Deep learning-based simple urinary flow test results learning and lower urinary tract symptoms diagnosis method

A deep learning-based method for diagnosing lower urinary tract symptoms using simple urinary flow test results addresses the invasiveness of conventional tests by accurately determining urinary system state through non-invasive data extraction and integration, reducing pain and infection risk.

JP7577850B2Active Publication Date: 2024-11-05SAMSUNG LIFE PUBLIC WELFARE FOUND
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
JP2023524820
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2020-10-23
Filing Date
2021-10-15
Publication Date
2024-11-05
Estimated Expiration
2041-10-15

Smart Images

  • Figure 0007577850000001
    Figure 0007577850000001
  • Figure 0007577850000002
    Figure 0007577850000002
  • Figure 0007577850000003
    Figure 0007577850000003
Patent Text Reader

Abstract

The present invention relates to a method for learning the results of a simple uroflow test and a method for diagnosing lower urinary tract symptoms, and more particularly, to a method for learning the results of a simple uroflow test, which is non-invasive data, to train a neural network and diagnose lower urinary tract symptoms using the trained neural network.The present invention provides a method for diagnosing lower urinary tract symptoms, which prevents patients from experiencing pain and embarrassment during the diagnosis of lower urinary tract symptoms and reduces the risk of secondary infection that occurs through invasive diagnostic methods, by generating a learning model using the results of a simple uroflow test, which is a non-invasive testing method, based on deep learning and diagnosing lower urinary tract symptoms using the learning model.
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a learning method of simple uroflow test results and a diagnosing method for lower urinary tract symptoms, and more particularly to a learning method of simple uroflow test results and a diagnosing method for lower urinary tract symptoms, in which a neural network is trained using simple uroflow test results, which are non-invasive data, and the trained neural network is used to diagnose lower urinary tract symptoms. [Background technology]

[0002] Lower urinary tract symptoms (LUTS) are a general term for various symptoms that appear in relation to the process of storing and discharging urine, such as difficulty in starting urination, residual urine, frequent urination, thin urine, straining during urination, bedwetting, urgent urination, and intermittent urination. Recently, the incidence and severity of lower urinary tract symptoms have been increasing due to various reasons, such as psychological stress caused by increased intake of animal fat and increased social complexity, smoking, drinking, weight gain, lack of rest, and lack of exercise. If lower urinary tract symptoms become severe, activity is restricted and patients are constantly in a state of anxiety and tension, causing great mental stress. Going to and from the toilet during sleep not only causes lack of sleep but also adds to physical fatigue, causing various physical problems.

[0003] Lower urinary tract symptoms are diagnosed using a urodynamic study (UDS) to evaluate bladder function. Urodynamic studies are primarily performed to determine prostate surgery and to distinguish between patients with only detrusor underactivity (DUA), for which surgery is unlikely to be effective, and patients with bladder outlet obstruction (BOO), for which surgery is known to be effective.

[0004] However, conventional urodynamic tests involve inserting tubes into the bladder and anus to measure pressure, gradually filling the bladder with saline solution to measure pressure, and then measuring the bladder pressure while the patient urinates. In other words, the urodynamic tests currently used to diagnose lower urinary tract symptoms are not only uncomfortable and embarrassing for patients, but also have problems such as the risk of infection and the pain and embarrassment caused by the long period of time spent with the catheter in place.

[0005] As a result, there is a continuing need for a method for diagnosing lower urinary tract symptoms that can accurately determine the state of the urinary system without causing pain or embarrassment to patients. Summary of the Invention [Problem to be solved by the invention]

[0006] In response to the demand for the development of a method for diagnosing lower urinary tract symptoms as described above, the present invention aims to provide a method for diagnosing lower urinary tract symptoms, which prevents the patient from experiencing pain and embarrassment during the diagnosis of lower urinary tract symptoms and reduces the risk of secondary infection that may occur through invasive diagnostic methods, by generating a learning model using the results of a simple urinary flow test, which is a non-invasive testing method based on deep learning, and diagnosing lower urinary tract symptoms using the learning model. [Means for solving the problem]

[0007] As one embodiment of the present invention, a deep learning-based method for learning simple urinary flow test results for diagnosing lower urinary tract symptoms is provided.

[0008] A method for learning a result of a simple urinary flow test according to an embodiment of the present invention includes a character extraction step of extracting character data from a result sheet obtained through a simple urinary flow test, and a character learning model generation step of generating a character learning model using the character data as learning data in order to extract feature points having a correlation with a cause of a lower urinary tract symptom from the character data.

