A cross-platform detection device for detecting fluorescent immunoassays and colloidal gold reagent strips

Abstract

The present invention provides a cross-platform detection device. Specifically, the detection device provided by the present invention includes: (a) a base (2); (b) a structure for fixing an object to be detected (3); (c) a photoelectric conversion module; (d) a light source module, wherein the light source module includes: (d1) a light source; the light source is a symmetrically arranged strip surface light source; and the light source can emit blue light or green light respectively. The cross-platform detection device of the present invention can be used for detection in a fluorescence mode and a colloidal gold mode, and the cross-platform detection device of the present invention has good stability, good repeatability, high resolution, good linearity, low cost, easy assembly, and is suitable for mass production.

Description

Technical Field

[0001] The invention belongs to the field of medical devices, and in particular relates to a cross-platform detection device for detecting fluorescent immunoassays and colloidal gold reagent strips. Background Art

[0002] At present, most IVD dry immunoassay instruments on the market are only suitable for single methodology reagent tablets, such as colloidal gold detection or fluorescence detection. However, due to different market demands and the iteration of new products, there is a demand for compatible platforms in the market, and cost factors should also be considered. In addition, in mass industrial production, due to the limitation of human and material resources, it is impossible to calibrate each module with reagent tablets separately. Therefore, in order to ensure the consistency of the final test results of different detection modules or devices, each batch of detection modules or devices must be calibrated to output consistent results. However, considering that there are deviations between different components (cameras, light sources, filters, lenses), after the modules are assembled, there are certain deviations between each batch of modules. If calibrated as a whole, there are often many problems, and it is difficult to take into account every detail (linearity, high and low value deviations), and the consistency results are often unsatisfactory.

[0003] In summary, the art needs a detection device for a dry immunolateral flow reagent strip that can perform detection based on both a colloidal gold method and a fluorescent method, and has accurate detection results and high sensitivity. Summary of the invention

[0004] The purpose of the present invention is to provide a fluorescence and colloidal gold cross-platform quantitative detection device with wide linearity, high precision and good output consistency.

[0005] In a first aspect of the present invention, a cross-platform detection device is provided, wherein the detection device comprises:

[0006] (a) a base (2);

[0007] (b) a structure for fixing the object to be tested (3), the structure for fixing the object to be tested being located on the base (2); and the structure for fixing the object to be tested being provided with a detection window (the detection window being used to expose the area of ​​the object to be tested (or referred to as the area to be tested) to the light emitted by the light source and the camera lens);

[0008] (c) a photoelectric conversion module, the photoelectric conversion module is used to convert the light signal emitted or reflected by the object to be tested into an electrical signal and obtain an image of the area to be tested containing the object to be tested; wherein the photoelectric conversion module comprises: a camera (preferably, one camera); and the camera is configured such that: the objective lens of the camera is located above the detection window;

[0009] (d) A light source module, comprising: (d1) a light source; the light source is a symmetrically arranged strip surface light source (51) (preferably, a pair (2) or more pairs of symmetrically arranged strip surface light sources); and the light source can emit blue light or green light respectively.

[0010] In another preferred embodiment, the detection device further comprises: (e) an outer box (1), wherein the outer box is used to form a darkroom required for detection.

[0011] In another preferred embodiment, the symmetry center of the symmetrical arrangement refers to the fixed structure; preferably, it refers to the AB center line of the fixed structure.

[0012] In another preferred example, the light source module includes a blue light channel and a green light channel, and emits blue light or green light respectively by switching the blue light channel and the green light channel.

[0013] In another preferred embodiment, the blue light refers to light centered at a wavelength of 450±10 nm and / or having a half-peak width of 20 to 40 nm.

[0014] In another preferred embodiment, the spectrum of the blue light is substantially as follows Figure 3A shown.

[0015] In another preferred embodiment, the green light refers to light centered at a wavelength of 525±10 nm and / or having a half-peak width of 20 to 40 nm.

[0016] In another preferred embodiment, the spectrum of the green light is substantially as follows Figure 3B shown.

[0017] In another preferred example, the light source is an LED light source; preferably, the light source is an LED array light source.

[0018] In another preferred example, the LED array light source refers to a light source comprising blue light beads and green light beads arranged in an alternating manner.

[0019] In another preferred example, the strip-shaped surface light source includes blue light beads and green light beads arranged in an alternating manner.

[0020] In another preferred embodiment, the blue light beads and the green light beads are arranged in a staggered array at a certain interval (preferably, the interval is 3 mm to 5 mm).

[0021] In another preferred embodiment, the blue light lamp bead is capable of emitting light with a wavelength of 450±10nm as the center and / or a half-peak width of 30±10nm (preferably, with a wavelength of 450±5nm as the center and / or a half-peak width of 30±1nm; more preferably, Figure 3A As shown) blue light lamp beads (preferably, LED lamp beads).

[0022] In another preferred embodiment, the green light lamp bead is capable of emitting light with a wavelength of 525±10nm as the center and / or a half-peak width of 30±10nm (preferably, the spectrum of the green light lamp bead is centered at 525±5nm and / or a half-peak width of 30±1nm; more preferably, Figure 3B Green light lamp beads (preferably, LED lamp beads) are provided.

[0023] In another preferred embodiment, the deviation of the luminous intensity mcd value of the blue light lamp bead or the green light lamp bead is ≤5%.

[0024] In another preferred embodiment, the light source module further comprises: (d2) a diffuse reflection plate (52); preferably, a diffuse reflection plate with a thickness of 1 to 3 mm; more preferably, a diffuse reflection plate with a thickness of 2±0.5 mm.

[0025] In another preferred embodiment, the surface of the strip-shaped surface light source forms an angle of 45±1 degrees with the plane where the detection window is located (equivalent to the surface of the reagent strip to be detected).

[0026] In another preferred example, the width of the surface of the strip-shaped surface light source is greater than or equal to the width of the detection window.

[0027] In another preferred embodiment, the detection window is substantially square.

[0028] In another preferred embodiment, the object to be tested includes: a reagent strip, a test piece, a fluorescence standard piece, and / or a grayscale piece.

[0029] In another preferred embodiment, the object-to-be-detected fixing structure is a groove, and the upper opening of the groove is the detection window.

[0030] In another preferred example, the width of the surface of the strip-shaped surface light source is greater than or equal to the width of the groove.

[0031] In another preferred embodiment, the camera is arranged through a camera fixing structure such that the objective lens of the camera is located above the detection window.

[0032] In another preferred example, the camera fixing structure comprises: a camera fixing plate (41) and a support column (42); preferably, the camera fixing plate is provided with an opening for fixing the camera and its four corners are provided with hole structures for connecting with the support column.

[0033] In another preferred example, the camera is a CMOS camera.

[0034] In another preferred embodiment, the CMOS camera unit is a high-sensitivity CMOS black-and-white camera unit with 12 bits A / D output.

[0035] In another preferred example, the photoelectric conversion module further includes: a lens assembly.

[0036] In another preferred embodiment, the lens assembly includes: a 2-4 times microscope magnification lens; preferably, a 3 times microscope magnification lens.

[0037] In another preferred example, the photoelectric conversion module further includes: a filter component; preferably, the filter component includes a green filter.

[0038] In another preferred example, the light source module further includes: (d3) a driving submodule, wherein the driving submodule is used to supply constant current to the light source.

[0039] In another preferred example, the driving submodule includes an LED driving circuit.

[0040] In another preferred embodiment, the driving submodule is further provided with a temperature compensation circuit; the temperature compensation circuit is used to keep the luminous intensity of the light source consistent at different temperatures (within the temperature range of 25° C.-40° C.).

[0041] In another preferred embodiment, the temperature compensation circuit collects temperature through a sensing network distributed inside the light source (preferably an LED light source).

[0042] In another preferred example, the driving submodule is further configured to have one or more of the following functions: PWM regulation, current regulation, and temperature acquisition.

[0043] In another preferred embodiment, the temperature is collected through a sensing network distributed inside the light source (preferably an LED light source).

[0044] In another preferred example, the sensing accuracy of the temperature acquisition unit is ±0.25°C.

[0045] In another preferred example, the driving submodule is configured to be able to perform 12-bit grayscale PWM adjustment on the blue light channel or the green light channel.

[0046] In another preferred example, the driving submodule is configured to be able to perform 6-point correction current adjustment for the blue light channel or the green light channel.

[0047] In another preferred example, the light source module further comprises: (d4) a light source fixing frame (53), the light source fixing frame is used to fix the light source, and the position of the fixing frame can be adjusted along the AB direction.

[0048] In another preferred embodiment, the light source module does not include a component for filtering light (eg, a filter).

[0049] In another preferred embodiment, the detection device further includes: (f) a control module, wherein the control module is configured to execute switching between a colloidal gold mode and a fluorescence mode.

[0050] In another preferred embodiment, the switching content includes: switching between a blue light channel and a green light channel in the light source module.

[0051] In another preferred embodiment, the switching content is the switching of the blue light channel and the green light channel in the light source module and the switching of the camera parameters.

[0052] In another preferred embodiment, the detection device further includes: (g) an image processing and analysis module; wherein the image processing and analysis module is used to obtain a detection result based on the image obtained by the photoelectric conversion module.

