention.

As shown in FIG. 1, the growth analysis prediction device 100 includes a preprocessor 110, a learning unit 120, a region detection unit 130, a bone maturity analysis unit 140, a growth characteristics analysis unit 150, and Includes an output unit 160.

For explanation purposes, they are referred to as a pre-processing unit 110, a learning unit 120, an area detection unit 130, a bone maturity analysis unit 140, a growth characteristics analysis unit 150, and an output unit 160, but these are at least one It is a computing device that runs on a processor. Here, the preprocessing unit 110, learning unit 120, area detection unit 130, bone maturity analysis unit 140, growth characteristics analysis unit 150, and output unit 160 are implemented in one computing device or separately. It can be distributed and implemented on computing devices.

For example, the pre-processing unit 110 and the learning unit 120 are implemented in separate computing devices, and the trained AI model is divided into a region detection unit 130, a bone maturity analysis unit 140, and a growth characteristics analysis unit 150. It can be implemented in conjunction with .

In the case of distributed implementation in a separate computing device like this, the preprocessing unit 110, the learning unit 120, the area detection unit 130, the bone maturity analysis unit 140, the growth characteristics analysis unit 150, and the output unit 160. can communicate with each other through a communication interface. The computing device is sufficient as a device capable of executing a software program written to carry out the present invention, and may be, for example, a server, a laptop computer, etc.

The pre-processing unit 110 collects data from the linked database and constructs learning data for learning an artificial intelligence model (AI) in the learning unit 120.

Here, the database stores medical images captured from medical imaging equipment, medical images read by multiple image interpretation specialists, and reading results for the medical images. At this time, medical images include x-ray X-ray imaging technology (radiography), CT (Computed Tomography), PET (Positron Emission Tomography), and Ultrasound (ultrasound imaging diagnosis) It represents a medical image among MRI (Magnetic Resonance Imaging), etc.

And the preprocessor 110 can construct learning data for learning an artificial intelligence model from among the stored data (medical images, reading results, read medical images, etc.).

For example, the preprocessor 110 may construct learning data by labeling a specific area selected as the detection area in the carpal bone image.

Alternatively, the preprocessor 110 may construct learning data by labeling the medical image for the detection area and the bone maturity level read from the corresponding medical image.

In addition, the preprocessor 110 can build learning data by matching growth delay, precocious puberty, bone growth rate, predicted adult height, etc. according to the distribution of bone maturity.

In this way, the pre-processing unit 110 builds learning data suitable for each artificial intelligence model based on the type, purpose, and function of the artificial intelligence model performing learning in the learning unit 120.

Additionally, the preprocessor 110 may preprocess medical images based on the input format of the artificial intelligence model so that the collected medical images can be input to the artificial intelligence model. Preprocessing of medical images is a step to increase the effectiveness of artificial intelligence learning and analysis, and includes processes such as image quality improvement and uniformity. For example, the accuracy of image analysis can be improved by applying CLAHE (Contrast Limited Adaptive Histogram Equalization) and histogram equalization as image quality improvement methods.

The learning unit 120 uses learning data to train an artificial intelligence model to detect a region of interest or derive analyzed data in response to an input image.

The model used in the learning unit 120 is a deep neural network composed of one or more layers, including multiple convolution layers and multiple convolutions that create a feature map for the features of the region of interest. It may be a convolutional neural network (CNN) that includes a pooling layer that performs subsampling between layers. A convolutional neural network can extract analysis results from an input image by alternately performing convolution and subsampling on the input image.

Specifically, in the case of detection learning, learning data that marks the areas to be automatically detected in the image is constructed, and these learning data are trained on an artificial intelligence detection model developed by optimizing the detection of specific areas. Create a detection model that has completed learning.

Analysis learning involves constructing learning data that applies classification criteria to the images and areas within the images to be analyzed, and training an artificial intelligence analysis model developed to automate classification for specific purposes with these learning data. Create a trained analysis model.

The detection model, one or more individual analysis models, and the integrated analysis model are implemented as artificial intelligence models, which may be independent artificial intelligence models, or may be implemented as one artificial intelligence model linked to each other. Accordingly, one or more artificial intelligence models corresponding to the above-described configurations may be implemented by one or more computing devices.

In this way, the learning unit 120 repeatedly learns each artificial intelligence model based on learning data, and can periodically retrain each artificial intelligence model based on a certain period of time or a specific standard.

Here, the artificial intelligence model represents a main region detection model, a region analysis model, and an integrated analysis model, but is not necessarily limited thereto. And the main region detection model, region analysis model, and integrated analysis model can each be implemented as one or more artificial intelligence models.

The region detection unit 130 inputs the input carpal image into a pre-learned main region detection model and detects regions where growth plates are present in the carpal image.

