Skin cholesterol noninvasive detection equipment for identifying fluorescence color based on deep learning

Abstract

The invention relates to the technical field of biological detection, and discloses skin cholesterol noninvasive detection equipment for recognizing fluorescence colors based on deep learning. According to the equipment, a sulfydryl-beta-cyclodextrin gold nano-cluster solution is used as a fluorescent probe, and the detection of skin cholesterol is realized by adopting the principle that the fluorescence intensity and color of the sulfydryl-beta-cyclodextrin gold nano-cluster solution can change along with the change of the concentration of cholesterol and an obvious linear relation exists in a specific color space. The equipment comprises a main control module, an acquisition module and an upper computer. The main control module is used for controlling the excitation light source, the CMOS camera and the USB communication module, and is provided with an LCD display module; the acquisition module is internally provided with a darkroom environment and is used for fixing a fluorescence excitation light source light path and a signal acquisition light path and filtering interference of excitation light through a built-in light filter; and the upper computer is used for processing the acquired picture data, including identifying and positioning the fluorescent probe in the acquired picture and extracting the chromatic value of the fluorescent probe, and finally calculating the skin cholesterol concentration.

Description

Technical Field

[0001] The present invention relates to the field of biological detection technology, and in particular to a non-invasive skin cholesterol detection device based on deep learning to identify fluorescent colors. Background Art

[0002] Cardiovascular disease is the leading cause of death worldwide, and its incidence continues to rise. Atherosclerosis is the pathological basis of cardiovascular disease, and high levels of skin cholesterol deposition are early warning signs of atherosclerosis. Recent clinical studies have shown that skin cholesterol can be used for risk screening of atherosclerosis and predict the risk of cardiovascular disease.

[0003] Conventional skin cholesterol determination mainly includes skin biopsy, tape stripping, and extraction. Skin biopsy causes pain to the subject when obtaining skin samples, and there is a wound at the biopsy site, which poses a risk of infection; tape stripping is a minimally invasive test; the chemical extract of the extraction method is generally toxic and irritating, which can cause damage to the skin. In recent years, fluorescence has played an increasingly important role in the field of bioanalysis research. At present, there have been reports on the use of gold nanoparticles modified with β-cyclodextrin (β-CD) for the fluorescence detection of human serum cholesterol content.

[0004] The previous research results of the inventor team of this application (publication number CN113801650A, patent number ZL202110988479.4) show that after the thiol-β-cyclodextrin gold nanoclusters (CD-β-AuNCs) are in contact with cholesterol, the fluorescence intensity of CD-β-AuNCs itself under ultraviolet light can be enhanced, which can be used to detect serum cholesterol content. In order to achieve non-invasive detection of skin cholesterol, the inventor team of this application further studied and found that the fluorescence color of CD-β-AuNCs with added cholesterol is linearly correlated with the cholesterol concentration. Through deep learning to identify fluorescent images, the skin cholesterol detection device based on HSV color detection of this application was established. Summary of the invention

[0005] The present invention is mainly based on the problems existing in the existing conventional skin cholesterol determination method, and designs a non-invasive skin cholesterol detection device based on deep learning to identify fluorescent color.

[0006] In order to design the above detection equipment, the present invention adopts the following technical solutions:

[0007] A non-invasive skin cholesterol detection device based on deep learning to identify fluorescent colors, as shown in Figure 1- Figure 7 As shown, it includes an acquisition module, a main control module and a host computer; among which:

[0008] The acquisition module consists of a cover plate, a slide, a main body and a bottom cover. The cover plate is provided with a detection hole and a control hole; a camera is provided on the shooting platform at the bottom of the main body, and the lens is fixed to the shooting hole to vertically collect fluorescent signals. The main body is spliced ​​with the slide to form a darkroom inside. The main body is also provided with a filter and an excitation light source. The slide is provided with quartz glass. The excitation light source is irradiated at a 45° angle on the collection area of ​​the slide, and then passes through the quartz glass to irradiate the detection hole and the control hole of the cover plate;

[0009] The main control module includes a main control module top cover, a main control circuit board and a main control module bottom cover. The main control information is displayed on the display screen.