[0053] In another preferred example, the image processing and analysis module is configured to: perform preprocessing on the image obtained by the photoelectric conversion module; wherein the preprocessing includes denoising the image.

[0054] In another preferred embodiment, the denoising process is performed based on wavelet transform soft threshold (ie, the denoising process is wavelet transform soft threshold denoising).

[0055] In another preferred embodiment, the image processing and analysis module is configured to perform the following steps to pre-process the image obtained by the photoelectric conversion module:

[0056] (i) Wavelet decomposition of the two-dimensional signal of the image: select the wavelet basis function (or simply wavelet basis) and the level N of wavelet decomposition, and then calculate the decomposition of the two-dimensional signal s to the Nth level;

[0057] (ii) Threshold quantization of high-frequency coefficients: For each layer from 1 to N, a threshold is selected and a soft threshold quantization is performed on the high-frequency coefficients of this layer, where N is 1, 2, 3, 4, or 5; and

[0058] (iii) Two-dimensional wavelet reconstruction: Based on the low-frequency coefficients of the Nth layer of wavelet decomposition and the modified high-frequency coefficients from the first layer to the second layer, the wavelet reconstruction of the two-dimensional signal is calculated to obtain the image signal reconstructed using the two-dimensional wavelet.

[0059] In another preferred embodiment, the wavelet basis function used in the denoising process (ie, wavelet denoising process) is coif; preferably, coif3.

[0060] In another preferred embodiment, the soft threshold used in the denoising process is calculated by Formula 1:

[0061] Threshold=sigma*sqrt(2ln(N D1 )) (Formula 1);

[0062] In the formula, Threshold represents the soft threshold, sigma represents the variance corresponding to the wavelet coefficient of the noise, and ND1 Represents the number of wavelet coefficients decomposed in the first layer;

[0063] Here, sigma=median(abs(wavelet coefficient of D1 layer)) / 0.6745.

[0064] In another preferred embodiment, the decomposition level N of the denoising process (ie, wavelet denoising process) is 2.

[0065] In another preferred example, the detection result includes: process result and / or target result.

[0066] In another preferred embodiment, the process results include: peak area, peak value, peak area ratio (C / T peak area ratio), peak value ratio (C / T peak value ratio), or a combination thereof; preferably, include: peak area ratio, peak value ratio, or a combination thereof.

[0067] In another preferred embodiment, the target result includes: concentration.

[0068] In another preferred embodiment, the image processing and analysis module is configured to perform the following steps:

[0069] (1) preprocessing the image obtained by the electrical conversion module; wherein the preprocessing includes denoising the image;

[0070] (2) optionally adjusting the image; wherein the adjustment includes performing rotation correction processing on the image so that the edge of the object to be measured or the area to be measured is parallel to the edge of the field of view (equivalent to being parallel to the edge of the acquired image);

[0071] (3) identifying and extracting the image processed by step (1) and optionally step (2) to obtain signal data; and

[0072] (4) Obtaining a detection result based on the signal data.

[0073] In another preferred embodiment, in step (1), the denoising process is performed based on wavelet transform soft threshold.

[0074] In another preferred embodiment, the image processing and analysis module is further configured to perform correction of the reaction signal value, thereby obtaining a signal true value estimation of the reaction signal, and the detection result is obtained based on the reaction signal true value estimation;

[0075] Wherein, the reaction signal is a peak area, a peak value, a ratio of peak areas, a ratio of peak values, or a combination thereof.

[0076] In another preferred embodiment, the image processing and analysis module calculates the signal true value (the signal true value is the signal value after removing background interference) through formula 2; wherein the calculated signal true value is the signal true value estimate in the formula;

[0077] Reaction signal value = signal true value estimate + b (signal true value estimate) * background signal value + e (Formula 2)

[0078] In the formula, b (signal true value estimation) represents the slope of the signal true value as the base changes, and e is the systematic random error; where Base refers to the background signal value (or base signal value).

[0079] In another preferred embodiment, the b (signal true value estimation) function is obtained by the following method: (i) processing known data; and (ii) fitting the processed data by a fitting model to obtain the b (signal true value estimation) function.

[0080] In another preferred example, the known data includes: a known signal true value and a slope b (known signal true value) corresponding to the known signal true value.

[0081] In another preferred embodiment, the known signal true value is obtained based on a known detection result (preferably, a known concentration).

[0082] In another preferred embodiment, the slope b (known signal true value) is obtained based on the reaction signal value (i.e., the reaction signal value without correction of the reaction signal value) and the background signal value (preferably, the reaction signal value and the background signal are obtained by the device as described in the first aspect).

[0083] In another preferred embodiment, the uncorrected reaction signal value and the background signal are calculated by formula 8 to obtain b (known signal true value);

[0084] Uncorrected reaction signal value = known signal true value + b (known signal true value) * background signal value + e (Formula 8)

[0085] In the formula, b (known signal true value) represents the slope of the known signal true value as the base changes, and e is the system random error; where base refers to the background signal value (or base signal value).

[0086] In another preferred embodiment, the existing detection data is processed by the following steps: obtaining a scatter plot based on known data (wherein a point in the scatter plot represents a pair of corresponding signal true values ​​and a slope b corresponding to the signal true value), and optionally removing abnormal data in the scatter plot.

[0087] In another preferred embodiment, the fitting model includes one or more of the following models:

[0088] y = a*x^b+c (Model 1);

[0089] y = (p1*x+p2) / (x+q1) (Model 2);

[0090] y = a*x^2 + b*x + c (Model 3); and / or

[0091] y=a*x+b(Model 4);

[0092] In this format, a, b, c, p1, and p2 are constants to be determined by fitting.

[0093] In another preferred embodiment, the b (signal true value estimation) function includes: Formula 3, Formula 4, Formula 5, and / or Formula 6;

[0094] b(That)=-0.001778*That^0.7807+1.118 (Formula 3)

[0095] b(That)=(-3.048*That+11780) / (That+10910) (Formula 4)

[0096] b(That)=0.00000002006*That^2-0.0003549*That+1.073 (Formula 5)

[0097] b(That)=-0.0002758*That+1.036 (Formula 6);

[0098] Among them, That refers to the true value estimation of the signal.

[0099] In another preferred example, the detection device further includes: (h) an output module, wherein the output module is used to output the detection result.

[0100] In another preferred embodiment, the outer box is a metal outer box; preferably, an aluminum metal outer box.

[0101] In another preferred embodiment, the outer box is a shielding outer box of 6061 aluminum alloy, which is used to create a darkroom environment to avoid the influence of external natural light, and at the same time has good anti-electromagnetic interference performance and excellent heat dissipation performance.

[0102] In another preferred embodiment, the outer box is further provided with a wiring hole (11).

[0103] In another preferred embodiment, the diameter of the wiring hole is 10-20 mm; preferably, 15 mm.

[0104] In another preferred embodiment, the light source is arranged at a height H L(measured at the center of the strip-shaped surface light source) is 0.1-0.3H (preferably, 0.15-0.20H); wherein H refers to the setting height of the objective lens of the camera of the photoelectric conversion module or the lens of the auxiliary lens assembly.

[0105] In another preferred embodiment, H=150-210 mm; preferably, H=185±2 mm.

[0106] In another preferred embodiment, H L =25~40mm; preferably, H L =35±1mm.

[0107] In another preferred embodiment, the distance between the strip-shaped surface light sources is 0.3-0.5H (preferably 0.4H), and the distance is measured from the center of the surface of the strip-shaped surface light source.

[0108] In another preferred embodiment, the distance between the strip-shaped surface light sources is 50-100 mm; preferably, 75±5 mm.

[0109] In another preferred embodiment, the detection device is suitable for detecting an analyte whose TAP is within the following range: TAP≥900, preferably, TAP≥1500; more preferably, TAP≥2000 and TAP≤4000.

[0110] In a second aspect of the present invention, a detection method is provided, comprising the steps of:

[0111] (1) Providing the detection device according to the first aspect of the present invention;

[0112] (2) Placing the object to be tested on the object fixing structure;

[0113] (3) Select colloidal gold mode or fluorescence mode;

[0114] (4) a photoelectric conversion module to obtain an image in a colloidal gold mode or a fluorescence mode; and

[0115] (5) Obtain detection results based on the obtained image.

[0116] In another preferred embodiment, in the method, the same analyte is detected in colloidal gold mode and / or fluorescence mode.

[0117] In another preferred embodiment, the method is non-diagnostic and non-therapeutic.

[0118] In another preferred embodiment, the method is an in vitro method or an ex vivo method.

[0119] In another preferred embodiment, the object to be tested includes: a reagent strip, a test piece, a fluorescence standard piece, and / or a grayscale piece.

[0120] It should be understood that within the scope of the present invention, the above-mentioned technical features of the present invention and the technical features specifically described below (such as embodiments) can be combined with each other to form a new or preferred technical solution. Due to space limitations, they will not be described one by one here. BRIEF DESCRIPTION OF THE DRAWINGS

[0121] Figure 1 A three-dimensional schematic diagram of the detection device of the present invention is shown (the outer box is not shown).

[0122] Figure 2 A front schematic diagram of the detection device of the present invention is shown.