In detail, the area detection unit 130 can automatically detect areas selected to analyze bone age (bone maturity) among the growth plate areas.

For example, areas such as the radius, ulna, distal phalange, middle phalange, proximal phalange, metacarpal, and carpal can be detected. You can.

The area detection unit 130 sequentially detects a plurality of main areas set in the carpal image, and each main area may overlap with each other depending on the size of the image and the size of the detected main area.

The bone maturity analysis unit 140 may input the detected major regions into region analysis models learned as input data and obtain bone maturity for each region.

The bone maturity analysis unit 140 can obtain the bone maturity in one main area by using a region analysis model learned to analyze the bone maturity in that main area.

Here, bone maturity can be derived by analyzing the growth plate and can be output as a bone age value.

The growth characteristics analysis unit 150 analyzes growth characteristics based on the distribution of bone age in each major region analyzed by the bone maturity analysis unit 140. In addition, the growth characteristics analysis unit 150 may analyze growth characteristics by further using basic information including the current height, actual age, weight, and obesity level of the subject in the carpal image.

Here, growth characteristics include, but are not necessarily limited to, growth retardation, precocious puberty, bone growth rate, bone age, and reverse age gradient.

The growth characteristics analysis unit 150 may analyze growth prediction values, including growth potential or predicted height upon completion of growth, based on the analyzed growth characteristics.

Here, the predicted key represents the height that can be estimated when the subject becomes an adult or when growth is complete, and may be expressed as a range including an error value or may represent the average value in the range.

And the output unit 160 can output the data analyzed by the bone maturity analysis unit 140 and the growth characteristics analysis unit 150 to a linked terminal or display.

At this time, the output unit 160 may display a graph, such as a bone maturation distribution chart for each major region, or provide analysis data or prediction data on growth characteristics by converting the results for each list into tables, graphs, figures, etc.

Figure 2 is an exemplary diagram showing an AI model according to an embodiment of the present invention.

Figure 2 (a) represents an automatic region detection model, (b) represents a region analysis model, and (c) represents an integrated analysis model.

Figure 2(a) shows a main area detection model that automatically detects areas where growth plates exist within the carpal bone using a carpal image as an input value.

In detail, the main region detection model is a model that detects regions of interest that serve as a basis for judging the growth characteristics to be analyzed in carpal images.

For example, the main area detection model can detect individual areas for the plurality of main areas described above.

The main area detection model can be implemented as an independent artificial intelligence model that detects each main area, or it can be implemented as a single artificial intelligence model and implemented as a plurality of artificial intelligence models that detect the area of interest for one medical image.

At this time, the main region detection model may be linked to detect the region of interest corresponding to the region to be analyzed in the region analysis model. Accordingly, when the region analysis model is added or changed, a function to automatically detect the image of the region of interest that becomes the input image of the major region detection model may be included, or an artificial intelligence model that detects the major region may be added. Also, conversely, when the main area detection model is added or changed, the area analysis model may be added or changed.

Figure 2(b) shows a region analysis model that outputs bone maturity by analyzing the growth plate area, size and state of bone, etc. located within the main region using the image of the main region as input.

At this time, the learned domain analysis model for each major area is matched one-to-one, and through this, bone maturity in each area can be inferred.

The region analysis model is a model that applies deep neural network techniques from the Convolutional Neural Network series, and bone maturity can be output as bone age value.

For example, the bone maturity level output from the radius (A) area analysis model that receives the radius (A) area can be expressed as a bone age value, such as 10 years. This bone age value expresses a certain standard in relation to age, and is a standard value set by the appearance, shape, or fusion of bone nuclei.

Figure 2(c) shows an integrated analysis model that analyzes growth characteristics or derives growth prediction using the distribution of bone maturity in major regions as input.

In addition to the distribution of bone maturity in major areas, the integrated analysis model can input basic information of the subject and analyze growth delay, precocious puberty, and bone growth rate.

Here, the bone growth rate can be output by dividing it into certain stages, such as fast or normal slow.

And the integrated analysis model can output a growth prediction indicating growth potential and a predicted value of adult height based on the distribution of the input bone maturity level, the subject's basic information, and the analyzed growth characteristics.

The predicted value of adult height represents the predicted height value at the time when growth is assumed to be completed.

The integrated analysis model can build a separate model for each growth characteristic and can be learned through supervised learning such as regression analysis, SVM, and neural network techniques.

Figure 3 is a flowchart showing a method for precise analysis of arthritis severity according to an embodiment of the present invention.

As shown in FIG. 3, the growth analysis prediction device 100 inputs the carpal image into a learned region detection model to obtain a plurality of main regions (S10).