[0009] In a method for learning the results of a simple urinary flow test according to an embodiment of the present invention, the text data includes at least one of a point of maximum urinary flow rate (Qmax) during the urination process, voiding time, post-void residual (PVR), and bladder filling volume (BFV).

[0010] A method for learning a result of a simple urinary flow test according to an embodiment of the present invention includes a graph extraction step of extracting graph data from a result sheet obtained through a simple urinary flow test, and a graph learning model generation step of generating a graph learning model using the graph data as learning data in order to extract feature points having a correlation with a cause of a lower urinary tract symptom from the graph data. In the method for learning the results of a simple uroflow test according to an embodiment of the present invention, the graph data includes a urination volume over time or a urination rate over time.

[0011] In a learning method for simple urinary flow test results according to one embodiment of the present invention, the graph extraction step further includes a start point extraction step of extracting a point in the graph data where the graph fluctuation begins as a start point where urination begins, an end point extraction step of extracting a point in the graph data where the graph fluctuation ends as an end point where urination ends, and a pre-processing step of extracting a section from the start point to the end point and inputting the section into the graph learning model. As one embodiment of the present invention, a deep learning-based method for diagnosing lower urinary tract symptoms is provided.

[0012] A method for diagnosing lower urinary tract symptoms according to one embodiment of the present invention includes a learning model generation step of generating a character learning model and a graph learning model using the learning method for a simple urinary flow test result according to one embodiment of the present invention, an input step of inputting a simple urinary flow test result paper to be diagnosed to a lower urinary tract symptom diagnosis system, a data extraction step of extracting the character data and the graph data from the result paper, a feature point integration step of extracting a plurality of feature points correlating with the cause of the lower urinary tract symptom from the character data and the graph data and integrating the feature points, and a diagnosis step of analyzing the correlation between the feature points and the lower urinary tract symptoms using the character learning model and the graph learning model to determine whether the result paper corresponds to a lower urinary tract symptom.

[0013] In a method for diagnosing lower urinary tract symptoms according to an embodiment of the present invention, the text data includes at least one of a point of maximum urinary flow rate (Qmax) during urination, a urination time, a post-urination residual volume (PVR), and a pre-urination urine volume (BFV).

[0014] In the method for diagnosing a lower urinary tract symptom according to an embodiment of the present invention, the graph data includes at least one of a urination volume over time and a urination rate over time.

[0015] In the method for diagnosing a lower urinary tract symptom according to one embodiment of the present invention, the data extraction step further includes a start point extraction step of extracting a point in the graph data where a graph fluctuation starts as a start point where urination starts, an end point extraction step of extracting a point in the graph data where a graph fluctuation ends as an end point where urination ends, and a pre-processing step of extracting a section from the start point to the end point and inputting the section into the graph learning model.

[0016] In the method for diagnosing a lower urinary tract symptom according to an embodiment of the present invention, the diagnosing step further includes a step of combining the result of the judgment with a symptom corresponding to the lower urinary tract symptom and expressing the result in a binary or quaternary system. According to one embodiment of the present invention, a computer readable recording medium having a program for implementing the above-described method is provided. Effect of the Invention

[0017] According to one embodiment of the present invention, lower urinary tract symptoms can be diagnosed using only non-invasive test result data while minimizing the physical / mental pain of the subject. In addition, the effort can be reduced by omitting unnecessary tests such as urodynamic tests. [Brief description of the drawings]

[0018] [Figure 1] This is an example of a typical simple uroflow test result sheet used to generate a learning module and diagnose lower urinary tract symptoms. [Diagram 2] 13 is a flowchart of a method for learning the results of a simple urinary flow test using character data. [Diagram 3] 13 is a flowchart of a method for learning the results of a simple urinary flow test using graph data. [Figure 4] 3 is a flow chart of the graph extraction steps according to the present invention; [Diagram 5] In the learning and diagnosis methods according to the present invention, this is a process of extracting graph data. [Figure 6] 1 is a flowchart of a deep learning-based lower urinary tract symptom diagnosis method according to the present invention. [Figure 7] In the diagnosis method according to the present invention, the extracted feature points are integrated. [Figure 8] 1 is a block diagram of a lower urinary tract symptom diagnostic system according to the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0019] Hereinafter, the embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily carry out the present invention. However, the present invention may be embodied in various different forms and is not limited to the embodiments described herein. In order to clearly explain the present invention in the drawings, parts that are not related to the description are omitted, and similar parts are designated by similar reference numerals throughout the specification.