[0123] The symbols in the figure are as follows: 1 is the outer box, 11 is the wiring hole, 2 is the base, 3 is the fixed structure, 4 is the camera, 41 is the camera fixing plate, 42 is the support column, 51 is the strip light source, 52 is the diffusion plate, and 53 is the light source fixing frame

[0124] Figure 3A The spectrum of the blue light (or blue light lamp beads) used in the detection device of the present invention is shown.

[0125] Figure 3B The spectrum of the green light (or green light lamp bead) used in the detection device of the present invention is shown.

[0126] Figure 4A A flow chart showing image acquisition of the detection device of the present invention.

[0127] Figure 4B A flow chart showing the image algorithm processing and image analysis of the detection device of the present invention.

[0128] Figure 5 Displays the parameters of the grayscale standard film of the Image Society of Japan.

[0129] Figure 6 It shows the image of the horizontal row average of the grayscale card under the use of a ring light source, and the brightness of the middle area is relatively low.

[0130] Fig. 7A A schematic diagram of a PTC test system for calibrating a camera of the detection device of the present invention is shown.

[0131] Figure 7B A consistency calibration flow chart of the detection device of the present invention is shown.

[0132] Figure 7C (i, ii and iii) respectively show the light response curves of different cameras used in the detection device of the present invention under the PTC detection platform.

[0133] Fig.7DThe light response curve of the camera used in the detection device of the present invention under the PTC detection platform is shown.

[0134] Figure 8 The linear test results of the detection device of the present invention in the colloidal gold mode are shown. DETAILED DESCRIPTION

[0135] After extensive and in-depth research, the inventors have developed for the first time a cross-platform detection device that is compatible with colloidal gold detection and fluorescence detection. The present invention uses a light source that can emit blue light and green light respectively, uses blue light for fluorescence detection and green light for colloidal gold detection, so that the optical signals required to be acquired by the photoelectric conversion module are all green light signals, so that the device of the present invention only requires a single photoelectric conversion module (or a single camera) to be compatible with colloidal gold detection and fluorescence detection, thereby greatly reducing the cost of the device. In particular, when switching between blue light mode and green light mode, the only camera parameter that needs to be adjusted is the camera gain, and there is no need to adjust parameters such as exposure time. In addition, the detection device of the present application uses a symmetrically arranged strip surface light source as a light source for blue light and green light. Compared with other types of light sources, the strip surface light source can provide a more uniform light field, thereby significantly improving the detection stability of the detection device of the present application. Based on this, the inventors have completed the present invention.

[0136] the term

[0137] As used herein, "length", such as the length of a reagent strip, a reagent strip slot, a detection area, a detection window and / or a strip-shaped surface light source, refers to the length along the direction from A to B (see Figure 1 ).

[0138] Similarly, "width" refers to the width along the AC direction (see Figure 1 ).

[0139] As used herein, the term "setting height" refers to the distance from a component (e.g., the lens of a camera, the lens of an auxiliary lens assembly, or a light source) to the plane where the detection window of the object to be detected is located. In this article, unless otherwise specified, the setting height of the light source is the distance from the center of the light source surface to the aforementioned plane.

[0140] As used in this article, "area to be tested" and "detection area" are used interchangeably and refer to the area where the object to be tested actually needs to be detected and analyzed (colloidal gold or fluorescence analysis), for example, the part of the reagent strip with the test line and the quality control line, or part or all of the test piece or grayscale piece.

[0141] As used herein, there is no special requirement for the control module (including the control circuit), which can be used to execute the switching of the apparatus of the present invention between two modes (eg, switching of blue light and green light paths, switching of camera parameters).

[0142] Unless otherwise specified, the terms and / or abbreviations used herein have the common meanings known to those skilled in the art.

[0143] For example, in this article, "wavelet denoising" or "wavelet transform soft threshold denoising" refers to a denoising method based on wavelet transform. Wavelet transform is a time-scale (time-frequency) analysis method for signals. It has the characteristics of multi-resolution analysis and has the ability to characterize local characteristics of signals in both time and frequency domains. It is a time-frequency localization analysis method with a fixed window size but a changeable shape, and both the time window and the frequency window can be changed. That is, it has a lower time resolution and a higher frequency resolution in the low-frequency part, and a higher time resolution and a lower frequency resolution in the high-frequency part. It is very suitable for analyzing non-stationary signals and extracting local characteristics of signals, so wavelet transform is known as a microscope for analyzing and processing signals. Usually, the signals collected from the equipment have a certain amount of noise. In most cases, this noise can be considered as Gaussian white noise. Signal contaminated by noise = clean signal + noise.

[0144] Since the signal has a certain continuity in space (or time domain), the modulus of the wavelet coefficients generated by the effective signal in the wavelet domain is often large; while Gaussian white noise has no continuity in space (or time domain), so after the noise is transformed by wavelet, it still shows strong randomness in the wavelet domain and is usually still considered to be Gaussian white noise. Then we get the following conclusion: in the wavelet domain, the coefficient corresponding to the effective signal is large, while the coefficient corresponding to the noise is small. The coefficient corresponding to the noise in the wavelet domain still satisfies the Gaussian white noise distribution. If in the wavelet domain, the variance corresponding to the wavelet coefficient of the noise is sigma, then according to the characteristics of the Gaussian distribution, most (99.73%) of the noise coefficients are in the interval [-3*sigma, 3*sigma] (Chebyshev inequality, 3sigma criterion). Therefore, as long as the coefficients in the interval [-3*sigma, 3*sigma] are set to zero (commonly used hard threshold function), the noise can be suppressed to the greatest extent, while only slightly damaging the effective signal. The commonly used soft threshold function is to solve the impact caused by the "one-size-fits-all" hard threshold function (all wavelet coefficients with a modulus less than the threshold are cut off, and all those with a modulus greater than the threshold are retained, which is bound to produce mutations in the wavelet domain, resulting in local jitter in the denoising result, similar to the step in the frequency domain in the Fourier transform that produces tailing in the time domain). The soft threshold function sets all wavelet coefficients with a modulus less than the threshold to zero, and performs a special treatment on those with a modulus greater than the threshold. The wavelet coefficients greater than the threshold are uniformly subtracted from the threshold, and the wavelet coefficients less than the threshold are uniformly added to the threshold. After the soft threshold function, the wavelet coefficients are relatively smooth in the wavelet domain, so the corresponding image looks very smooth.

[0145] As used in this article, OpenCV is a cross-platform computer vision library released under the BSD license (open source) and can run on Linux, Windows, Android and Mac OS operating systems. It is lightweight and efficient - consisting of a series of C functions and a small number of C++ classes, while providing interfaces for languages ​​such as Python, Ruby, and MATLAB, and implementing many common algorithms in image processing and computer vision. In the development of computer vision projects, OpenCV, as a more popular open source library, has a rich library of commonly used image processing functions, written in C / C++, and can run on operating systems such as Linux / Windows / Mac, and can quickly implement some image processing and recognition tasks.

[0146] As used in this article, the term "bilinear interpolation" is also called bilinear interpolation. Mathematically, bilinear interpolation is an extension of linear interpolation with an interpolation function of two variables. The core idea is to perform linear interpolation in two directions. As an interpolation algorithm in numerical analysis, bilinear interpolation is widely used in signal processing, digital image and video processing, etc.

[0147] Cross-platform detection device for detecting fluorescent immunoassays and colloidal gold reagent strips

[0148] In order to solve the above problems, the present invention provides a cross-platform detection device, which integrates an immunofluorescence detection system and a colloidal gold detection system. The blue-green light channels are switched by designing a special optical system (including a light source and a special optical path system) and cooperating with a commercially available or self-developed control module (preferably including a control circuit). In addition, the interference signals generated are removed and / or filtered by the built-in post-processing algorithm (such as a denoising algorithm, etc.) in the image processing and analysis module and the narrow-band filter (i.e., a green filter) in the photoelectric conversion module, thereby solving the problems of sensitivity and accuracy in the quantitative test of dry immunolateral flow reagent strips, and optimizing the differences in the reading results between quantitative detection devices through consistency calibration.

[0149] When the detection device of the present invention is in the fluorescence mode, the light source module is used to emit a blue excitation light beam A to the test area of ​​the reagent strip, so that the test area emits a fluorescent signal B; when in the colloidal gold mode, the light source module assembly is used to emit a green excitation light beam C to the test area on the reagent strip, and the test area reflects a signal D. The photoelectric conversion module is used to receive the reflected fluorescent signal B and signal D and perform photoelectric signal conversion to obtain an image for analysis.

[0150] Preferably, in the fluorescence mode, the light source module assembly emits a 30nm blue light with a wavelength of 450nm as the center and a half-peak width of Figure 3AWhen the object to be tested (such as a reagent strip) is irradiated with an excitation light beam A, its fluorescent substance emits a 525nm light wave, namely a fluorescent signal B; the emitted light wave is captured by the camera through a filter assembly (such as a green filter centered at 520nm) and a lens assembly, thereby obtaining an image for analysis.

[0151] Preferably, in the colloidal gold mode, the light source module emits green light with a wavelength of 525 nm as the center and a half-peak width of 30 nm (such as Figure 3B When the object to be tested (such as a reagent strip) is irradiated with a green excitation light beam C, it will reflect the same light wave, i.e., a signal D reflected from the area to be tested; the reflected light wave is optionally captured by a camera through a filtering component (such as a green filter centered at 520 nm) and a lens component, thereby obtaining an image for analysis.