[0010] The host computer is used to receive the fluorescent image signal collected by the main control module and analyze and process it;

[0011] In the technical solution described in the present invention, the main control module chip uses STM32F407ZGT6 and is written in Keil-uVision 5.

[0012] In the technical solution described in the present invention, the host computer applies Yolov8n, a target detection algorithm based on deep learning, and uses PyCharm to write the host computer program according to the target recognition model trained according to the data set collected in this experiment.

[0013] In the technical solution described in the present invention, a USB serial port communication mode is adopted between the host computer and the main control module.

[0014] In the technical solution described in the present invention, the excitation light source is two ultraviolet LED lamps with the same wavelength of 365nm or 370nm, which are respectively located on the front and back sides of the acquisition module body.

[0015] In the technical solution described in the present invention, the filter selects a 400nm-450nm high-pass filter to filter out 365nm violet light interference, leaving blue-green fluorescence around 450nm;

[0016] In the technical solution described in the present invention, the camera selects an OV5640 camera, which is controlled by a main control module, and the control interface is displayed on an LCD display screen;

[0017] The present invention also provides a method for measuring skin cholesterol concentration using the above-mentioned device, comprising the following steps:

[0018] Step 1: Insert the quartz glass into the glass slot, then place the cover plate on the upper part of the slide slot and turn on the two excitation light sources. Click the LED control button on the main control module to adjust the LED light intensity until the LCD display shows a value of 300.

[0019] Step 2: Click the camera interface display button to turn on the camera, transmit the real-time image of the camera to the LCD display, and adjust the position of the cover plate so that the detection hole and the control hole are located in the center of the two excitation light sources respectively;

[0020] Step 3: Open the host computer, select the serial port and set the baud rate to 1500000, and open the serial port transmission;

[0021] Step 4: Place the colorless sodium alginate control solution without CD-β-AuNCs (hereinafter referred to as the control solution) in the control well, and place the light yellow sodium alginate detection solution with CD-β-AuNCs (hereinafter referred to as the detection solution) in the detection well. Then, place the hypothenar of the hand over the two wells, click the USB transfer button to start USB data transmission, and transfer the real-time image collected by the camera to the host computer.

[0022] Step 5: Locate the control well based on the deep learning training model, extract the RGB value of the control solution at this time, and click white balance calibration on the host computer;

[0023] Step 6: Apply 70% to 76% ethanol solution on the hypothenar eminence of the hand for 1 minute, then cover the hypothenar eminence over the test well and control well for 10 minutes;

[0024] Step 7: Follow the steps in step 4 to transmit the fluorescence image of the test liquid collected by the camera to the host computer in real time. From the 8th minute to the 10th minute, save the image every 30 seconds. Extract the RGB values ​​of the 5 test liquid images and convert them to HSV values. Click "Calculate Results" in the host computer to get the skin cholesterol measurement value at this time. Click "HSV Curve" to get the change curve of the HSV value within 2 minutes. Calculate the average of the last 5 measurement data as the final skin cholesterol concentration value.

[0025] The CD-β-AuNCs solution involved in the present invention is preferably prepared according to the method described in the invention patent application with publication number CN113801650A.

[0026] The preparation method of the detection solution involved in the present invention is as follows: add 2.5g of sodium alginate solid to 50ml of pure water, stir at room temperature for 1h, and then place in a refrigerator for 30min at low temperature to form a colorless sodium alginate solution; take 4ml of the prepared CD-β-AuNCs solution, add 1ml of Triton X-100 solution, mix and stir for 5min to form a mixed solution. Take 2ml of the prepared CD-β-AuNCs mixed solution, add 3ml of a colorless sodium alginate solution with a concentration of 90.8mmol / L, mix and stir for 30min, and then place in a refrigerator for 30min at low temperature to form a detection solution.