The growth analysis and prediction device 100 may acquire individual key regions in a carpal image through a region detection model learned for a plurality of key regions that are set to be highly related to the growth plate.

Next, the growth analysis prediction device 100 inputs the learned region analysis model for each of the plurality of major regions to obtain bone maturity in the major regions (S20).

The growth analysis and prediction device 100 can output bone maturity in each major area related to the growth plate as bone age. At this time, accurate bone maturity can be obtained for each region through a region analysis model learned specifically for each major region.

Then, the growth analysis prediction device 100 analyzes growth characteristics through an integrated analysis model learned based on the bone maturity distribution of major areas (S30).

Here, the bone maturity distribution value

[Equation 1]

X = [F1, F2, … , Fh,… , F.H.],

Here, H represents the number of regions of interest (main growth plates), Fh represents the bone maturity value of region of interest h, and h represents the number of major regions.

In this way, the growth analysis prediction device 100 can infer growth characteristics by inputting the bone maturity distribution value into the integrated analysis model. Here, the integrated analysis model can infer individual growth characteristic output values by building a separate model for each growth characteristic.

In addition, the growth analysis prediction device 100 uses the bone maturity distribution and the subject's basic information such as current height, age, weight, and obesity to determine growth, including whether there is growth delay, precocious puberty, bone growth rate, etc. Characteristics can be analyzed.

Next, the growth analysis and prediction device 100 provides data on the analyzed growth characteristics and predicted growth data (S40).

The growth analysis and prediction device 100 can convert and provide analyzed and predicted data into various forms of tables, graphs, figures, etc., based on the subject's basic information.

Figure 4 is an exemplary diagram showing main areas according to an embodiment of the present invention.

As shown in Figure 4, the main region detected through the automatic region detection model can be confirmed in the carpal image.

In general, the growth plate is a cartilage tissue remaining between the two ends of a bone and is the part where growth occurs along the length of the bone, so the area between the bones can be set as the main area.

Accordingly, as described above, the growth analysis prediction device 100 includes the radius, ulna, distal phalange, middle phalange, proximal phalange, metacarpal, and Areas such as the carpal bones can be detected.

Specifically, the radius, which is the part between the wrist joint of the bone located on the outside of the two bones that make up the forearm bones, and the ulna, which is the part between the wrist joints that is located on the inside of the two bones that make up the forearm bones. ulna), and the carpal bone, which represents the eight short bones that make up the wrist.

In addition, the metacarpal (MC1), which represents the metacarpal bone between the wrist bone and the finger bone, the proximal phalange (PP-MC), which is the first joint bone among the finger bones that touches the metacarpal bone, and the middle joint bone among the finger bones. It can be detected for the middle phalange (MP), which is a bone, and the distal phalange (DP), which is the most distal bone of the finger bone.

At this time, each main area can be detected by overlapping certain areas. In Figure 4, MC1, PP-MC, MP, and DP are each shown as one area, but each main area (MC1, PP-MC, MP, DP) can be detected.

Figure 5 is an exemplary diagram showing the distribution of various performance ages for each area of interest according to an embodiment of the present invention.

Figures 5 (a), (b), (c), and (d)) are exemplary diagrams showing bone age distributions derived from major regions based on the carpal regions of each different subject.

As shown in Figure 5, in the carpal bone image, the bone age in each region related to the growth plate appears different based on the growth plate, showing a variety of bone age distributions.

In Figure 5 (a), it can be seen that the bone age (dp, mp) of the growth plate in the finger bone is high, but the bone age (mc, radius, ulna) in the wrist region is relatively low. On the contrary, in figure (c), it can be seen that the bone age (mc, radius, ulna) of the growth plate in the finger bone is high, but the bone age (dp, mp) in the wrist region is relatively low.

In addition, there can be various bone age distributions, such as a bone age distribution such as (d) where the bone age is high only in certain areas, and a bone age distribution such as (b) where the bone age is the same in all major areas.

Accordingly, the growth analysis and prediction device 100 analyzes growth characteristics based on bone age distribution for major areas, enabling precise analysis of the subject's growth characteristics or growth potential.

Figure 6 is a hardware configuration diagram of a computing device according to an embodiment of the present invention.

As shown in FIG. 6, the hardware of the computing device 200 may include at least one processor 210, memory 220, storage 230, and communication interface 240, and may be connected through a bus. there is. In addition, hardware such as input devices and output devices may be included. The computing device 200 may be equipped with various software, including an operating system capable of running programs.

The processor 210 is a device that controls the operation of the computing device 200, and may be various types of processors 210 that process instructions included in a program, for example, a CPU (Central Processing Unit), MPU ( It may be a Micro Processor Unit (Micro Processor Unit), Micro Controller Unit (MCU), Graphic Processing Unit (GPU), etc. The memory 220 loads the corresponding program so that instructions described to execute the operations of the present invention are processed by the processor 210. The memory 220 may be, for example, read only memory (ROM), random access memory (RAM), etc. The storage 230 stores various data, programs, etc. required to execute the operations of the present invention. The communication interface 240 may be a wired/wireless communication module.