[0152] In another preferred embodiment, the detection device is a dry immunoquantitative detection device.

[0153] In a specific embodiment, Figure 1 , Figure 2 As shown, the detection device of the present invention comprises:

[0154] (a) base 2;

[0155] (b) a fixed structure 3 for the object to be tested, which is located on the base 2; and the fixed structure 3 for the object to be tested is provided with a detection window (the detection window is used to expose the area of ​​the object to be tested (or referred to as the area to be tested) to the light emitted by the light source and the camera lens);

[0156] (c) a photoelectric conversion module, the photoelectric conversion module is used to convert the light signal emitted or reflected by the object to be tested into an electrical signal and obtain an image of the area to be tested containing the object to be tested; wherein the photoelectric conversion module includes: a camera; and the camera (preferably, through a camera fixing structure) is configured such that: the objective lens of the camera is located above the detection window; preferably, the camera fixing structure includes: a camera fixing plate 41 and a support column 42; preferably, the camera fixing plate is provided with an opening for fixing the camera and its four corners are provided with hole structures for connecting with the support column;

[0157] (d) a light source module, the light source module comprising: (d1) a light source; the light source is a pair (2) of symmetrically arranged strip-shaped surface light sources 51; and the light sources can emit blue light or green light respectively;

[0158] Preferably, the light source module of the present invention also includes: (d2) a diffuse reflection plate (52); preferably, a diffuse reflection plate with a thickness of 1 to 3 mm; more preferably, a diffuse reflection plate with a thickness of 2±0.5 mm; (d3) a driving submodule, the driving submodule is used to supply constant current to the light source; and / or (d4) a light source fixing frame (53), the light source fixing frame is used to fix the light source, and the fixing frame can adjust the position along the AB direction.

[0159] Preferably, it also includes: (e) an outer box 1, which is used to form a darkroom required for detection. More preferably, the outer box is a metal outer box; preferably, an aluminum metal outer box, for example, a shielding outer box of 6061 aluminum alloy; the outer box is used to create a darkroom environment to avoid the influence of natural light from the outside, and at the same time has good anti-electromagnetic interference performance and excellent heat dissipation performance. Preferably, the outer box is also provided with a wiring hole 11 (the diameter of the wiring hole is 10 to 20 mm, such as 15 mm).

[0160] Preferably, it also includes: (g) an image processing and analysis module; wherein the image processing and analysis module is configured to obtain a detection result based on the image obtained by the photoelectric conversion module.

[0161] Preferably, the device further comprises: (h) an output module, the output module being used to output the test results. More preferably, the output module comprises a micro printer device for printing the test results.

[0162] Photoelectric conversion module

[0163] In a specific embodiment, the camera is a CMOS camera. Preferably, the performance of the CMOS camera in the photoelectric conversion module is 12 bits output, and the resolution is 1920*1080, 1272*1016, 660*504bin, 648*472ROI.

[0164] More preferably, the camera has one or more of the following parameters: (a) Gain value adjustment range 6.04-14.42dB, default 6.04; (b) Gamma value adjustment range 0.20-1.50, default 1.0; (c) in 1272×1016 mode, the frame rate can reach 6fps; (d) 12bits A / D output, output grayscale value range 0-4096; (e) output format raw, tif, jpg, png; in 1272×1016 mode, at 5.4fps, exposure range 0-401ms; (f) Contrast adjustment range 0-255, default 128; (g) Black level adjustment range 0-511, default 0; and / or (e) Stretch color level adjustment range 0-255, default 25. In a specific embodiment, the CMOS camera includes a Sony IMX290 sensor.

[0165] Light source module

[0166] The detection device of the present invention can switch between fluorescence and colloidal gold quantitative detection items according to different selected test items, thereby maximizing the optional types of test items, and switching of the excitation light source is achieved by controlling the switching of the LED light source channels (including the blue light channel and the green light channel) of the circuit board through the control module.

[0167] In a specific embodiment, the light source module provided by the present invention includes two symmetrically arranged LED strip surface light sources 51. Preferably, each LED strip surface light source is formed by a staggered array arrangement of blue and green LED lamp beads (referred to as blue-green lamp beads), wherein one channel for blue light and one channel for green light, and these two channels can be switched by an LED circuit control board. Among them, the blue LED lamp beads (or blue light lamp beads) emit blue light, and the green LED lamp beads (or green light lamp beads) emit green light.

[0168] Preferably, the blue-green lamp beads in the strip-shaped surface light source are arranged in an array, and the blue-green lamp beads are powered by a constant current power supply (for example, powered by a driving circuit), and a temperature compensation circuit is built in, so as to ensure the consistency of the light intensity under the condition of 25°-40°. The temperature compensation circuit can also be used to reduce the impact of the external environment temperature change on the CMOS camera.

[0169] In the present invention, a thermistor is added to the LED driving circuit of the driving submodule for providing constant current power supply to the light source of the present invention to compensate for the LED driving current. In order to achieve quantitative and precise control of the unit light field intensity, each LED can independently perform 6-bit point correction current adjustment and 12-bit grayscale PWM adjustment, and the maximum output power of a single channel is 30mA / 13V. The LED conduction time can reach a minimum of 0.5us and provide a 30MHz digital interface at the same time, so it can meet the demand for a single-channel LED light intensity to respond quickly according to system instructions. The driving circuit used in the present invention can adopt a commercially available LED driving unit or circuit with a temperature compensation function, or it can be designed by itself. Preferably, the present invention establishes a micro-sensing network inside the LED light source, and the data collected by the sensing network at multiple points provides a rich state judgment basis for multi-modal control. At the same time, the temperature acquisition resolution is increased from the conventional 12 bits to 16 bits, and its accuracy is increased from ±1°C to ±0.25°C.

[0170] Image processing and analysis module

[0171] The image processing and analysis module of the present application further improves the signal-to-noise ratio of the image by using wavelet transform soft threshold denoising. In addition, the module further corrects and compensates for the background processing of the reagent sheet through the post-processing algorithm, eliminating the impact of the change of the reagent sheet background on the true value of the signal.

[0172] Detection Methods

[0173] The present invention also provides a detection method using the detection device as described in the first aspect (the general process is as follows Figure 4A As shown), comprising the steps of:

[0174] (1) Providing a detection device as described in the first aspect;

[0175] (2) placing the object to be tested (such as a test piece, a pre-reacted reagent strip, etc.) on the object fixing structure;

[0176] (3) Select colloidal gold mode or fluorescence mode;

[0177] in,

[0178] In the fluorescence mode, the light source of the light source module of the detection device emits blue light, so that the fluorescent markers (such as fluorescent microspheres) on the test area of ​​the test object emit fluorescent signals under the action of the blue light;

[0179] In the colloidal gold mode, the light source of the light source module of the detection device emits green light, and the colloidal gold marker on the detection area of ​​the object to be detected absorbs part of the green light and reflects the remaining green light (i.e., the reflection signal);

[0180] (4) receiving the reflection signal or the fluorescence signal through the photoelectric conversion module, thereby obtaining an image (or original image) in the colloidal gold mode or the fluorescence mode;

[0181] (5) Obtaining signal spectra and / or detection results based on the image (or original image), which is post-processing algorithm.

[0182] In another preferred embodiment, in the fluorescence mode, the camera parameters are exposure time: 4.93ms, camera gain: 7.93, other camera parameters default values ​​(see camera parameters), and 12V voltage input.

[0183] In another preferred embodiment, in the colloidal gold mode, the camera parameters are exposure time: 4.93ms, camera gain: 6.04, other camera parameters default values ​​(see camera parameters), and 12V voltage input.

[0184] In another preferred embodiment, the method for post-processing algorithm includes the following steps (roughly as follows: Figure 4B shown):

[0185] (5.1) Image preprocessing: preprocessing the obtained image (or original image) (preferably, the preprocessing is wavelet denoising);

[0186] (5.2) Image adjustment: adjust the acquired image (including rotation correction of the image);

[0187] (5.3) Identify and extract the image to obtain the signal spectrum.

[0188] Preferably, the post-processing algorithm further comprises the step of: (5.4) performing Gaussian smoothing on the signal peaks of the signal spectrum.

[0189] Preferably, the post-processing algorithm further comprises the step of: (5.5) calculating the signal peak value, peak area, C / T peak ratio and / or C / T peak area ratio based on the signal spectrum (preferably, the signal spectrum after smoothing).

[0190] Preferably, the post-processing algorithm further comprises the step of: (5.6) calculating the concentration value according to the signal peak value, peak area, C / T peak ratio and / or C / T peak area ratio.

[0191] Preferably, step (5.6) includes the steps of: optionally (5.6.1) calculating the corresponding signal true value estimate (i.e., elimination of matrix effect or correction of reaction signal) based on the signal peak value, peak area, C / T peak ratio and / or C / T peak area ratio; and

[0192] (5.6.2) Calculate the concentration based on the signal peak value, peak area, C / T peak ratio and / or C / T peak area ratio (when matrix effects are not significant), or the corresponding estimate of the true signal value (when matrix effects are significant).

[0193] In another preferred embodiment, the detection device of the present invention is particularly suitable for detecting an analyte having a TAP within the following range: TAP≥900, preferably, particularly suitable for TAP≥1500 (for example, TAP≥2000 or TAP≥2500 and TAP≤4000 (or TAP≤3500, or TAP≤3000).