[0020] The terms used in the present invention will be briefly explained, and the present invention will be specifically described. As terms used in the present invention, common terms currently in wide use have been selected as much as possible in consideration of the functions of the present invention, but these may change depending on the intentions of those skilled in the art, precedents, the emergence of new technologies, etc. In addition, in certain cases, there are terms arbitrarily selected by the applicant, and in such cases, the meanings are described in detail in the description of the invention. Therefore, the terms used in the present invention should be defined based on the meanings that the terms have and the overall content of the present invention, rather than simply the names of the terms.

[0021] Throughout the specification, when a part "includes" a certain component, this does not mean excluding other components, but may further include other components, unless otherwise specified. In addition, the terms "unit" and "module" in the specification mean a unit that processes at least one function or operation, which is embodied in hardware or software, or a combination of hardware and software. In addition, throughout the specification, when a part is "connected" to another part, this includes not only the case where the part is "directly connected" but also the case where the part is connected "through other elements therebetween."

[0022] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 shows an example of a typical simple uroflow test result sheet used to generate a learning module and diagnose lower urinary tract symptoms. Referring to FIG. 1, in the method for learning a simple uroflow test result and the method for diagnosing a lower urinary tract symptom according to the present invention, a simple uroflow test result sheet 1 used as stored data includes character data 2, graph data 3, and patient personal information 4. The simple uroflow test result form 1 is a document in a format for reporting result data generated through a simple uroflow test.

[0023] The text data 2 includes at least one of the point of maximum urinary flow rate (Qmax) during the urination process, voiding time, post-void residual (PVR), and bladder filling volume (BFV). The graph data 3 includes the voided volume over time or the voiding rate over time.

[0024] The patient personal information 4 includes at least one of the subject's age, height, weight, urination pattern of the subject obtained through a simple uroflow test, urinary flow test index, prostate symptom score, past medical history, and urination efficiency.

[0025] In the text data 2, the point (Qmax) having the maximum urinary flow rate during the urination process and the urination time are used as elements for quantifying the height and width of the graph in the graph data 3.

[0026] In addition, since pre-void urine volume (BFV) is the sum of voided volume and post-void residual volume (PVR), the degree of bladder fullness before voiding can be used as additional information under the assumption that peak urinary flow rate and urinary flow rate patterns may change depending on bladder fullness.

[0027] In addition, since post-void residual volume (PVR) (or voiding efficiency) is the value obtained by dividing the amount of urine voided by the pre-void urine volume (BFV), the post-void residual volume (PVR) can be used as additional information. This allows the voiding efficiency to be shown as a separate value that is not reflected in the graph.

[0028] The method for learning simple uroflow test results according to the present invention extracts character data and graph data 3 from the simple uroflow test result paper and generates a character learning model and a graph learning model which learn what correlation each test data has with lower urinary tract symptoms, and the method for diagnosing lower urinary tract symptoms uses the learning model generated by the method for learning simple uroflow test results to diagnose whether or not there is a lower urinary tract symptom using only the simple uroflow test result paper.

[0029] FIG. 2 is a flowchart of a method for learning the results of a simple urinary flow test using character data. Referring to FIG. 2, a method for learning a result of a simple urinary flow test using character data 2 according to an embodiment of the present invention includes a character extraction step (S111) of extracting character data 2 from a result sheet obtained through a simple urinary flow test, and a character learning model generation step (S112) of generating a character learning model using the character data 2 as learning data in order to extract feature points having a correlation with the cause of a lower urinary tract symptom from the character data 2.

[0030] The character extraction step (S111) extracts character data 2 from the simple uroflow test result sheet, the character data 2 including the patient's urination information such as the point of maximum urinary flow rate (Qmax) during the urination process, urination time, post-urination residual volume (PVR), and pre-urination urine volume (BFV).

[0031] Methods for extracting character data 2 include finding a portion of the entire simple urinary flow test result sheet that contains the same character string as the information to be extracted and extracting the urination information contained in that portion, finding the position of character data 2 on the simple urinary flow test result sheet and extracting information that falls within a specified range based on that position, or using a function to find character data 2.

[0032] The character learning model generation step (S112) automatically extracts features of multiple pieces of urination information contained in the input character data 2 using deep learning.

[0033] That is, the characteristics of the urination information that are meaningful in diagnosing lower urinary tract symptoms can be known without a human being having to directly compare, analyze, and judge the criteria used to diagnose lower urinary tract symptoms by looking at the urination information.

[0034] FIG. 3 is a flowchart of a method for learning the results of a simple urinary flow test using the graph data 3. Referring to FIG. 3, a method for learning a result of a simple urinary flow test based on deep learning using graph data 3 according to an embodiment of the present invention includes a graph extraction step (S121) of extracting graph data 3 from a result sheet obtained through a simple urinary flow test, and a graph learning model generation step (S122) of generating a graph learning model using the graph data 3 as learning data in order to extract feature points having a correlation with the cause of a lower urinary tract symptom from the graph data 3.

