Endothelial function detection method based on machine vision and embedded control

Abstract

The invention discloses an endothelial function detection method based on machine vision and embedded control, and belongs to the field of digital image processing and the field of embedded control. The method mainly aims at solving the problems existing in endothelial function detection that the endothelial function serves as an evaluation index of cardiovascular capacity and has important reference value, and a traditional method has the defects of wound, complexity and man-made interference. According to the method, the cuff is accurately pressurized in an automatic pressurization mode, meanwhile, fingertip videos are collected through a camera, FMD indexes of the collected videos are calculated through a digital image processing technology based on machine vision, and therefore the simple, convenient and accurate endothelial function evaluation method is completed through evaluation.

Description

Technical Field

[0001] The present invention belongs to the fields of digital image processing and embedded control, and specifically embodies an automatic pressurization function completed by a relay, a motor, an air pump, a pressure gauge, a pressure sensor, and a cuff. A camera and a light source simultaneously perform video acquisition, and finally a method for calculating the endothelial function evaluation index FMD using digital image processing technology. Technical Background

[0002] In addition to some invasive methods, traditional endothelial function detection methods mainly include those for large blood vessels and microvessels. For example, the FMD method introduced in 1992 for large blood vessels uses ultrasound to measure the diameter of the brachial artery. It not only requires experienced operators, but also high-quality ultrasound equipment, as well as the judgment of medical staff on ultrasound images, which leads to a large interference from human factors in this method. Summary of the Invention

[0003] The present invention mainly aims at the problems existing in endothelial function detection: endothelial function, as an evaluation index of cardiovascular ability, has important reference value, while traditional methods have the disadvantages of being invasive, complex, and subject to human interference. This method uses an automatic pressurization method to accurately pressurize the cuff. At the same time, the camera collects fingertip videos, and the collected videos are used to calculate the FMD index through digital image processing technology based on machine vision, so as to complete a simple and accurate endothelial function evaluation.

[0004] The technical solution of the present invention is an endothelial function detection method based on machine vision and embedded control. The detection device includes: a power supply, a camera, a host, a relay, an air pump, a solenoid valve, a cuff, a pressure sensor, and an ADC module. The power supply supplies power to the solenoid valve, the air pump, the relay, and the host. The host controls the relay, the camera, the pressure sensor, and the ADC module. The relay controls the air pump and the solenoid valve. The air pump and the solenoid valve jointly control the inflation and deflation of the cuff. The pressure sensor collects the pressure of the cuff, and the ADC module performs analog-to-digital conversion on the data collected by the pressure sensor. The camera is used to collect fingertip images and transmit the collected images to the host, and the host analyzes the endothelial function detection result.

[0005] Furthermore, the detection method of this device is as follows:

[0006] Step 1: Select one arm and put on the cuff, place the cuff on the upper arm, about two or three centimeters above the elbow fossa.

[0007] Step 2: Insert the index fingers of both arms into the fixed sponge of the image acquisition device respectively, and the camera in the image acquisition device collects the images of the index finger pulp.

[0008] Step 3: The camera starts to collect images; before the air pump pressurizes the cuff, the camera collects fingertip static images for a period of time.

[0009] Step 4: During the pressurization stage, the air pump inflates the cuff, and the finger needs to be kept stable during this stage.

[0010] Step 5: After the pressurization stage ends, it enters the final static stage. After a period of time in the static stage, the camera is turned off; the cuff is deflated, and the data collected by the camera is transmitted to the host computer.

[0011] Step 6: The host computer analyzes the data and obtains the endothelial function assessment and FMD value through an algorithm.

[0012] Further, the specific steps of Step 3 are as follows:

[0013] Step 3-1: Enter the static stage.

[0014] Step 3-2: Turn off the air pump, close the solenoid valve, and turn on the camera for collection.

[0015] Step 3-3: Obtain the time of the static stage according to the parameters set in the software.

[0016] Step 3-4: The default duration of the static stage is 3 minutes and 30 seconds, and the camera continuously collects videos during the static stage.