[0027] In the detection system, the concentration of CD-β-AuNCs in the detection system is preferably 1.9×10 -5 ~5.8×10 -5 mol / L (calculated as Au). If the concentration is too high, it can be diluted with pH=6.4 PBS buffer as a solvent or dispersant. In this step, the reaction time of CD-β-AuNCs and cholesterol is usually greater than or equal to 8 min, preferably 8 min.

[0028] Compared with the prior art, the measuring device of the present invention uses an OV5640 camera sensor to detect the fluorescent color of the test liquid after the reaction of CD-β-AuNCs and skin cholesterol, obtains the values ​​of the three primary colors of RGB, and then converts them into the HSV color space through microprocessor processing. The applicant found that the hue value in the HSV color space is negatively correlated with the corresponding cholesterol concentration and has a good logarithmic linear relationship, so it is possible to construct a cholesterol concentration-hue value standard curve, and then obtain the hue value of the test liquid after the reaction of skin cholesterol and CD-β-AuNCs, and then combine the above standard curve to obtain the cholesterol concentration of the skin. The measuring device of the present invention is simple and easy to operate, with high sensitivity and high accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1A This is the front side of the structure diagram of the acquisition module of an embodiment of the non-invasive skin cholesterol detection device based on deep learning to identify fluorescent color according to the present invention.

[0030] Figure 1B This is the back side of the structure diagram of the acquisition module of an embodiment of the non-invasive skin cholesterol detection device based on deep learning to identify fluorescent color according to the present invention.

[0031] Figure 2 This is a structural diagram of the main control module of the non-invasive skin cholesterol detection device based on deep learning to identify fluorescent color according to the present invention.

[0032] Figure 3 This is the host computer main interface of the non-invasive skin cholesterol detection device for recognizing fluorescent colors based on deep learning as described in the present invention.

[0033] Figure 4 This is an assembly diagram of the acquisition module of an embodiment of the non-invasive skin cholesterol detection device based on deep learning to identify fluorescent color according to the present invention.

[0034] Figure 5 This is an assembly diagram of the main control module of the non-invasive skin cholesterol detection device for identifying fluorescent colors based on deep learning according to the present invention.

[0035] Figure 6 The main control module of the present invention Figure 2Enlarged diagram of the main control button in the middle.

[0036] Figure 7 The acquisition module of the present invention Figure 1A A top view of the top cover plate 1 of the acquisition module.

[0037] Figure 8 The correlation curve between the concentration of pig skin cholesterol and the corresponding hue value H in the HSV color space

[0038] Fig. 9 It is a schematic diagram of the integrated circuit of the present invention.

[0039] Fig.10 It is a schematic diagram of the working process of the present invention.

[0040] The numbers in the figure are:

[0041] 1 Cover plate, 2 Slide slot, 3 Main body, 4 Excitation light source, 5 Bottom cover, 6 Detection hole, 7 Control hole, 8 Darkroom, 9 Excitation light source, 10 Quartz glass, 11 Quartz glass slot, 12 Collection area, 13 Excitation light source, 14 Filter, 15 Filter slot, 16 Camera, 17 Shooting platform, 18 Display screen, 19 Main control module top cover, 20 Main control circuit, 21 Main control module bottom cover, A Main control button, 22 Main interface button, 23 LED control button, 24 USB data transmission button, 25 Camera interface display button. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] The skin cholesterol non-invasive detection device based on deep learning to identify fluorescent color of the present invention comprises a collection module ( Figure 1A ), main control module ( Figure 2 ) and the host computer ( Figure 3), the detection process is carried out in a dark room, and the data obtained by the detection is transmitted to the host computer for processing through the main control module, and the results are displayed on the host computer. Specifically, the detection obtains the fluorescent image of the detection liquid in the detection hole 6. The main control module is used to control the light intensity of the excitation light source, the control of the camera, the collection and transmission of the fluorescent image signal, and the LCD displays the main control information in real time, including the main control interface, the light intensity of the excitation light source, and the camera transmission image target display. The host computer is used to receive the collected fluorescent image, locate the fluorescent position through the target recognition model of the host computer, and extract the RGB value of the fluorescent image and convert it into an HSV value. According to actual needs, the change curve of the HSV value with the reaction time of cholesterol to the fluorescent probe is drawn. And the skin cholesterol concentration value is obtained based on the stable hue value.

