WO2023095950A1

WO2023095950A1 - Device for cardiac quantitative analysis and self-diagnosis using artificial intelligence technology

Info

Publication number: WO2023095950A1
Application number: PCT/KR2021/017539
Authority: WO
Inventors: 이준호; 최관용; 이정민
Original assignee: 이준호; 최관용; 이정민
Priority date: 2021-11-25
Filing date: 2021-11-25
Publication date: 2023-06-01
Also published as: KR20230077065A

Abstract

Disclosed is a device for cardiac quantitative analysis and self-diagnosis using artificial intelligence technology. The device for cardiac quantitative analysis and self-diagnosis of the present invention comprises: a communication unit which receives an ultrasound image of the myocardium of the heart, captured by an ultrasound probe; and a control unit which divides the myocardium into predetermined regions on the basis of the ultrasound image to measure the maximal diastolic wall thickness and maximal systolic wall thickness of the left ventricle for each region, quantitatively analyzes the wall motions of the left ventricle on the basis of information on the measured wall thicknesses, and diagnoses the condition of the heart by applying the quantitative analysis results to an artificial intelligence model.

Description

Cardiac quantitative analysis and self-diagnosis device using artificial intelligence technology

The present invention relates to a self-diagnosis technology for the heart, and a device for quantitative analysis and self-diagnosis of the heart using artificial intelligence technology for self-diagnosis of the heart by measuring the wall thickness of the myocardium using ultrasound and applying the measurement result to an artificial intelligence model. It is about.

The heart is an essential organ that supplies blood to the entire body through myocardial contraction. Myocardium is supplied with an energy source through blood vessels called coronary arteries. If essential energy sources are not supplied due to coronary artery stenosis, myocardial infarction may occur and death may occur.

The heart has two ventricles, the left ventricle and the right ventricle, to supply blood to the body. The left ventricle plays an important role in delivering oxygenated blood from the lungs to the capillaries through contraction.

In general, in the case of a normal person, the heart beats approximately 60 times per minute with a period of 1 second. During one heart cycle, the atria and ventricles repeatedly relax and contract, and during diastole (the process of filling the ventricles), the volume of the atria and ventricles is maximized, and during systole (the process of sending blood from the ventricles to the whole body). ) the volume of the atrium and ventricle is minimal. At this time, ejection fraction, left ventricle wall thickness, left ventricular wall motion, etc. are used as the left ventricular function measurement values of the heart.

On the other hand, echocardiography is widely used to analyze the function of the left ventricle of the heart. Since echocardiography uses ultrasound waves, it is completely harmless to the human body and has high accuracy in analyzing left ventricular function.

However, since such echocardiography requires expertise, it is difficult for the general public to examine and diagnose it by themselves.

The technical problem to be achieved by the present invention is to measure the wall thickness of each of the 17 regions of the left ventricle using ultrasound, and apply the measured results to an artificial intelligence model to help the user diagnose the condition of the heart by himself. Its purpose is to provide a heart quantitative analysis and self-diagnosis device using intelligent technology.

In order to achieve the above object, an apparatus for quantitative analysis and self-diagnosis of the heart using artificial intelligence technology according to the present invention includes a communication unit that receives an ultrasound image of the myocardium of the heart captured by an ultrasound probe and a preset myocardium based on the ultrasound image. It is divided into regions to measure the wall thickness for the maximum diastole and maximum systole of the left ventricle for each region, quantitatively analyze the wall motion of the left ventricle based on the information on the measured wall thickness, and analyze the quantitative analysis result with artificial intelligence. It includes a control unit that diagnoses the state of the heart by applying it to the model.

The apparatus may further include a sensor unit for measuring an electrocardiogram of the heart, and the control unit determines a maximum diastolic period and a maximum systolic period of the left ventricle by using at least one of an image analysis of an ultrasound image and an electrocardiogram analysis.