[0035] The graph extraction step (S121) extracts graph data 3 from the simple urinary flow test result sheet. The graph data 3 can be extracted by inputting a position of the graph data 3 on the simple urinary flow test result sheet and extracting information that falls within a predetermined range based on the position, or by using a function to find the graph data 3.

[0036] The graph learning model generation step (S122) uses deep learning to automatically extract features of the input graph data 3. A convolutional neural network (CNN), which is a neural network that borrows the principle by which the visual cortex of the brain processes and recognizes images, may be used to generate the graph learning model.

[0037] FIG. 4 is a flow chart of the graph extraction steps according to the present invention. FIG. 5 shows a process of extracting graph data 3 in the learning method and diagnostic method according to the present invention. Referring to FIG. 4, the graph extraction step (S121) further includes a start point extraction step (S1211) of extracting a point where the graph fluctuation starts in the graph data 3 as a start point where urination begins, an end point extraction step (S1212) of extracting a point where the graph fluctuation ends in the graph data 3 as an end point where urination ends, and a pre-processing step (S1213) of extracting a section from the start point to the end point and inputting it into the graph learning model.

[0038] 5, the graph data 3 extracted from the simple uroflow test result paper records the amount and speed of urine discharged into the simple uroflow test device from the start to the end of the test. Generally, the start of the test does not coincide with the start of urination, and the end of urination does not coincide with the end of the test, so the graph data 3 may contain unnecessary information between the start of the test and the start of urination, and between the end of urination and the end of the test.

[0039] In order to remove unnecessary information, in the present invention, the point where the graph fluctuation starts and the point where the graph fluctuation ends are defined as start point A and end point B, respectively, in the graph data 3, and the section between the start point and the end point is extracted and used to generate a learning model. Therefore, the present invention can obtain more accurate information regarding the patient's urination volume and urination rate.

[0040] FIG. 6 is a flowchart of a deep learning-based method for diagnosing lower urinary tract symptoms according to the present invention. Referring to FIG. 6, the method for diagnosing a lower urinary tract symptom according to an embodiment of the present invention includes a learning model generation step (S10) of generating a character learning model and a graph learning model using the learning method of a simple urinary flow test result according to an embodiment of the present invention, an input step (S20) in which a simple urinary flow test result paper to be diagnosed is input to a lower urinary tract symptom diagnosis system, a data extraction step (S30) of extracting the character data 2 and the graph data 3 from the result paper, a feature point integration step (S40) of extracting a plurality of feature points having a correlation with the cause of the lower urinary tract symptom from the character data 2 and the graph data 3 and integrating the feature points, and a diagnosis step (S50) in which the character learning model and the graph learning model analyze the correlation between the feature points and the lower urinary tract symptoms to determine whether the result paper corresponds to a lower urinary tract symptom.

[0041] The learning model generating step (S10) uses both the character learning model generated according to an embodiment of the present invention and the graph learning model generated according to an embodiment of the present invention.

[0042] In other words, the present invention does not use only character data 2 or only image data, but uses both character and image data, thereby reducing errors that occur when only image data is used and errors that occur when only character data 2 is used.

[0043] The input data input in the input step (S20) is a simple uroflow test result sheet, and includes data measured only by the process of urinating into a test toilet connected to a computer recording device, without the process of inserting a tube into the patient's urethra and anus.

[0044] In other words, there is no risk of pain or urinary tract infection, which are complications associated with tube insertion, and it is possible to accurately diagnose lower urinary tract symptoms through deep learning.In addition, it reduces the effort required to diagnose lower urinary tract symptoms by omitting unnecessary steps such as inserting a tube into the patient's urethra and anus.

[0045] The data extraction step (S30) extracts character data 2 including the patient's urination information and graph data 3 including information regarding the patient's urination volume and urination rate from the simple uroflow test result sheet.

[0046] The feature point integration step (S40) is a step of integrating feature points of the character data and the graph data generated by the character learning model and the graph learning model, which will be described in detail below with reference to FIG.

[0047] The diagnosis step (S50) further includes a step of combining the result of the judgment with symptoms corresponding to lower urinary tract symptoms and expressing the result in a binary or quaternary system. Examples of symptoms corresponding to lower urinary tract symptoms include bladder outlet obstruction (BOO) and detrusor underactivity (DUA).