[0044] The assembly method of the acquisition module of the present invention is as follows Figure 4 , the quartz glass 10 and the filter 14 are inserted into the corresponding quartz glass groove 11 and the filter groove 15 respectively. The slide groove 2 is spliced ​​on the upper part of the main body 3, wherein the collection area 12 of the slide groove 2 faces the darkroom 8. The camera 16 is installed on the shooting platform 17, and the bottom cover 5 is closed to complete the assembly. The main control module top cover 19 and the main control module bottom cover 20 fix the main control circuit 20 in the center to form a main control box such as Figure 5 . During the detection, the cover plate 1 is installed on the slide 2, the detection hole 6 of the cover plate 1 is filled with the detection liquid, and the control hole 7 is filled with the control liquid. At this time, the palm of the hand covers the detection hole 6 and the control hole 7. At this time, the main body 3 is completely closed to form a light-shielding darkroom 8. Adjust the main control module, turn on the excitation light source 4 and the excitation light source 13, and adjust the light intensity to the maximum value. At this time, the excitation light source is 45° obliquely illuminated on the collection area 12 and illuminates the cover plate 1 through the quartz glass 10. Turn on the camera 16, adjust the cover plate 1, and keep the excitation light irradiated on the position of the detection hole 6 or the control hole 7. At this time, the detection liquid produces fluorescence under the irradiation of the excitation light source. Click the USB data transmission button 24 to transmit the fluorescence image signals of the detection liquid and the control liquid collected by the camera sensor in real time to the host computer through the USB to serial port module.

[0045] The detection optical path of the excitation light source of the present invention is an oblique optical path, that is, the excitation light source 4 and the excitation light source 13 are respectively irradiated at an angle of 45° to the collection area 12, forming two circular excitation areas. The camera sensor is arranged directly below the collection area 12, and the image information of the excitation area is vertically collected after filtering out the interference of the excitation light through a 400nm to 450nm high-pass filter 14.

[0046] Furthermore, the skin cholesterol non-invasive detection device based on deep learning to identify fluorescent color of the present invention needs to be calibrated in advance for white balance due to the different skin colors of different people. That is, the RGB image of the control liquid when the palm covers the control hole 7 is extracted during the first detection and transmitted to the host computer for white balance calibration.

[0047] The main control module chip in the technical solution of the present invention is STM32F407ZGT6, which is an existing conventional choice that can realize data processing. The display screen is an LCD touch display screen.

[0048] In a specific embodiment, the excitation light source 4 and the excitation light source 13 are ultraviolet lamps, and further use ultraviolet lamps with a wavelength of 365nm or 370nm.

[0049] In a specific embodiment, the camera 14 preferably adopts OV5640.

[0050] The present invention also provides a method for measuring skin cholesterol concentration using the above device, comprising the following steps:

[0051] 1) Place the control solution in the control well 7, place the detection solution in the detection well 6, place the hypothenar of the palm above the detection well 6 and the control well 7, and turn on the excitation light source 4 and the excitation light source 13.