In addition, the control unit measures the maximum diastolic left ventricular wall thickness (EDT) and maximum systolic left ventricular wall thickness (EST), calculates a difference between the maximum diastolic left ventricular wall thickness and the maximum systolic left ventricular wall thickness, and determines the maximum systolic left ventricular wall thickness as After dividing by the difference value, it is characterized in that quantitative analysis is performed by percentage.

In addition, the control unit may learn the artificial intelligence model based on the diagnosed result.

The controller may further include an output unit for outputting information related to the diagnosis, and the control unit outputs a guide description for assisting operation of the ultrasound probe to a portion of the output unit so that an ultrasound image capable of measuring a wall thickness of each region is received. It is characterized by doing

The method for quantitative analysis and self-diagnosis of the heart using artificial intelligence technology according to the present invention includes the steps of receiving an ultrasound image in which the myocardium of the heart is photographed by an ultrasound probe in the device for quantitative analysis and self-diagnosis of the heart, and the device for quantitative analysis and self-diagnosis of the heart. dividing the myocardium into preset regions based on the ultrasound image and measuring the wall thickness of the left ventricle for each region during maximum diastole and maximum systole; Quantitatively analyzing the wall motion of the left ventricle based on the information, and diagnosing a state of the heart by applying the quantitatively analyzed result to an artificial intelligence model by the device for quantitatively analyzing and self-diagnosing the heart.

The cardiac quantitative analysis and self-diagnosis device using the artificial intelligence technology of the present invention guides even a novice to easily measure the wall thickness of the left ventricle using ultrasound.

In addition, by automatically diagnosing the condition of the heart by applying the result of the wall thickness of the measured left ventricle to the artificial intelligence model, the user can check the condition of the heart by himself even without professional knowledge.

1 is a configuration diagram illustrating a system for quantitative cardiac analysis and self-diagnosis according to an embodiment of the present invention.

2 is a block diagram illustrating an apparatus for quantitative cardiac analysis and self-diagnosis according to an embodiment of the present invention.

3 is a diagram for explaining how the myocardium of the heart is divided into 17 regions.

4 is a diagram for explaining the functions of a device for quantitative cardiac analysis and self-diagnosis according to an embodiment of the present invention.

5 is a diagram for explaining the screen output of the heart quantitative analysis and self-diagnosis device according to an embodiment of the present invention.

6 is a diagram for explaining the configuration of an artificial neural network according to an embodiment of the present invention.

7 is a diagram for explaining a node performing an operation to which a weight is applied according to an embodiment of the present invention.

8 is a flowchart illustrating a method for quantitative cardiac analysis and self-diagnosis according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. First, in adding reference numerals to the components of each drawing, it should be noted that the same components have the same numerals as much as possible even if they are displayed on different drawings. In addition, in describing the present invention, if it is determined that a detailed description of a related known configuration or function is obvious to those skilled in the art or may obscure the gist of the present invention, the detailed description will be omitted.

Referring to FIG. 1 , the heart quantitative analysis and self-diagnosis system 300 helps beginners to take an ultrasound image of the heart by themselves and automatically diagnoses the state of the heart using the captured image information. The heart quantitative analysis and self-diagnosis system 300 diagnoses conditions using an artificial intelligence model, thereby providing an accurate diagnosis to the user. The heart quantitative analysis and self-diagnosis system 300 includes a heart quantitative analysis and self-diagnosis device 100 (hereinafter referred to as 'device') and an ultrasound probe 200.

The device 100 communicates with the ultrasound probe 200, and for this purpose, a program (or application) supporting interworking with the ultrasound probe 200 is installed. The device 100 may perform self-diagnosis of the heart according to the present invention through an installed program.