[0017] Further, the specific steps of Step 4 are as follows:

[0018] Step 4-1: Conduct the pressurization stage.

[0019] Step 4-2: The air pump is powered to inflate the cuff, the solenoid valve is closed to prevent air leakage, and the cuff air pressure value is collected through a pressure sensor.

[0020] Step 4-3: Obtain the time of the pressurization stage according to the parameters set in the software. The default time of the pressurization stage is 5 minutes.

[0021] Step 4-4: Obtain the pressurization pressure according to the parameters set in the software. The default pressurization pressure is 220 mmHg.

[0022] Step 4-5: The cuff continues to inflate until it reaches 220 mmHg.

[0023] Step 4-6: After the inflation reaches 220 mmHg, the air pump stops working.

[0024] Step 4-7: During the pressurization stage, the air pump will start intermittently to keep the pressure between 217 mmHg and 223 mmHg.

[0025] Step 4-8: The camera continuously works to collect videos during the pressurization stage.

[0026] Furthermore, the specific steps of step 5 are as follows:

[0027] Step 5-1: After the pressurization stage ends, the solenoid valve is energized, and the cuff is deflated until the air pressure shown on the pressure gauge is 0, and then the solenoid valve is closed;

[0028] Step 5-2: Enter the final static stage, keep the measured arm and finger static, and the camera continues to collect fingertip videos;

[0029] Step 5-3: Obtain the time of the static stage according to the parameters set in the software, with a default of 3 minutes and 30 seconds

[0030] Step 5-4: Wait for 3 minutes and 30 seconds to complete the final static stage, and control the camera to end the collection;

[0031] Step 5-5: Save the collected videos and transmit them to the host computer.

[0032] Furthermore, the specific steps of step 6 are as follows:

[0033] Step 6-1: Through the adaptive ROI selection and dynamic tracking algorithm based on deep learning for the collected videos, extract the ROI region;

[0034] Step 6-2: Perform method processing through the video network amplification structure based on deep learning to extract RGB video and infrared thermal image data;

[0035] Step 6-3: Draw an envelope line for the data obtained in 6-2;

[0036] Step 6-4: Calculate the equivalent FMD value and output it for display on the software interface.

[0037] The present invention is an endothelial function detection scheme based on machine vision and embedded control. This scheme first uses rk3588 to control the air pump and solenoid valve respectively to complete the inflation and deflation of the cuff during the pre-pressurization static stage, pressurization stage, and post-pressurization static stage. At the same time, it controls the camera to collect fingertip videos. Finally, it uses deep learning and image processing technologies to process the collected videos to obtain accurate evaluation of the FMD index and endothelial function. The present invention has developed a new evaluation method for endothelial function based on machine vision. Compared with some previous invasive methods and traditional FMD methods, it no longer requires high-precision acquisition equipment such as ultrasonic waves and complex manual operations, which can make the evaluation of endothelial function simpler, enabling more and more people to evaluate through simple equipment, playing a role in early prevention of cardiovascular diseases. Brief Description of the Drawings

[0038] Figure 1 It is a schematic diagram of the physical structure of the present invention.

[0039] Figure 2 It is a schematic diagram of the software interface of the present invention.

[0040] Figure 3 It is a flowchart of the automatic pressurization module. Specific implementation manners

[0041] This method uses the rk3588 to control the air pump and solenoid valve to automatically pressurize the arm, and at the same time controls the camera to collect images of the fingertips. It performs image processing on the collected fingertip videos before and after congestion, calculates the FMD index, avoids using ultrasonic equipment to collect artery images, makes the operation simpler, and does not reduce the accuracy of the results. The software is used to start and stop the device and output the results, avoiding dependence on manual operations. In this way, the endothelial function is evaluated, making the detection of endothelial function more convenient and more conducive to the prevention of cardiovascular diseases.