[0194] Image preprocessing

[0195] Photoelectric noise will affect the final result in the quantitative analysis of gold immunochromatographic reagent strips. Photoelectric noise mainly comes from gold immunochromatographic test strips, photoelectric sensors, and external light source fluctuations.

[0196] The noise generated by these three factors usually exists in the form of random noise (white noise). For accurate quantitative analysis, the image needs to be preprocessed (including denoising). In the present invention, wavelet transform is used to remove Gaussian white noise. Compared with other common denoising methods, wavelet transform denoising can obtain the best image signal-to-noise ratio.

[0197] The filtering algorithm in the present invention is implemented on the C# platform. Based on the principle of wavelet transform soft threshold denoising, the wavelet basis function coif is used to obtain a very good denoising effect. According to the multi-resolution analysis theory, the wavelet coefficients of the high-level decomposition correspond to the low-frequency part, and the low-frequency part is mainly composed of signals. Therefore, the higher the decomposition level, the more low-frequency components are removed, the more obvious the denoising effect is, but the distortion is also increased accordingly. Through the original colloidal gold reagent image library test, the best balance point can be obtained when the decomposition level N=2.

[0198] In a specific embodiment, the image preprocessing comprises the steps of:

[0199] (5.1.1) Wavelet decomposition of two-dimensional signal Select the wavelet basis and the level N of wavelet decomposition (preferably, N=2), and then calculate the decomposition of signal s to the Nth level.

[0200] (5.1.2) Threshold quantization of high-frequency coefficients. For each layer from 1 to N, select a threshold and perform soft threshold quantization on the high-frequency coefficients of this layer.

[0201] (5.1.3) Two-dimensional wavelet reconstruction. Based on the low-frequency coefficients of the Nth layer of wavelet decomposition and the modified high-frequency coefficients from the first layer to the second layer, the wavelet reconstruction of the two-dimensional signal is calculated.

[0202] Preferably, the wavelet soft threshold is estimated using the following formula in the present invention:

[0203] Threshold=sigma*sqrt(2ln(N D1 )),N D1 is the number of wavelet coefficients after the first layer decomposition

[0204] Where sigma = median (abs (wavelet coefficient of D1 layer)) / 0.6745

[0205] Preferably, the present invention adopts wavelet transform soft threshold denoising, wherein the wavelet base adopts coif3 and the wavelet decomposition level N=2.

[0206] Image Adjustment

[0207] Since the signal in the detection area needs to be processed by taking the row average in the later stage, in the camera's shooting field of view, the edge of the detection area often has a certain angle with the edge of the field of view, such as 1°-2°. Directly taking the average of the detection area will affect the true reading of the signal, so the image is rotated and corrected.

[0208] In a specific embodiment, the image adjustment comprises the steps of:

[0209] (5.2.1) Call OpenCV to perform horizontal edge detection on the reagent strip in the detection image.

[0210] (5.2.2) Get the straight line of the edge, calculate the slope, and calculate the angle.

[0211] (5.2.3) Use interpolation to rotate the image so that the edge of the reagent strip is parallel to the edge of the field of view.

[0212] Preferably, the OpenCV library is used for image post-processing;

[0213] Preferably, a Sobel operator is used for horizontal edge detection;

[0214] Preferably, a bilinear interpolation method is used to rotate the image.

[0215] Matrix effect elimination algorithm

[0216] The present invention also provides a result correction algorithm to further reduce the impact of background or matrix effects on target results (such as concentration).

[0217] Since the reaction signal of the reagent sheet is a colorimetric reaction, the corresponding reaction signal value will be superimposed by the background signal. Based on big data, this algorithm makes statistics on the change of the signal with the background, fits a curve between the slope of the signal value with the background change and the true value estimate, and designs a formula based on this to obtain an accurate estimate of the true signal, which greatly eliminates the influence of the background on the reaction signal. The reaction signal refers to the fluorescent signal and the colloidal gold signal. The true value of the reaction signal includes: the fluorescent signal with background interference removed, and the colloidal gold signal with background interference removed.

[0218] Reaction signal value = signal true value estimate + b (signal true value estimate) * background signal value + e (Formula 2)

[0219] Among them, b (signal true value estimate) represents the slope of the signal true value as the base changes (preferably, a point set diagram is obtained through big data analysis, and the relationship between the true value estimate and the changing slope is fit through the point set diagram), e is the systematic random error, and Base refers to the background signal value (or base signal value).

[0220] In the present invention, b (signal true value estimation) (function) will change with the detection reagent and / or the detection target product. Preferably, the slope of the signal true value changing with base (i.e., b (signal true value estimation) function) is a large data pool established by a series of clinical simulation data of a specific product (e.g., data obtained for a specific detection target and / or using a specific detection reagent); and the known signal true value of different target detection results (such as concentration) and the slope (i.e., b (known signal true value)) that changes with base are estimated (e.g., the known signal true value is calculated according to the known concentration, and the slope b (known signal true value) is calculated by formula 2 (wherein, the signal true value estimation is replaced with the known signal true value) according to the detection result obtained by the device of the present invention) (wherein, the signal true value estimation is replaced with the known signal true value)), thereby generating a scatter plot of the signal true value and the slope that changes with base; after removing the abnormal points, fitting matching is performed by fitting model, thereby obtaining a function for calculating b (signal true value estimation) function. The fitted b (signal true value estimate) function is substituted into the formula for back calculation to obtain the signal true value (i.e., the signal true value estimate in Formula 2). In addition, to further ensure the robustness of the model, the data pool may also include data obtained using other detection reagents and / or other detection instruments.

[0221] In a specific embodiment, the present invention provides a fitting model partially suitable for b (signal true value estimation) function fitting and the b (signal true value estimation) function obtained by fitting:

[0222] 1.y=a*x^b+c(Model 1)→b(That)=-0.001778*That^0.7807+1.118(Formula 3)

[0223] 2.y=(p1*x+p2) / (x+q1)(Model 2)→b(That)=(-3.048*That+11780) / (That+10910)(Formula 4)

[0224] 3.y=a*x^2+b*x+c(Model 3)→b(That)=0.00000002006*That^2-0.0003549*That+1.073(Formula 5)

[0225] 4.y=a*x+b(Model 4)→b(That)=-0.0002758*That+1.036(Formula 6)

[0226] Among them, formulas 1-4 are all obtained by fitting different models based on the test data obtained from Airuid's PCT procalcitonin colloidal gold kit (available commercially) (for other test products (lateral flow reagents), single T lines and double T lines can be tested by this method to establish a data pool, fit with different models, obtain appropriate parameters, and obtain the true value estimate of the signal by solving the formula.

[0227] It should be understood that the b (signal true value estimation) function for different types of products such as reagent strips can also be obtained using a similar method based on other detection data (such as data obtained using other detection reagents and / or for other detection targets).

[0228] Different b(That) formulas are used to solve the equations and obtain true value estimates. The true value estimates are used to make calibration curves and back-calculate sample concentrations, which are in good agreement with the sample target values ​​of Roche chemiluminescence.

[0229] The main advantages of the present invention include:

[0230] The quantitative detection device of the present invention adopts a specially designed optical system and temperature compensation circuit and a post-processing algorithm and correction of matrix effects, as well as consistency calibration between modules, so as to make immunoquantitative detection more sensitive and accurate.

[0231] (a) Compared with other detection devices, the linear range is wider, the sensitivity is higher, and the repeatability is good. The fluorescence detection has a wide linear range (20-120000), high sensitivity (up to 0.02% of full scale), and high repeatability (CV < 1%).

[0232] (b) It is compatible with both fluorescent and colloidal gold reagent tablets.

[0233] (c) Under the same ambient temperature, the modules have good consistency.

[0234] (d) Compared with the traditional photoelectric detection head, the camera shooting of the present invention can identify the complete detection window area (including more details), and promptly detect abnormal samples caused by matrix effects. For samples with obvious matrix effects, correction processing is performed on the algorithm so that the concentration calculation can eliminate the influence of matrix effects or background.

[0235] (e) The present invention uses a constant current source and a multi-bead array LED surface light source to ensure the uniformity of the light field and the stability of the lighting (even if the light source is turned on for 24 hours a day and is always on for 1 year, the intensity of the light source after attenuation is still ≥80%). Specifically, the deviation of the mcd value of the luminous intensity of the same color lamp beads used in the present invention is 5% (selected by a spectrometer), which has better luminous consistency, and a 32-bit high-density array LED design (blue and green lamp beads are arranged in an alternating manner) is selected to replace the traditional single-point light source design, greatly improving the uniformity of the light field.

[0236] (f) Later algorithms such as wavelet denoising can maximize the signal-to-noise ratio;

[0237] (g) The quality control line and the reaction line are considered simultaneously when calculating the reaction concentration (i.e., the signal is calculated using the C / T peak value ratio or peak area ratio) to offset the attenuation of the light source.

[0238] (h) When the detection device of the present invention switches between the two modes, only a few contents need to be switched, and only the light source needs to be switched and the parameters of the camera need to be simply adjusted (only the exposure time needs to be adjusted).