The device 100 measures the wall thickness of the left ventricle based on an ultrasound image of the myocardium of the heart captured by the ultrasound probe 200 . Here, the wall thickness of the left ventricle means the distance between the left ventricle inner membrane and the left ventricle outer membrane. The apparatus 100 divides the myocardium into preset regions and measures the wall thickness of the left ventricle for each region. There may be 17 preset areas. The device 100 quantitatively analyzes wall motion of the left ventricle based on the measured wall thickness information. The device 100 analyzes the wall motion of the left ventricle using the maximal diastole and maximal systole of the left ventricle. At this time, the device 100 may determine the maximum diastolic period and the maximum systolic period through image analysis such as the maximum and minimum sizes of the left ventricle, or may determine the maximum diastolic period and maximum systolic period using information on the electrocardiogram, but is not limited thereto. The device 100 diagnoses the condition of the heart by applying the result of quantitative analysis to an artificial intelligence model. Here, the artificial intelligence model may be various artificial intelligence models such as an artificial neural network (ANN), machine learning, and deep learning. Device 100 may be a computing system including a smart phone, smart watch, desktop, laptop, tablet PC, handheld PC, or the like.

The ultrasound probe 200 transmits and receives data through communication with the device 100 . At this time, the ultrasonic probe 200 is installed with a program (or application) supporting interworking with the device 100 . The ultrasound probe 200 generates ultrasound waves to image a heart, and generates an ultrasound image of the imaged heart. The ultrasound probe 200 transmits the generated ultrasound image to the device 100 in real time. Here, the ultrasound probe 200 may be portable, and preferably may be a wireless probe.

Meanwhile, in the heart quantitative analysis and self-diagnosis system 300 , a communication network 350 may be established between the device 100 and the ultrasound probe 200 . The communication network 350 may support wired/wireless communication, preferably short-distance wireless communication, but is not limited thereto.

FIG. 2 is a block diagram for explaining a heart quantitative analysis and self-diagnosis device according to an embodiment of the present invention, FIG. 3 is a diagram for explaining how the myocardium of the heart is divided into 17 regions, and FIG. It is a diagram for explaining the functions of the device for quantitative analysis and self-diagnosis of the heart according to an embodiment of the present invention, and FIG. 5 is a diagram for explaining the screen output of the device for quantitative analysis and self-diagnosis of the heart according to an embodiment of the present invention. 6 is a diagram for explaining the configuration of an artificial neural network according to an embodiment of the present invention, and FIG. 7 is a diagram for explaining a node performing an operation to which weights are applied according to an embodiment of the present invention.

1 to 7, the apparatus 100 measures the wall thickness of 17 regions of the left ventricle using sound waves, and applies the measured results to an artificial intelligence model to diagnose the state of the heart by the user himself. help you do it The device 100 includes a communication unit 10 and a control unit 40, and may further include an input unit 20, a sensor unit 30, an output unit 50, and a storage unit 60.

The communication unit 10 performs communication with the ultrasonic probe 200 . The communication unit 10 receives an ultrasound image from the ultrasound probe 200 in real time. Here, the ultrasound image may be an image of the myocardium of the heart.

The input unit 20 receives a user input for driving the device 100 . The input unit 20 supports various input methods such as voice input, touch input, and keypad input, and may include a microphone, a touch screen, a keypad, and a mouse for this purpose. Here, when the input unit 20 is implemented as a touch screen, the input unit 20 and the output unit 50 may be implemented in one configuration.

The sensor unit 30 measures the electrocardiogram of the heart. The sensor unit 30 may implement an electrocardiogram sensor in a wearable form, preferably in a wristwatch form, but is not limited thereto.

The controller 40 performs overall control of the device 100 . The control unit 40 controls the output unit 50 to output the ultrasound image received from the communication unit 10 in real time. At this time, the controller 40 controls the region for measuring the wall thickness of the left ventricle for each region to be performed later as a specific mark (color, section, point, etc.) 42, and measures the wall thickness of the left ventricle in the correct direction. A guide description assisting the operation of the ultrasound probe 200 is controlled to be output on a part 44 of the output screen. Here, the guide description may be in the form of a compass, or may be in the form of indicating a direction or rotation of at least one of the X-axis, Y-axis, Z-axis, roll, pitch, and yaw.