In this way, according to the present invention, the level of bone maturity in each major region can be analyzed very precisely in the major region detected from the carpal image through an artificial intelligence model.

In addition, by analyzing the distribution of bone maturation, an accurate prediction of the subject's growth potential can be made and data on the basis for the prediction can be clearly provided.

Although one preferred embodiment of the present invention has been described in detail above, the scope of the present invention is not limited thereto, and various modifications and improvements can be made by those skilled in the art using the basic concept of the present invention defined in the following claims. It falls within the scope of rights of the present invention.

Claims (10)

As a growth analysis prediction device,
memory containing instructions, and
Comprising a processor that executes the commands to analyze growth characteristics in the input carpal image,
The processor is
Detect the main areas where the growth plates are located in the carpal image, obtain the bone maturity level of each main area, generate a bone maturity distribution chart connecting the bone maturity levels of the main areas obtained from the carpal image, and grow from the input bone maturity distribution chart. Obtaining the subject's growth characteristic values for the carpal image inferred from the bone maturation distribution map of the carpal image using a model trained to infer characteristic values,
The growth characteristic value includes one or more of growth delay, precocious puberty, bone growth rate, and growth prediction value.
In paragraph 1:
The processor,
Two or more of the following major regions are the radius, ulna, distal phalange, middle phalange, proximal phalange, metacarpal, and carpal. set,
A growth analysis and prediction device that detects the main regions in the carpal image through a trained main region detection model.
In paragraph 1:
The processor,
A growth analysis prediction device that derives bone maturity through analysis of the growth plate for each major region through region analysis models that have been trained to derive bone maturity for the major regions.
In paragraph 1:
The processor,
If the model is an integrated analysis model trained to infer growth characteristic values by receiving user information including at least one of height, age, or weight along with the input skeletal maturity distribution, the skeletal maturity distribution of the carpal image A growth analysis prediction device for deriving growth characteristic values of the subject by inputting the subject's user information about the carpal image into the integrated analysis model.
In paragraph 1:
The processor,
Collect multiple training carpal images, label major regions highly related to the growth plate for each training carpal image, and label bone maturity for key regions to construct learning data.
Using the learning data, a main region detection model is trained to derive the main regions from the learning carpal image, and a region analysis model is trained to derive bone maturity labeled in the main region,
An integrated analysis model is learned to derive the growth characteristic value based on user information including at least one of the skeletal maturity distribution of major regions obtained from the learning carpal image and the height, weight, or age of the subject of the carpal image. A growth analysis and forecasting device.
As a method of operating a growth analysis prediction device,
By collecting multiple training carpal images, major regions are labeled for each training carpal image, bone maturity is labeled for each major region, and growth characteristic values are labeled for each bone maturity distribution that connects the bone maturity of major regions to create learning data. building steps,
Using the learning data, a main region detection model is trained to derive the main regions from the learning carpal image, a region analysis model is trained to derive bone maturity labeled in the main region, and a region analysis model is trained to derive bone maturity labeled in the key region. A step of learning an integrated analysis model to derive growth characteristic values, and
When the target carpal image of the subject to be predicted is input, the main region detection model and the region analysis model are used to generate a bone maturity distribution chart connecting the bone maturity levels of the main regions obtained from the target carpal image, and the integrated analysis is performed. Obtaining growth characteristic values of the subject inferred from the bone maturation distribution of the target carpal image using a model, and providing growth characteristic values of the subject,
The growth characteristic value includes one or more of growth delay, precocious puberty, bone growth rate, and growth prediction value.
In paragraph 6:
The step of constructing the learning data is,
An operation method for performing an image quality improvement or equalization operation on the carpal bone image based on input formats of the main region detection model, the region analysis model, and the integrated analysis model.
In paragraph 6:
The learning step is,
The integrated analysis model is used to derive the growth characteristic value based on user information including at least one of the skeletal maturity distribution of major regions obtained from the learning carpal image and the height, weight, or age of the subject of the carpal image. Learning operation method.
In paragraph 6:
The steps provided above are:
An operation method for analyzing and providing a growth prediction value including growth potential or predicted height when growth is completed based on one or more data among the growth delay, precocious puberty, and bone growth rate.
In paragraph 7:
The steps provided above are:
An operation method that provides information by converting one or more data from among the bone maturation distribution by major area, growth delay, precocious puberty, bone growth rate, reverse age change in bone age, and predicted height into a table, graph, or graphic format.

Images