[0239] Therefore, the present invention is compatible with at least two reagent sheets of different methodologies, achieving high-sensitivity, high-accuracy, stable and reliable detection.

[0240] The present invention will be further described below in conjunction with specific examples. It should be understood that these examples are intended only to illustrate the present invention and are not intended to limit the scope of the present invention. The experimental methods for the unrecorded specific conditions in the following examples are usually based on conventional conditions or the conditions recommended by the manufacturer. Unless otherwise stated, percentages and parts are weight percentages and weight parts.

[0241] Test piece / standard piece description

[0242] 1. Fluorescence standard film

[0243] The test data of the corresponding No. 1-6 fluorescent standard films under the photoelectric detection probe of QIAGEN Company of Germany are shown in Table 1A below.

[0244] Table 1A Test data of fluorescence standard film on German QIAGEN photoelectric detection probe

[0245]

[0246] 2. Grayscale Standard Film of the Japanese Society of Image Studies

[0247] The standards and specifications of the grayscale sheets used in this application are shown in Table 1B and Figure 5 shown.

[0248] Table 1B

[0249]

[0250] 3. Quality control film

[0251] The standards and specifications of the quality control plates used in this application are shown in Table 1C

[0252] Table 1C Light reflectance of quality control film

[0253] Quality control film number 1 2 3 4 5 6 Light reflectivity 0.2 0.35 0.36 0.5 0.65 0.8

[0254] Example 1 Screening of light sources

[0255] In the selection and setting of light sources, point light sources, line light sources, surface light sources, ring light sources, coaxial light sources and other light sources were considered, and relevant experiments and research were carried out;

[0256] Test environment: 23 degrees.

[0257] Test equipment: Cintu sCMOS Dhyana 401DS camera (16-bit output), lens: PUMIS 4 / 3” 35mm fixed focus lens.

[0258] Test reference: Grayscale standard film of Japan Image Society, among which the light reflectivity of 45% and 40% were selected as test films.

[0259] The height of the camera and lens from the object plane (i.e. the setting height of the camera or auxiliary lens assembly) is fixed at 185mm±2mm, the focal length of the lens is 35.2mm, and the resolution test card ISO12233 is used for testing. The test results show that the resolution is qualified and the captured image is clear.

[0260] In order to achieve the best light field illumination and the best cost performance, LED array light source is selected as the illumination light source in the module.

[0261] In order to verify the influence of different types of light sources on the uniformity of the light field, the following comparison was made according to the current mainstream LED array design, using CV to evaluate the uniformity of rows and columns, where CV = STD / AVG, i.e. standard deviation / mean:

[0262] (1) Ring light source test:

[0263] The light source height is 30mm, the light intensity is 255, the aperture is 5.6, and a 2mm thick diffuser is added to the light source module. The light intensity is calibrated with a grayscale sheet with a 40% light reflectance, and the exposure time is adjusted. The grayscale mean is 32593 and the exposure time is 146μs.

[0264] The pixel row CV of the light reflector with a light reflectivity of 40% is 3.36% and the column CV is 3.35%.

[0265] The pixel row CV of the light reflector with a light reflectivity of 45% is 3.36% and the column CV is 3.25%.

[0266] During the test, it was found that there was a dark spot in the center of the ring light source, resulting in the image when the horizontal row average of the grayscale card was taken (grayscale film with a reflectivity of 45%) as shown below: Figure 6 As shown, the grayscale value of the central area is low, the grayscale values ​​on both sides are high, and the dark spot falls in the center of the detection area.

[0267] Keeping other factors such as lens height unchanged, increase the light source height to expand the range of black spots, raise the light source to 70mm, and let the black spots cover the 8mm distance between the two detection lines. The area of ​​the black spot makes the gray value change radially from the center, which cannot provide a highly uniform initial light field.

[0268] (2) Coaxial light source test:

[0269] The initial setting is that the LED coaxial light source intensity is 200 (digital control, upper limit 255), the light source height is 35mm, the aperture is 5.6, and the exposure time is adjusted to perform median calibration on the 40% light reflector so that its grayscale mean is around 32000. The exposure time at this time is fixed to 106μs.

[0270] Shooting under the same test environment, the test results are as follows:

[0271] The pixel row CV of the light reflector with a light reflectivity of 40% is 4.07% and the column CV is 3.66%

[0272] The pixel row CV of the light reflective sheet with a light reflectivity of 45% is 3.81% and the column CV is 3.40%

[0273] (3) Comparison test of two bar light sources:

[0274] Initially set the light intensity of two control LED strip light sources to 255 (digital control, upper limit 255), the light source height to 35mm, the aperture to 5.6, and a 2mm thick diffuser plate to the light source module. Adjust the exposure time to perform median calibration on the 40% light reflector so that its grayscale mean is 32809, and fix the exposure time at this time to 106μs.

[0275] Shooting under the same test environment, the test results are as follows:

[0276] The pixel row CV of the light reflector with a light reflectivity of 40% is 3.93% and the column CV is 1.41%

[0277] The pixel row CV of the light reflective sheet with a light reflectivity of 45% is 3.46% and the column CV is 2.05%

[0278] (4) Comparison test of two strip light sources:

[0279] Initially set the light intensity of two control LED strip light sources to 255 (digital control, upper limit 255), light source height 35mm, aperture 11, and a 2mm thick diffuser plate outside the light source module. Adjust the exposure time to perform median calibration on the 40% light reflector so that its grayscale mean is 33158, and fix the exposure time at this time, the exposure time is 8ms10μs.

[0280] Shooting under the same test environment, the test results are as follows:

[0281] The pixel row CV of the light reflector with a light reflectivity of 40% is 1.31% and the column CV is 1.35%

[0282] The pixel row CV of the light reflective sheet with a light reflectivity of 45% is 1.39% and the column CV is 1.41%

[0283] In summary, comparing different light source module settings, the two strip surface light sources have the smallest CV and a relatively uniform light field. At the same time, the uniformity of the LED light source is fully considered, and the design of the array LED module adopts a patch LED light source. A 2mm thick diffuser is added to the LED light source module to improve the diffusion effect of the light source, improve the uniformity of the light source, and further improve the uniformity of the light field. Taking into account that the standard film itself also has a certain CV, the strip surface light source used in the present invention has excellent uniformity, which significantly improves the uniformity of the light field required for detection, thereby significantly reducing the coefficient of variation (CV) and improving the detection stability of the cross-platform detection device of the present invention.

[0284] Example 2: Method for calibrating consistency between cross-platform detection devices

[0285] In order to make the module output results under the same environment highly consistent, the present invention achieves the consistency of the output results of the integrated system from the perspective of individually calibrating each component.

[0286] i. Camera calibration platform

[0287] Platform Description:

[0288] For CMOS cameras, the differences between each image chip and electronic components lead to differences in the performance of each batch of array detectors. The photon transfer curve (PTC) is an effective method widely used to evaluate the performance of array detectors. The present invention uses a PTC test system to verify the performance of the test camera. The consistency calibration method is basically as follows: Figure 7B In addition, the cross-platform detection module of the present invention can also be calibrated by other calibration methods in the art.

[0289] The PTC test system mainly consists of a high-brightness halogen lamp (200W), a narrow-band filter (central wavelength 594nm, passband 40nm), and an integrating sphere with an outlet diameter of 8cm. Fig. 7A For each different incident signal intensity, a uniform image was collected for data analysis.

[0290] ii. Test Results

[0291] pass Fig. 7A The system in the paper tests three CMOS cameras. The light intensity is changed by changing the exposure time. The analyzed data are all half the size of the camera's field of view (the middle position, i.e. 508 pixels * 636 pixels). The test results are as follows: Figure 7C and 7D shown.

[0292] Depend on Figure 7C and 7D It can be seen that the fitting degree of the light response curves of the above three cameras has reached above 0.999. When the exposure time is 211ms, the grayscale values ​​of the three cameras are 3883, 3854, and 3814 respectively, and the difference is about ±35.

[0293] iii. Camera calibration adjustment

[0294] The same batch of cameras were tested on the calibration platform, and the cameras at the median value were selected as standard cameras under high exposure (211ms, close to the saturated grayscale value). Under the premise that each camera had good linearity, calculations were performed according to Formula I.

[0295] Y 标准 =K*Y 调整 (Formula I)

[0296] Where K is the camera calibration coefficient, Y 标准 is the grayscale value of the standard camera under high exposure (211ms), Y 调整It is the grayscale value of the camera to be adjusted under the same high exposure (211ms). By calculating the calibration coefficient K, the output result of each camera is multiplied by the coefficient K to achieve normalization processing of different cameras.

[0297] 1) Field of view center calibration

[0298] Assemble the camera on the machined structural fixture, turn on the camera to shoot the fixed reference object, calculate the center of the reference object based on the shot, and fine-tune the camera screw hole fixing position to ensure that the deviation between the center of the object to be tested and the center of the shooting field of view of each module does not exceed ±10 pixels (<0.1mm).

[0299] 2) Use special calibration sheets to calibrate the light intensity of colloidal gold mode and fluorescence mode respectively

[0300] Since there is a certain deviation in the transmittance of light from the filter and lens, the filter, lens and light source are adjusted as a whole, and the standard film is fixed in the center of the detection module's shooting field of view.