The controller 40 divides the myocardium into 17 predetermined regions based on the received ultrasound image, and measures the wall thickness of the left ventricle for each region. Here, as shown in FIG. 3, the 17 regions are divided into basal, mid, and apical based on the short axis and long axis of the heart, and the divided regions are It means an area divided into 17 parts. At this time, each region is connected to blood vessels of the heart, and in detail, the first region, the second region, the seventh region, the eighth region, the thirteenth region, and the fourteenth region are connected to LAD (left anterior descending a.), The third region, the ninth region, and the fifteenth region are connected to left circumflex a. (LCX) and LAD, the fourth region and the tenth region are connected to LCX and right coronary a. (RCA), and the fifth region, The sixth region, the eleventh region, the twelfth region, and the sixteenth region are connected to the RCA. When measuring the wall thickness of the left ventricle, the controller 40 measures the maximum diastolic left ventricular wall thickness (EDT) and maximum systolic left ventricular wall thickness (EST) for each region. In detail, the controller 40 determines the maximum diastolic period and maximum systolic period of the left ventricle by using at least one of image analysis of an ultrasound image and electrocardiogram analysis, and measures the thickness at the determined time point. That is, the control unit 40 analyzes the ultrasound image to detect when the left ventricle is at its maximum or minimum, measures the wall thickness at the detected time point ( FIG. 3A ), or uses the information on the electrocardiogram to determine the maximum diastolic and maximum diastolic phases. The wall thickness can be measured by detecting the systole and measuring the wall thickness at the detected time point ( FIG. 3B ), or by using both of the above methods.

The controller 40 quantitatively analyzes the wall motion of the left ventricle based on the measured wall thickness information. Here, the wall motion means the change rate of the wall thickness. The controller 40 may perform quantitative analysis by calculating a difference between the maximum diastolic left ventricular wall thickness and the maximum systolic left ventricular wall thickness, dividing the maximum systolic left ventricular wall thickness by the calculated difference value, and then calculating a percentage (%). That is, the control unit 40 may perform quantitative analysis as shown in [Equation 1].

Here, EDT is the maximum diastolic left ventricular wall thickness, and EST is the maximum systolic left ventricular wall thickness.

The control unit 40 diagnoses the condition of the heart by applying the result of the quantitative analysis to the artificial intelligence model. Here, the artificial intelligence model may be various artificial intelligence models such as artificial neural networks, machine learning, and deep learning.

For example, FIG. 6 illustrates an artificial neural network, which is an example of an artificial intelligence model, and as shown, the artificial neural network includes a plurality of layers. These plurality of layers include an input layer (IL), a hidden layer (HL1 to HLk), and an output layer (OL).

In addition, each of the plurality of layers (IL, HL, OL) includes one or more nodes. For example, as shown, the input layer IL may include n input nodes i1 to in, and the output layer OL may include one output node v. In addition, among the hidden layers HL, the first hidden layer HL1 includes a number of nodes h11 to h1a, the second hidden layer HL2 includes b nodes h21 to h2b, and the kth hidden layer (HLk) may include c nodes (hk1 to hkc).

Each of the nodes of the plurality of layers performs an operation. In particular, a plurality of nodes of different layers are connected by a channel (shown as a dotted line) having a weight (w: weight). In other words, the calculation result of any one node is input to the node of the next layer after applying the parameters of the artificial neural network including the weight (w) or threshold (b). This connection relationship will be described based on nodes.

7 shows an example of a node h according to an embodiment of the present invention. Although the node of FIG. 7 is described as one of the hidden nodes h, it can be commonly applied to all nodes included in the artificial intelligence model. The operation of this node (h) is performed according to the following [Equation 2].

Here, function F means an activation function. In addition, x is a node value according to an operation result of each of a plurality of nodes of the previous layer, and denotes an input to a node h of the next layer. These inputs are x=[x1, x2, … , xn]. w is the weight corresponding to the input x, and w=[w1, w2, . . . , wn]. According to [Equation 2], node h is an input x = [x1, x2, . , xn] with weights w=[w1, w2, … , wn], the result is taken as the function F. Here, the parameter b is a threshold, and the threshold b is in [Equation 2]

It serves to prevent the corresponding node from being activated when the value of is smaller than the threshold.