[0042] The following will combine with the accompanying drawings to elaborate in detail on a detection scheme for endothelial function based on machine vision and embedded control in the present invention:

[0043] Step 1: Select one arm and put on the cuff, place the cuff on the upper arm, about two or three centimeters above the elbow pit;

[0044] Step 2: Adjust the tightness of the cuff to achieve a comfortable state and ensure that it will not loosen or slip during the detection process;

[0045] Step 3: Insert the index fingers of both arms into the fixed sponge of the collection device, and stretch the fingers forward as much as possible until reaching the bottom;

[0046] Step 4: Rotate the fingers to adjust the angle so that the palm is placed flat upwards, that is, the position of the index finger pulp is facing the collecting camera, and the whole arm is in a static and flat state, waiting for the start of collection;

[0047] Step 5: Click the start collection button on the software interface. The software interface will simultaneously prompt to wear the cuff and fix the fingers properly. Check whether the cuff on the arm and the fingers in the collection device are correct. If adjustment is still needed, click the forced stop button on the software interface, and after readjusting, click the start collection again;

[0048] Step 6: Enter the static stage before pressurization. The camera starts to collect fingertip images. The tested person needs to keep the fingers fixed, and at the same time keep breathing smoothly and the mind calm, without strong emotional fluctuations. If necessary, the forced stop button can also be clicked for adjustment;

[0049] Step 6-1: The software interface prompts to enter the static stage;

[0050] Step 6-2: The rk3588 controls to turn off the air pump, close the solenoid valve, and turn on the camera for acquisition;

[0051] Step 6-3: The static stage lasts for 3 minutes and 30 seconds (which can be modified through software, and 3 minutes and 30 seconds is the default duration). During the static stage, the camera continuously acquires videos;

[0052] Step 7: After the static stage ends, it automatically enters the pressurization stage. The cuff will automatically inflate, and the finger needs to remain stable during the pressurization stage;

[0053] Step 7-1: At the start of the pressurization stage, the software will simultaneously prompt the start of the pressurization stage;

[0054] Step 7-2: The air pump is powered to inflate the cuff, the solenoid valve is closed to prevent air leakage, and the pressure at this time can be known through the pressure gauge;

[0055] Step 7-3: The cuff continues to inflate until it reaches the default value of 220 mmHg (this air pressure can be set through software before testing). During the pressurization process, the subject may feel slightly uncomfortable, but the measured arm and finger need to remain stationary at all times;

[0056] Step 7-4: After the inflation reaches 220 mmHg, the air pump stops working. During the pressurization stage, the air pressure may drop, and the air pump will start at any time to ensure that the air pressure is stable at about 220 mmHg;

[0057] Step 8: After the pressurization stage ends, it enters the final static stage, which is the same as the static stage before pressurization. Do not have violent emotional fluctuations and keep the finger stable;

[0058] Step 8-1: After the pressurization stage ends, the solenoid valve is energized, and the cuff deflates until the air pressure shown on the pressure gauge is 0, then the solenoid valve is closed;

[0059] Step 8-2: Enter the final static stage, keep the measured arm and finger stationary, and the camera continues to acquire fingertip videos;

[0060] Step 8-3: Similar to the previous static stage, wait for 3 minutes and 30 seconds (the time of these two static stages can also be set through software before testing). After completing the final static stage, the rk3588 controls the camera to end the acquisition;

[0061] Step 8-4: Save the acquired videos and wait for the algorithm part to read;

[0062] Step 9: After the software prompts the end of the acquisition stage, the finger can leave the acquisition device, and at the same time, the cuff can be removed;

[0063] Step 10: After waiting for the algorithm to perform image processing on the collected video, specific endothelial function evaluation and FMD value can be obtained on the software interface;

[0064] Step 10-1: Extract the ROI region from the collected video through an adaptive ROI selection and dynamic tracking algorithm based on deep learning;

[0065] Step 10-2: Process through a video network amplification structure based on deep learning to extract RGB video and infrared thermal image data;

[0066] Step 10-3: Draw an envelope for the data obtained in Step 10-2; Step 10-4: Calculate the equivalent FMD value and output it for display on the software interface.