[0052] 2) Click the camera interface display button 25 to turn on the camera 16, transmit the real-time image of the camera to the display screen 18, and adjust the position of the cover plate so that the detection hole 6 and the control hole 7 are respectively located in the center of the two excitation light sources;

[0053] 3) Open the host computer, select the serial port and set the baud rate to 1500000, open the serial port transmission, and then click the USB transmission button 24 to receive the fluorescence image signals of the control solution and the test solution;

[0054] 4) Based on the deep learning training model, locate and identify the control solution and the test solution, extract the RGB value of the control solution at this time, and click "Calibrate" on the host computer to complete the white balance calibration;

[0055] 5) Remove the hand and apply 70% to 76% ethanol solution on the hypothenar eminence of the hand for 1 minute, then cover the hypothenar eminence over the detection well 6 and the control well 7 for 10 minutes;

[0056] 6) According to step 4), the fluorescent image of the fluorescent probe detection liquid collected by the camera (16) is transmitted to the host computer in real time. From the 8th minute to the 10th minute, the image is saved every 30 seconds. The RGB values ​​of the fluorescent probe images are extracted 5 times and converted to HSV values. Click "Calculate Results" in the host computer to obtain the skin cholesterol measurement value at this time. Click "HSV Curve" to obtain the change curve of the HSV value within 2 minutes. The average value of the last 5 measurement data is calculated as the final skin cholesterol concentration value.

[0057] The CD-β-AuNCs solution involved in the present invention is preferably prepared according to the method described in the invention patent application with publication number CN113801650A. Specifically, as follows:

[0058] Preparation of CD-β-AuNCs solution: The reduction method was used for synthesis. The specific synthesis steps were as follows: All glassware was washed clean with freshly prepared aqua regia. The water bath temperature was adjusted to 90 °C. 5.0 mL of HAuCl 4 ·3H 2 O aqueous solution (1.0 mM, 90°C) was vigorously stirred for 10 min, and 2.0 mL of mono(6-mercapto-6-deoxy)-β-cyclodextrin solution (20.0 mM, 90°C) was added. After stirring for 5 min, 1.5 mL of sodium hydroxide aqueous solution (1.0 mol·L -1 ), and then the bottle mouth was sealed with aluminum film. After reacting for 4 hours, the obtained solution was a yellow clear liquid. Under ultraviolet light (365nm), strong green fluorescence could be observed, indicating that a green CD-β-AuNCs solution was obtained. The concentration of CD-β-AuNCs in the solution was 5.8×10 -4 mol / L. The concentration of CD-β-AuNCs is preferably 1.9×10 -5 ~5.8×10 -5 mol / L (calculated as Au). If the concentration is too high, PBS buffer with pH = 6.4 can be used as a solvent or dispersant for dilution.

[0059] The preparation methods of the control solution and the detection solution of the present invention are as follows:

[0060] Add 0.7g sodium alginate solid to 50ml pure water, stir at room temperature for 1h, and place in a refrigerator for 30min to form a colorless sodium alginate control solution; take 4ml of the prepared CD-β-AuNCs solution and add 1ml of Triton X-100 solution and mix and stir for 5min to form a mixed solution. Take 2ml of the prepared CD-β-AuNCs mixed solution and add 3ml of 90.8mmol / L colorless sodium alginate solution and mix and stir for 30min, then place in a refrigerator for 30min to form a light yellow CD-β-AuNCs sodium alginate detection solution.

[0061] In the method of the present invention, the fluorescence intensity of the reaction time between skin cholesterol and CD-β-AuNCs remains substantially unchanged after 8 minutes. Therefore, the hypothenar eminence of the hand usually needs to be covered on the cover plate for more than or equal to 8 minutes, preferably 10 minutes.

[0062] In the method of the present invention, in order to establish a quantitative relationship between the enhancement of CD-AuNCs and the cholesterol concentration of the skin, 16 cholesterol solutions of different concentrations (0, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.9, 1.9 mmol / L) were prepared in the concentration range of 0-1.9 mmol / L, and pig skin was taken as the experimental test sample. The ambient temperature was 23-26 ° C, the pork skin was heated to the temperature of 36-37 ° C when the human skin was in a normal state, 0.1 mL of the above concentration of cholesterol was applied to the pork skin, and then covered on the cover plate filled with the detection solution and the control solution, and the reaction time was 8 minutes. From the 8th minute to the end of the 10th minute, the fluorescence emission spectrum and fluorescence intensity (fluorescence color) were measured every 30 seconds, and converted into the HSV color space. The parameter values ​​of the fluorescence color and the HSV color space are shown in Table 1.