For example, it is assumed that there are 3 nodes in the hierarchy prior to node h (n=3). Accordingly, there are three inputs x1, x2, and x3 and three weights w1, w2, and w3 for node h. The node h receives a value obtained by multiplying the three inputs x1, x2, and x3 by weights w1, w2, and w3, sums them all, and substitutes the summed value into a transfer function to calculate an output. Specifically, it is assumed that the inputs [x1, x2, x3] = 0.5, -0.3, 0, and the weights w = [w1, w2, w3] = 4, 5, 2. In addition, assuming that the activation function F is 'sgn()' and the threshold is 0.01, the output is calculated as follows through the operation according to [Equation 2].

x1 × w1 = 0.5 × 0.19 = 0.095

x2 × w2 = -0.3 × 0.25 = -0.075

x3 × w3 = 0 × 0.66 = 0

0.095 + (-0.075) + 0 = 0.02

When b=0.01, sgn(0.02-0.01) = 1

In this way, each of the plurality of nodes for the plurality of layers generated from the artificial neural network receives a value obtained by applying weights w1, w2, w3 and threshold b, which are parameters of the artificial neural network, to node values x1, x2, and x3 of the previous layer, Calculate the output value OUT by performing the operation by the activation function F. When the next layer exists, the calculated output value OUT becomes an input to a node of the next layer. That is, any one node of any one layer of the artificial intelligence model receives a value obtained by applying a weight or threshold to the output of the node of the previous layer, performs an operation by the activation function F, and transfers the operation result to the next layer. .

The overall calculation of the artificial intelligence model is performed as follows. First, the results of quantitative analysis are used as input data for the artificial intelligence model. The input data may be converted into an input feature vector and then input. The input feature vector IV has a plurality of element values iv1 to ivn corresponding to a plurality of input nodes i1 to in of the input layer IL of the artificial intelligence model. Accordingly, when the input feature vector IV = [iv1 to ivn] is input to the plurality of input nodes i1 to in of the input layer IL, the plurality of first hidden nodes h11 to h1a of the first hidden layer HL1 Each applies weights and thresholds to a plurality of element values iv1 to ivn of a plurality of input nodes (i1 to in), and performs an operation according to an activation function for each of the plurality of element values of the input feature vector to which the weights and thresholds are applied. Thus, a plurality of first hidden node values are calculated.

Then, each of the plurality of second hidden nodes h21 to h2b of the second hidden layer HL2 applies a weight and a threshold value to each of the plurality of first hidden node values of the plurality of first hidden nodes h11 to h1a, and the weight and calculating a plurality of second hidden node values by performing an operation according to an activation function for each of the plurality of first hidden node values to which the threshold value is applied. In this way, in the hidden layer (HL), a previous node value is transmitted with a weight applied, and a current node value is calculated through an operation. By repeating this process, a plurality of k-th hidden node values of the plurality of k-th hidden nodes (hk1 to hkc) of the k-th hidden layer (HLk) may be calculated.

Accordingly, the output node v of the output layer OL applies a weight w = [w1, w2, . . . , wc] is applied (indicated by a dotted line), the input values are summed up, the threshold value is subtracted, and the corresponding value is calculated according to the activation function to calculate the output value. The output value calculated by the output node v may be an estimated value of the driving control value.

As described above, the controller 40 estimates the diagnosis of the heart condition through the artificial neural network, but is not limited thereto and may estimate the diagnosis of the heart condition using various artificial intelligence models.

The control unit 40 generates diagnostic information tailored to the user based on the estimated diagnosis result. Diagnosis information is generated based on the user's personal information such as age, gender, and educational background, and refers to information summarized so that one can intuitively recognize where the heart is hurting just by checking the contents.