Claims

1. An endothelial function detection method based on machine vision and embedded control. In this detection method, the following device is used for data acquisition, and the device includes: Power supply, camera, host, relay, air pump, solenoid valve, cuff, pressure sensor, ADC module. The power supply powers the solenoid valve, air pump, relay, and host. The host controls the relay, camera, pressure sensor, and ADC module. The relay controls the air pump and solenoid valve. The air pump and solenoid valve jointly control the inflation and deflation of the cuff. The pressure sensor collects the pressure of the cuff. The ADC module performs analog-to-digital conversion on the data collected by the pressure sensor. The camera is used to collect fingertip images and transmit the collected images to the host, and the host analyzes the endothelial function detection results; The detection method is as follows: Step 1: Put on the cuff on one side of the arm, place the cuff on the upper arm, about two or three centimeters above the elbow fossa; Step 2: Insert the index fingers of both arms into the fixed sponge of the image acquisition device respectively, and the camera in the image acquisition device collects the images of the index finger pulp; Step 3: The camera starts to collect images; before the air pump pressurizes the cuff, the camera collects the fingertip static images for a period of time; Step 4: In the pressurization stage, the air pump inflates the cuff, and the fingers also need to be kept stable during the pressurization stage; Step 5: After the pressurization stage ends, enter the final static stage. After a period of time in the static stage, the camera is turned off; the cuff is deflated, and the data collected by the camera is transmitted to the host. Step 6: The host analyzes the data and obtains the endothelial function evaluation and FMD value through the algorithm.

2. The endothelial function detection method based on machine vision and embedded control according to claim 1, characterized in that The specific steps of Step 3 are as follows: Step 3-1: Enter the static stage; Step 3-2: Turn off the air pump, turn off the solenoid valve, and turn on the camera for collection; Step 3-3: Obtain the static stage time according to the parameters set in the software; Step 3-4: The static stage lasts for 3 minutes and 30 seconds by default, and the camera continuously collects videos during the static stage.

3. The endothelial function detection method based on machine vision and embedded control according to claim 1, wherein, The specific steps of Step 4 are as follows: Step 4-1: Enter the pressurization stage; Step 4-2: The air pump is powered to inflate the cuff, the solenoid valve is closed to prevent air leakage, and the pressure value of the cuff is collected through the pressure sensor; Step 4-3: Obtain the pressurization stage time according to the parameters set in the software, and the pressurization stage time is 5 minutes by default; Step 4-4: Obtain the pressurization pressure according to the parameters set in the software, and the pressurization pressure is 220 mmHg by default; Step 4-5: The cuff continues to inflate until it reaches 220 mmHg; Step 4-6: After the inflation reaches 220 mmHg, the air pump stops working; Step 4-7: During the pressurization stage, the air pump will start intermittently to keep the pressure between 217 mmHg and 223 mmHg; Step 4-8: The camera continuously works to collect videos during the pressurization stage.

4. The endothelial function detection method based on machine vision and embedded control according to claim 1, characterized in that, The specific steps of Step 5 are as follows: Step 5-1: After the pressurization stage ends, the solenoid valve is powered on, the cuff is deflated, and until the air pressure shown on the pressure gauge is 0, the solenoid valve is closed; Step 5-2: Enter the final static stage, keep the measured arm and finger static, and the camera continues to collect fingertip videos; Step 5-3: Obtain the static stage time according to the set parameters, which is 3 minutes and 30 seconds; Step 5-4: Wait for 3 minutes and 30 seconds to complete the final static stage, and control the camera to end the collection; Step 5-5: Save the collected videos and transmit them to the host.

5. The endothelial function detection method based on machine vision and embedded control according to claim 1, wherein, The specific steps of step 6 are as follows: Step 6-1: Extract the ROI region from the collected video through an adaptive ROI selection and dynamic tracking algorithm based on deep learning; Step 6-2: Perform method processing through a video network amplification structure based on deep learning to extract RGB video and infrared thermal image data; Step 6-3: Draw an envelope line for the data obtained in 6-2; Step 6-4: Calculate the equivalent FMD value and output it for display on the software interface.

Images