[0048] That is, the diagnosis step (S50) expresses the result of the diagnosis in a binary system such as normal vs. abnormal / BOO vs. Non-BOO / DUA vs. Non-DUA, or in a quaternary system such as normal vs. BOO vs. DUA vs. BOO&DUA.

[0049] FIG. 7 shows an embodiment of a process of integrating extracted feature points in the method for diagnosing a lower urinary tract symptom according to the present invention. Referring to FIG. 7, in the method for diagnosing a lower urinary tract symptom according to an embodiment of the present invention, the feature point integration step (S40) integrates features extracted from character data 2 by applying a concatenate function to an intermediate layer of a graph learning model using graph data 3 as an input, or by performing element-wise addition.

[0050] In addition, the features extracted from the graph data 3 and the features extracted from the character data 2 are integrated by sequentially applying a function immediately before the output layer of the character learning model or by performing element-by-element calculations.

[0051] FIG. 8 is a block diagram of a lower urinary tract symptom diagnostic system according to the present invention. Referring to FIG. 8, a lower urinary tract symptom diagnosis system according to an embodiment of the present invention includes a character learning model 10 that extracts character data 2 from a simple uroflow test result paper and learns feature points, a graph learning model 20 that extracts graph data 3 from the result paper and learns feature points, a data input unit 40 to which the simple uroflow test result paper is input, a data extraction unit 50 that extracts character data 2 and graph data 3 including the patient's urination information from the simple uroflow test result paper and transmits the data to the learning model or main processor through a learning or diagnosis process, and a main processor 30 that analyzes the simple uroflow test result paper using the feature points learned by the learning model and determines whether the patient has a lower urinary tract symptom.

[0052] The contents of the above-mentioned method are applied to the lower urinary tract symptoms diagnosis system according to the present invention, so that the description of the same parts of the lower urinary tract symptoms diagnosis system as those of the above-mentioned method will be omitted.

[0053] The above description of the present invention is for illustrative purposes only, and those skilled in the art will understand that the present invention can be easily modified into other specific forms without changing the technical idea or essential features of the present invention. Therefore, it should be understood that all the above-described embodiments are illustrative and not limiting. For example, each component described as a single type may be implemented in a distributed manner, and similarly, each component described as distributed may be implemented in a combined form.

[0054] The scope of the present invention is defined by the claims set forth below rather than the above detailed description, and all modifications and variations that fall within the meaning and scope of the claims and their equivalents should be construed as being included within the scope of the present invention.

Claims

1. A lower urinary tract symptom diagnostic system, comprising: an input step of receiving an input of a simple urinary flow test result sheet to be diagnosed; wherein the lower urinary tract symptom diagnosis system includes a character learning model that extracts character data from the simple uroflow test result paper and extracts feature points correlating with causes of lower urinary tract symptoms from the character data, and a graph learning model that extracts graph data from the simple uroflow test result paper and extracts feature points correlating with causes of lower urinary tract symptoms from the graph data, a feature point integration step in which the lower urinary tract symptom diagnosis system integrates a plurality of the feature points extracted from the character learning model and a plurality of the feature points extracted from the graph learning model; A deep learning-based method for diagnosing lower urinary tract symptoms, comprising a diagnosis step in which the character learning model and the graph learning model analyze the correlation between the integrated feature points and lower urinary tract symptoms to determine whether the simple urinary flow test result paper corresponds to a lower urinary tract symptom.

2. 2. The deep learning-based method for diagnosing lower urinary tract symptoms according to claim 1, wherein the text data includes at least one of a point of maximum urinary flow rate (Qmax) during urination, a voiding time, a post-void residual (PVR), and a bladder filling volume (BFV).

3. The deep learning-based method for diagnosing lower urinary tract symptoms according to claim 1 , wherein the graph data includes at least one of a urination volume over time and a urination rate over time.

4. The graph learning model extracting the graph data comprises: a start point extraction step of extracting a point where a graph fluctuation starts in the graph data as a start point of urination; an end point extraction step of extracting a point where the graph fluctuation ends in the graph data as an end point of urination; The deep learning-based lower urinary tract symptom diagnosis method according to claim 1 , further comprising a preprocessing step of extracting a section from the start point to the end point and inputting the section into the graph learning model.

5. The diagnosis step includes: The deep learning-based method for diagnosing lower urinary tract symptoms according to claim 1 , further comprising a step of combining the result of the judgment with symptoms corresponding to lower urinary tract symptoms and expressing the result in a binary or quaternary system.

6. A computer-readable recording medium having a program recorded thereon for implementing the method of claim 1.

Citation Information

Patent Citations

  • Urination management system

    JP2007233949A

  • Analysis data analytics method and analysis data analytics device

    WO2018134952A1