The control unit 40 transfers the diagnostic information to the output unit 50 and controls it to be output, and learns the diagnostic information at the same time. That is, the control unit 40 improves the accuracy of a diagnosis result to be performed later by learning an artificial intelligence model based on diagnosis information, which is a diagnosis result.

The output unit 50 outputs an ultrasound image 41, a quantitative analysis table 43, and diagnosed information. In addition, the output unit 50 may display a region for measuring the wall thickness of the left ventricle with a specific display 42 and output a guide description 44 for assisting the operation of the ultrasound probe 200 on a portion of the screen. The output unit 50 may be a display, such as a liquid crystal display (LCD), a thin film transistor-liquid crystal display (TFT LCD), or an organic light-emitting diode (OLED). , a flexible display, and a 3D display.

The storage unit 60 stores programs or algorithms for driving the device 100 . The storage unit 60 stores ultrasound images, electrocardiogram information, quantitatively analyzed information, and diagnosed information. The storage unit 60 is a flash memory type, a hard disk type, a multimedia card micro type, a card type memory (eg SD or XD memory, etc.), RAM (Random Access Memory, RAM), SRAM (Static Random Access Memory), ROM (Read-Only Memory, ROM), EEPROM (Electrically Erasable Programmable Read-Only Memory), PROM (Programmable Read-Only Memory), magnetic memory, It may include at least one storage medium of a magnetic disk and an optical disk.

Claims

a communication unit for receiving an ultrasound image of a myocardium of a heart photographed by an ultrasound probe; and

Based on the ultrasound image, the myocardium is divided into preset regions to measure the wall thickness of the left ventricle during maximum diastole and maximum systole for each region, and quantitatively analyze the wall motion of the left ventricle based on the information on the measured wall thickness. and a controller for diagnosing the state of the heart by applying the result of the quantitative analysis to an artificial intelligence model;

Cardiac quantitative analysis and self-diagnosis device using artificial intelligence technology including.
According to claim 1,

It further includes; a sensor unit for measuring the electrocardiogram of the heart;

The control unit,

A heart quantitative analysis and self-diagnosis device using artificial intelligence technology, characterized in that for determining the maximum diastolic period and maximum systolic period of the left ventricle using at least one of image analysis and electrocardiogram analysis of an ultrasound image.
According to claim 1 or 2,

The control unit,

The maximum diastolic left ventricular wall thickness (EDT) and maximum systolic left ventricular wall thickness (EST) were measured, the difference between the maximum diastolic left ventricular wall thickness and the maximum systolic left ventricular wall thickness was calculated, and the maximum systolic left ventricular wall thickness was divided by the difference. After, quantitative analysis of the heart and self-diagnosis device using artificial intelligence technology, characterized in that the quantitative analysis is performed by percentage.
According to claim 1,

The upper control unit,

Cardiac quantitative analysis and self-diagnosis device using artificial intelligence technology, characterized in that for learning the artificial intelligence model based on the diagnosed result.
According to claim 1,

Further comprising an output unit for outputting contents related to the diagnosis;

The control unit,

A heart quantitative analysis and self-diagnosis device using artificial intelligence technology, characterized in that a guide description for assisting the operation of the ultrasound probe is output to a part of the output unit so that an ultrasound image capable of measuring wall thickness for each region is received.
receiving an ultrasound image of myocardium captured by an ultrasound probe by a device for quantitative cardiac analysis and self-diagnosis;

dividing the myocardium into preset regions based on the ultrasound image, and measuring, by the device for quantitative cardiac analysis and self-diagnosis, wall thicknesses of the left ventricle during maximum diastole and maximum systole for each region;

quantitatively analyzing wall motion of the left ventricle based on the measured wall thickness information by the heart quantitative analysis and self-diagnosis device; and

diagnosing a state of the heart by applying the quantitative analysis result to an artificial intelligence model by the heart quantitative analysis and self-diagnosis device;

Cardiac quantitative analysis and self-diagnosis method using artificial intelligence technology including.