[0301] In the colloidal gold test mode, take a picture of the colloidal gold standard film, calculate the corresponding grayscale value, and compare it with the grayscale value corresponding to the standard film (median calibration). Change the supply current of the colloidal gold light source by adjusting the duty cycle (PWM) of the LED light source until the grayscale value calculated by the standard film and the corresponding reference value are within ±5 pixels.

[0302] In the fluorescence test mode, take pictures with the fluorescence median standard film, calculate the corresponding grayscale value, and compare it with the grayscale value corresponding to the fluorescence standard film (median calibration). Change the power supply current of the light source of the fluorescence channel by adjusting the duty cycle (PWM) of the LED light source until the grayscale value calculated by the standard film and the corresponding reference value are within ±5 pixels.

[0303] The preferred colloidal gold standard film in this embodiment is a grayscale film (No. 9 film) with a light reflectance of 40%, and a corresponding grayscale value of 2200 (Japan Image Society). For parameters and specifications, refer to the aforementioned Japan Image Society grayscale standard film.

[0304] The preferred fluorescence standard film in this embodiment is a homemade fluorescence standard film, and the parameters thereof refer to the aforementioned fluorescence standard film.

[0305] Example 3 Image processing and analysis

[0306] Image preprocessing

[0307] Photoelectric noise will affect the final result in the quantitative analysis of colloidal gold immunoassay strips. Photoelectric noise mainly comes from gold immunochromatographic test strips, photoelectric sensors, and external light source fluctuations.

[0308] The noise generated by these three factors usually exists in the form of random noise (white noise). For accurate quantitative analysis, the inventors found that the best image signal-to-noise ratio (SNR) can be obtained by using wavelet transform to remove Gaussian white noise.

[0309] 3.1.1 Screening of several denoising methods

[0310] The present invention evaluates and compares several denoising methods and calculates the corresponding SNRs, and the results are shown in Table 2A.

[0311] Table 2A

[0312] Mainstream filtering denoising algorithms SNR Weiner Filter 41.1916 NLM Filtering 46.4456 Gaussian filtering 33.7241 Median Filter 36.9251 Mean Filter 32.0622 Gaussian low-pass filter 32.3668 Wavelet soft thresholding 50.6413

[0313] As shown in Table 2A, for the filtering processing of the captured images of the colloidal gold reagent tablets, the SNR value after wavelet soft threshold processing is significantly higher than the SNR values ​​after processing using several other processing methods, and is even better than the NLM filtering method which has a high time complexity and cannot achieve real-time image denoising.

[0314] In addition, in this embodiment, the denoising performance of different types of orthogonal wavelet bases is also evaluated, and the results are shown in Table 2B:

[0315] Table 2B

[0316] Haar wavelet haar SNR 47.6923 Symlets sym1 sym2 sym3 sym4 sym5 SNR 47.6923 49.6907 50.2514 50.4517 50.5173 Daubechies wavelet db1 db2 db3 db4 db5 SNR 47.6923 49.6907 50.2514 50.4392 50.4723 Coiflets Wavelet coif1 coif2 coif3 coif4 coif5 SNR 49.773 50.4996 50.6413 50.7006 50.712

[0317] It can be seen from Table 2B that the denoising performance of the coif wavelet is relatively good year-on-year, so the coif series of wavelets are selected. At the same time, the contribution of wavelets above coif3 to SNR does not increase significantly, so the coif3 wavelet is used in this preferred example.

[0318] Example 4 Device Detection Performance

[0319] Outer box: 6061 aluminum alloy shielding outer box;

[0320] Light source: LED symmetrical strip light source (the surface of the sample under test is at an angle of 45 degrees, with a light-emitting surface of 48mm*18mm; the light source height is 35mm.

[0321] Camera: 12bitsA / D output high-sensitivity CMOS black and white camera, the camera is equipped with a 3x microscope magnifier of 525BP20OD6T85 (parameters: central wavelength of 525, bandpass 20nm, OD value 6, transmittance 85%) with a built-in green filter;

[0322] 4.1 Colloidal gold detection part

[0323] The performance of the colloidal gold module is tested in accordance with the requirements of the industry standard YY T 1582-2018 colloidal gold immunochromatography analyzer. Since the accuracy in the test involves reagent products, the performance test of this platform does not involve accuracy. The colloidal gold platform is mainly evaluated from four aspects: resolution, repeatability, linearity, and stability.

[0324] The quality control film with a light reflectivity of 0.2-0.8 used in the evaluation was prepared according to the industry standard and the light reflectivity was measured by the Shanghai Metrology Institute. The specific parameters of the quality control film are shown in Table 1C.

[0325] The inspection requirements for YY T 1582-2018 colloidal gold immunochromatography analyzer are as follows:

[0326] Resolution: Test quality control strips: Repeat the measurement 10 times, and it should be able to distinguish a pair of quality control strips with a reflectivity range of [0.30, 0.40] and a reflectivity difference of no more than 0.01. Converted into a calculation formula, it should specifically meet the following requirements:

[0327] 0<A1-A2≤0.01, S1-S2<0 is satisfied every time. Among them, A1 is the reflectivity of quality control strip 1; S1 is the response value of quality control strip 1; A2 is the reflectivity of quality control strip 2; S2 is the response value of quality control strip 2. That is, take the No. 2 film (A2) and the No. 3 film (A1) and repeat the test 10 times each, and S1 is satisfied every time. <S2。

[0328] Repeatability: Test the quality control strips: Test the three quality control strips with a reflectivity range of [0.20, 0.80], medium, high, and low reflectivity, respectively, calculate the mean (M) and standard deviation (SD) of the 10 measurement results, and calculate the coefficient of variation (CV) according to formula (2), which should be no greater than 3%. CV - coefficient of variation; SD - standard deviation of the 10 measurement results; M - arithmetic mean of the 10 measurement results. That is, take the No. 1, No. 4, and No. 6 pieces and repeat the test 10 times each, calculate the CV, and the CV should be less than 3%

[0329] Linearity: Test quality control strips: Test at least 5 quality control strips with reflectivity evenly distributed in [0.20, 0.80]. Test each quality control strip 3 times. Calculate the linear regression correlation coefficient (r) according to formula (3), which should be no less than 0.990. xi——reflectivity; yi——arithmetic mean of the measurement results. That is, take slides 1, 2, 4, 5, and 6 and repeat the test 3 times each, take the average, and make a linear analysis of the 5 points. Check the R square, which should be greater than 0.990*0.990.

[0330] Stability: Test the quality control strip whose reflectivity signal value is within the measurement range of the instrument. The quality control strip needs to enter / exit the chamber for each test. Test 10 times in a row. The interval between two tests should not be less than 3 minutes. The relative range (R) of the measurement result is calculated according to formula (5). It should not be greater than 5%. That is, take the No. 1 and No. 6 pieces and repeat the test 10 times, each with an interval of three minutes. Calculate the range / mean of the 10 results, which should be less than 5%.

[0331] This test uses the inversely calculated grayscale value, i.e. 4096-real grayscale value. The test results are shown in Tables 3A, 3B, 3C and Figure 8 shown.

[0332] Table 3A Resolution

[0333] Light reflectivity 35％ 36％ Passed No.1 2616 2599 Pass No.2 2616 2597 Pass No.3 2613 2595 Pass No.4 2615 2597 Pass No.5 2617 2593 Pass No.6 2621 2595 Pass No.7 2617 2597 Pass No.8 2615 2597 Pass No.9 2618 2592 Pass No.10 2615 2593 Pass

[0334] The 35% and 36% quality control films were tested 10 times in a row. It can be seen from Table 3A that the detection device of the present invention can completely distinguish the 35% and 36% quality control films without any crossover phenomenon.

[0335] Table 3B Repeatability

[0336] Light reflectivity 20％ 50％ 80％ No.1 3209 2175 1026 No.2 3208 2177 1016 No.3 3209 2175 1022 No.4 3207 2178 1020 No.5 3208 2181 1020 No.6 3208 2179 1029 No.7 3207 2179 1024 No.8 3208 2176 1022 No.9 3207 2177 1026 No.10 3208 2174 1019 AVG 3207.9 2177.1 1022.4 STD 0.7 2.071232 3.693237 CV 0.02％ 0.10％ 0.36％

[0337] The light reflectance was tested 10 times in the linear range, and the coefficient of variation (CV) was calculated. The results are shown in Table 3B. The CV of the detection device of the present invention is less than 0.5%, which is far lower than the less than 3% specified in the industry standard. The detection device has good repeatability.

[0338] Table 3C Stability

[0339] Light reflectivity 20％ 80％ No.1 3209 1026 No.2 3208 1016 No.3 3209 1022 No.4 3207 1020 No.5 3208 1020 No.6 3208 1029 No.7 3207 1024 No.8 3208 1022 No.9 3207 1026 No.10 3208 1019 AVG 3207.9 1022.4 Very bad 2 13 maximum deviation 0.06％ 1.27％

[0340] Table 3C shows the stability test results of the detection device of the present invention. The maximum deviation of the detection device of the present invention in the upper and lower limits of the linear range does not exceed 1.5%, which is far lower than the maximum deviation of less than 5% specified in the industry standard, and has good stability.

[0341] Figure 8 The linear test results of the detection device of the present invention (y = -3674.7x + 3936.5 R 2 =0.9954). R in the linear range 2 >0.990, fully meeting the industry standard requirements.