[0063] Table 1:

[0064]

[0065] The experimental results show:

[0066] (1) The addition of cholesterol can enhance the fluorescence intensity of the detection solution, and as the cholesterol concentration increases, the fluorescence intensity of the detection solution increases.

[0067] (2) The addition of cholesterol can reduce the hue value (H) of the test solution in the HSV color space, and as the cholesterol concentration increases, the hue value of the test solution decreases.

[0068] The concentration of cholesterol in the solution added in Table 1 and the corresponding hue value H in the HSV color space are used to draw a correlation curve, and the correlation coefficient between the two is 0.96. It can be seen that the device of the present invention has good accuracy and stability, the test results can be displayed in real time, and rapid quantitative detection of skin cholesterol can be achieved.

Claims

1. A non-invasive skin cholesterol detection device based on deep learning to identify fluorescent colors, characterized by: It includes an acquisition module, a main control module and a host computer. The acquisition module consists of a cover plate, a slide, a main body and a bottom cover. The cover plate is provided with a detection hole and a control hole; a camera is provided on the shooting platform at the bottom of the main body, and the lens is fixed to the shooting hole to vertically collect fluorescent signals. The main body is spliced ​​with the slide to form a darkroom inside. The main body is also provided with a filter and an excitation light source. The slide is provided with quartz glass. The excitation light source is irradiated at a 45° angle on the collection area of ​​the slide, and then passes through the quartz glass to irradiate the detection hole and the control hole of the cover plate; The main control module includes a top cover, a main control circuit board and a bottom cover. The main control information is displayed on a display screen. The host computer is used to receive the fluorescent image signal collected by the main control module and analyze and process it.

2. According to claim 1, a non-invasive skin cholesterol detection device based on deep learning to identify fluorescent colors is characterized in that The main control module chip uses STM32F407ZGT6 and is programmed using Keil uVision5.

3. According to claim 1, a non-invasive skin cholesterol detection device based on deep learning to identify fluorescent color is characterized in that The host computer (D) applies Yolov8n, a target detection algorithm based on deep learning, and trains a target recognition model according to the data set collected in this experiment, and uses PyCharm to write the host computer program.

4. According to claim 1, a non-invasive skin cholesterol detection device based on deep learning to identify fluorescent color is characterized in that The host computer and the main control module use a USB serial port communication mode.

5. According to claim 1, a non-invasive skin cholesterol detection device based on deep learning to identify fluorescent color is characterized in that The excitation light source is two ultraviolet LED lamps with the same wavelength of 365nm or 370nm, which are respectively located at the front and back sides of the main body.

6. According to claim 1, a non-invasive skin cholesterol detection device based on deep learning to identify fluorescent color is characterized in that The filter is a 400nm-450nm high-pass filter to filter out 365nm violet light interference and leave blue-green fluorescence around 450nm.

7. According to claim 1, a non-invasive skin cholesterol detection device based on deep learning to identify fluorescent color is characterized in that The camera is an OV5640 camera, which is controlled by the main control module. The control interface is displayed on the LCD display screen. When the host computer transmits, the collected image is displayed in real time on the host computer.

8. A non-invasive skin cholesterol detection device based on deep learning to identify fluorescent colors, characterized in that: The fluorescent probe for detecting cholesterol, thiol-β-cyclodextrin gold nanocluster (CD-β-AuNCs) solution, is preferably prepared according to the method described in the invention patent application with publication number CN113801650A. Its concentration is preferably configured according to the method described in the invention patent application with publication number CN114646600A. The preparation method of CD-β-AuNCs sodium alginate detection solution is as follows: add 2.5g sodium alginate solid to 50ml pure water, stir at room temperature for 1h, and then place in a refrigerator for 30min to form a colorless sodium alginate control solution (referred to as control solution); take 4ml of the prepared CD-β-AuNCs solution, add 1ml of Triton X-100 solution, mix and stir for 5min to form a mixed solution. Take 2 ml of the prepared CD-β-AuNCs mixed solution, add 3 ml of a colorless sodium alginate solution with a concentration of 90.8 mmol / L, mix and stir for 30 minutes, and then place it in a refrigerator at low temperature for 30 minutes to form a light yellow CD-β-AuNCs sodium alginate detection solution (referred to as detection solution).

9. A non-invasive skin cholesterol detection device based on deep learning to identify fluorescent colors, characterized in that: The following steps are involved: Step 1: Insert the quartz glass into the glass slot, then place the cover plate on the upper part of the slide slot and turn on the two excitation light sources. Use the main control module to adjust the LED light intensity until the LCD display shows a value of 300. Step 2: Click the camera interface display button to turn on the camera, transmit the real-time image of the camera to the LCD display, and adjust the position of the cover plate so that the detection hole and the control hole are located in the center of the two excitation light sources respectively; Step 3: Open the host computer, select the serial port and set the baud rate to 1500000, and open the serial port transmission; Step 4: Place the control solution in the control well and the test solution in the test well. Then, place the hypothenar of the hand over the two wells. Click the USB transfer button on the main control module to start USB data transmission and transfer the real-time image collected by the camera to the host computer. Step 5: Locate the control well based on the deep learning training model, extract the RGB value of the colorless sodium alginate solution at this time, and click white balance calibration on the host computer; Step 6: Apply 70% to 76% ethanol solution on the hypothenar eminence of the hand for 1 minute, then cover the hypothenar eminence over the test well and control well for 10 minutes; Step 7: Follow the steps in step 4 to transmit the test liquid image collected by the camera to the host computer in real time. From the 8th minute to the 10th minute, save the image every 30 seconds. Extract the RGB values ​​of the test liquid images 5 times and convert them to HSV values. Click "Calculate Results" in the host computer to get the skin cholesterol measurement value at this time. Click "HSV Curve" to get the change curve of the HSV value within 2 minutes. Calculate the average of the last 5 measurement data as the final skin cholesterol concentration value.

10. A non-invasive skin cholesterol detection device based on deep learning to identify fluorescent colors, characterized in that: In order to establish the quantitative relationship between the enhancement of CD-AuNCs and the cholesterol concentration of the skin, 16 cholesterol solutions of different concentrations (0, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.9, 1.9 mmol / L) were prepared in the concentration range of 0-1.9 mmol / L, and pig skin was taken as the experimental test sample. The ambient temperature was 23-26 °C, and the pork skin was heated to the normal temperature of human skin at 36-37 °C. 0.1 mL of the above concentration of cholesterol was applied to the pork skin, and then covered on the cover plate filled with the test solution and the control solution. The reaction time was 8 minutes, and then the fluorescence emission spectrum and fluorescence intensity (fluorescence color) were measured, and converted into HSV color space. The values ​​of the parameters of fluorescence color and HSV color space are shown in Table 1. Table 1:

11. A non-invasive skin cholesterol detection device based on deep learning to identify fluorescent colors, characterized in that: The correlation curve is drawn with the concentration of cholesterol in the solution added in Table 1 and the corresponding hue value H in the HSV color space, and the linear equation is H=-25.71ln(C)+304.22, where H is the hue value of the fluorescent probe, C is the cholesterol concentration, and the correlation coefficient is 0.

96. It can be seen that the device of the present invention has good accuracy and stability, the test results can be displayed in real time, and rapid quantitative detection of skin cholesterol can be achieved.

Images