[0342] 4.2 Fluorescence detection part

[0343] By using a special fluorescent standard film and testing this platform and the detection platform of a first-class foreign manufacturer (QIAGEN) at the same time, the following comparative data and conclusions were obtained.

[0344] 4.2.1 Linear range (TA, reaction peak area)

[0345] The test uses the same fluorescent standard film, changes the camera parameters (exposure time and gain), makes the fluorescence detection line reach the minimum limit and the maximum limit respectively, tests every 30 seconds, repeats 10 times, and calculates the CV of TA (see Table 4A for the results). Under the premise of ensuring that CV satisfies CV<=1%, the detection range of the detection device of the present invention is [1500-120000], the minimum detection line is 20, and CV<=8% is guaranteed.

[0346] Table 4A

[0347]

[0348] 4.2.2 Linearity

[0349] After calibration, the linearity of the detection device of the present invention is above 0.999. Fig.7D .

[0350] 4.2.3 Comparison results of fluorescence standard film detection

[0351] According to the fluorescence performance of the module, six fluorescence standard films (evenly distributed from low value to high value, with the same parameters as before) were used, numbered 1-6, and each film was tested 10 times under the module to calculate the coefficient of variation CV, CV = SD / Average; wherein CY and TY refer to the peak values ​​of the quality control line C and the reaction line T, respectively, Base is the base signal value, Tarea and Carea are the areas of the T line and C line after deducting the base, respectively; TAP = Tarea / (Tarea+Carea)*5000.

[0352] The test results of the fluorescent standard film using the detection device of the present invention are shown in Tables 5A-5F: Table 5A Repeatability test data of fluorescent standard film No. 1

[0353]

[0354] Table 5B Repeatability test data of No. 2 fluorescent standard film

[0355]

[0356] Table 5C No. 3 Fluorescence Standard Plate Repeatability Test Data

[0357]

[0358] Table 5D No. 4 Fluorescence Standard Plate Repeatability Test Data

[0359]

[0360] Table 5E No. 5 Fluorescence Standard Plate Repeatability Test Data

[0361]

[0362] Table 5F No. 6 Fluorescence Standard Plate Repeatability Test Data

[0363]

[0364] The results of the comparison of high and low values ​​of fluorescence detection CV (under the same environment) are shown in Tables 6A and 6B and Tables 7A and 7B:

[0365] Table 6A Stability test data of low-value tablets (i.e., No. 6 fluorescent standard tablets) of Qiagen detection device

[0366]

[0367] Table 6B Stability test data of high value tablets (i.e. No. 1 fluorescent standard tablets) of Qiagen detection device

[0368]

[0369] Table 7A TAP comparison results of Qiagen and the detection device of the present invention (high value tablets and low value tablets)

[0370]

[0371] Table 7B Comparison of fluorescence detection discrimination between Qiagen and the detection device of the present invention

[0372] TAP QIAGEN Discrimination This platform Discrimination 1 2803.26 25.4％ 2872.42 25.0％ 2 2219.00 5.4％ 2285.93 3.1％ 3 2095.65 24.2％ 2212.70 25.7％ 4 1537.04 27.9％ 1611.33 27.8％ 5 893.53 17.1％ 959.14 18.4％ 6 499.23 527.98

[0373] The results show that when the detection device of the present invention is used for detection, the CV of the low-value standard film No. 6 is about 0.5%, and the other high-value or medium-value standard films (No. 1 to No. 5) have good repeatability, with CVs of 0.02%, 0.06%, 0.10%, 0.11%, and 0.17%, respectively.

[0374] As can be seen from Table 7A, when detecting low values ​​of fluorescence, the detection device of the present invention has a CV comparable to that of the detection platform (Qiagen) of a world-class manufacturer. In addition, unexpectedly, when detecting high values ​​of fluorescence, the CV value of the detection device of the present application is extremely excellent (e.g., when TAP is about 2800, the CV (0.02%) of the device of the present invention is only 1 / 6 of that of the control (0.12%). This indicates that the detection device of the present invention is particularly suitable for detection under the conditions of TAP of ≥900, ≥1500, ≥2000, ≥2500 and ≤4000, preferably ≤3500, and more preferably ≤3000.

[0375] As can be seen from Table 7B, the discrimination of the detection device of the present application is comparable to that of Qiagen's detection platform.

[0376] In summary, it can be seen from various embodiments that the device of the present invention has good repeatability, excellent discrimination, and greatly reduced price cost.

[0377] All documents mentioned in the present invention are cited as references in this application, just as each document is cited as reference individually. In addition, it should be understood that after reading the above teachings of the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms also fall within the scope defined by the claims attached to this application.

Claims

1. A cross-platform detection device, characterized in that: The detection device comprises: (a) a base (2); (b) a test object fixing structure (3), the test object fixing structure being located on the base (2); and the test object fixing structure being provided with a detection window; (c) a photoelectric conversion module, the photoelectric conversion module is used to convert the light signal emitted or reflected by the object to be tested into an electrical signal and obtain an image of the area to be tested containing the object to be tested; wherein the photoelectric conversion module comprises: a camera, and the camera is a CMOS camera; and the camera is configured such that the objective lens of the camera is located above the detection window; (d) a light source module, the light source module comprising: (d1) a light source; the light source is a symmetrically arranged strip surface light source (51); and the light source can emit blue light or green light respectively; wherein the strip surface light source comprises blue light beads and green light beads arranged in an interlaced manner; and the surface of the strip surface light source forms an angle of 45±1 degrees with the plane where the area to be measured is located; and, (d2) a diffuse reflection plate (52) having a thickness of 1 to 3 mm; (g) an image processing and analysis module; wherein the image processing and analysis module is used to obtain a detection result according to the image obtained by the photoelectric conversion module; wherein the image processing and analysis module is configured to: perform preprocessing on the image obtained by the photoelectric conversion module; wherein the preprocessing includes denoising the image; and the denoising is performed based on a wavelet transform soft threshold and the wavelet basis function used in the denoising is coif3; The image processing and analysis module is configured to perform the following steps to pre-process the image obtained by the photoelectric conversion module: (i) Wavelet decomposition of the two-dimensional signal of the image: select the wavelet basis function and the level N of wavelet decomposition, and then calculate the decomposition of the two-dimensional signal s to the Nth level; (ii) Threshold quantization of high-frequency coefficients: For each layer from 1 to N, a threshold is selected and a soft threshold quantization is performed on the high-frequency coefficients of this layer, where N is 2; and (iii) Two-dimensional wavelet reconstruction: Based on the low-frequency coefficients of the Nth layer of wavelet decomposition and the modified high-frequency coefficients from the first layer to the second layer, the wavelet reconstruction of the two-dimensional signal is calculated to obtain an image signal reconstructed using the two-dimensional wavelet; The soft threshold used in the denoising process is calculated by formula 1: Threshold=sigma*sqrt(2ln(N D1 )) (Formula 1); In the formula, Threshold represents the soft threshold, sigma represents the variance corresponding to the wavelet coefficient of the noise, and N D1 Represents the number of wavelet coefficients decomposed in the first layer; Here, sigma=median(abs(wavelet coefficient of D1 layer)) / 0.6745.

2. The detection device according to claim 1, characterized in that The light source module has one or more of the following features: (i) the blue light refers to light with a wavelength of 450±10 nm as the center and / or a half-peak width of 20 to 40 nm; and / or (ii) The green light refers to light centered at a wavelength of 525±10 nm and / or having a half-peak width of 20 to 40 nm.

3. The detection device according to claim 1, characterized in that: The light source module also includes: (d2) a diffuse reflection plate with a thickness of 2±0.5 mm.

4. The detection device according to claim 1, characterized in that: The light source module also includes: (d2) a 2 mm thick diffuse reflection plate.

5. The detection device according to claim 1, characterized in that: The photoelectric conversion module further includes: a lens assembly.

6. The detection device according to claim 5, characterized in that: The lens assembly includes: a 2-4 times microscope magnification lens.

7. The detection device according to claim 5, characterized in that: The lens assembly includes: a 3x microscope magnification lens.

8. The detection device according to claim 5, characterized in that: The photoelectric conversion module further includes: a filter component.

9. The detection device according to claim 8, characterized in that: The filter assembly includes a green filter.

10. The detection device according to claim 1, characterized in that: The detection results include: process results and / or target results.

11. The detection device according to claim 10, characterized in that: The process result includes: peak area, peak value, ratio of peak areas, ratio of peak values, or a combination thereof; and / or, the target result includes: concentration.

12. The detection device according to claim 10, characterized in that: The process result is selected from the group consisting of a ratio of peak areas, a ratio of peak values, or a combination thereof; and, the target result is concentration.

13. The detection device according to claim 1, characterized in that: The detection device further comprises: (e) an outer box (1), wherein the outer box is used to form a darkroom required for detection.

14. A detection method, characterized in that: Includes steps: (1) Providing a cross-platform detection device as claimed in claim 1; (2) Placing the object to be tested on the object fixing structure; (3) Select colloidal gold mode or fluorescence mode; (4) a photoelectric conversion module to obtain an image in a colloidal gold mode or a fluorescence mode; and (5) Obtain detection results based on the obtained image.

Images