Digital analysis method and system based on nano lens high signal-to-noise ratio holographic imaging

Abstract

The invention discloses a digital analysis method and system based on nano lens high signal-to-noise ratio holographic imaging, and belongs to the field of biochemical analysis. According to the invention, the microsphere nanoprobe is combined with a high signal-to-noise ratio holographic microscopic imaging technology, the number of microsphere signal probes in a supernatant / immune complex is changed through a series of biochemical reactions, and high signal-to-noise ratio holographic imaging is carried out on the PS microsphere nanoprobe in the supernatant after magnetic separation; in the holographic imaging process, the surface acoustic wave-interdigital transducer is used for driving and forming a high-signal-to-noise-ratio lensless holographic microscope of a nano lens; and obtaining the number of the microsphere signal probes in the holographic image by using a lightweight unsupervised learning target detection algorithm, and finally realizing accurate quantitative detection of the target object to be detected. Experimental results show that the detection sensitivity and efficiency are effectively improved, and meanwhile the detection accuracy is improved.

Description

Technical Field

[0001] The invention relates to a digital analysis method and system based on nano-lens high signal-to-noise ratio holographic imaging, belonging to the field of biochemical analysis. Background Art

[0002] Digital detection is a new biosensing technology with single-molecule detection capability, which has outstanding advantages such as data visualization, high detection sensitivity, and fast detection speed. At present, various digital detection methods have been developed for precise quantitative detection of various types of targets, such as nanopore array biosensors, biosensors based on fluorescent molecules / droplets / microspheres, biosensors based on gold / upconversion nanoparticles, and smart phone imaging mobile detection platforms. However, such methods also have many limitations, such as the precision and expensive imaging equipment and small FOV area, the difficulty of digital image processing, and the low efficiency of biosensors, which hinder the development of digital detection methods. Among them, FOV and SNR are particularly important for digital detection. A wider FOV and a higher SNR represent a "double improvement" in signal probe throughput and sensitivity, but they are limited by the optical objective and the Abbe diffraction limit (~200nm), so it is difficult to have both at the same time. In addition, lighter image processing and more efficient biosensors also represent a significant improvement in the efficiency of digital detection. Therefore, it has become an urgent need to develop a new sensitive and fast digital detection platform. Holographic imaging equipment with ultra-wide FOV and high SNR, lightweight image processing algorithms and efficient and stable biosensing strategies are the three key points to achieve accurate detection.

[0003] At present, there are many ways to implement immunoassay methods based on optical imaging visualization microsphere counting. For example, patent CN 112964868A discloses a biochemical analysis method based on magnetic separation to simultaneously detect multiple targets, and patent CN117551743A discloses a digital biochemical analysis method based on holographic imaging and its supporting detection equipment. Although the above methods successfully achieve accurate quantitative detection of the target to be tested based on PS microsphere imaging counting technology, the detection cost is lower and the operation is more convenient compared with the traditional Coulter hole microsphere resistance method and differential impedance method. However, as a type of optical imaging detection technology, these methods have failed to break the limitation between FOV area and imaging SNR, and realize PS microsphere signal probe imaging with ultra-wide FOV and high SNR in digital detection, which makes it difficult to further improve the accuracy of digital detection. Summary of the invention

[0004] In order to further improve the accuracy of digital detection, the present invention provides a digital analysis method and system based on nano-lens high signal-to-noise ratio holographic imaging, and the technical solution is as follows:

[0005] The digital analysis method based on nanolens of the present invention comprises:

[0006] Step 1: For the target bacteria to be detected, the bacteriophage corresponding to the target bacteria to be detected is coupled to the surface of the nanomagnetic particle MNP to form an MNP-bacteriophage conjugate; for the non-nucleic acid target to be detected, the biological recognition element that recognizes the non-nucleic acid target is coupled to the surface of the nanomagnetic particle to form an MNP-biological recognition element conjugate;

[0007] Step 2: adding the MNP-bacteriophage conjugate to the target bacteria to be tested for reaction, the bacteriophage specifically captures the active target bacteria, and then magnetically separates and retains the MNP-bacteriophage-target bacteria complex, and after standing to lyse and release DNA, collects the reaction solution for later use;

[0008] Step 3: For the target bacteria to be tested, the target bacterial DNA extracted in step 2 is mixed with the Cas12a enzyme assembled with multiple crRNAs to enhance its trans-cutting ability with activity, and then the reaction solution of activating the Cas12a enzyme is added to the MNP-ssDNA-biotin conjugate to trans-cut the ssDNA probe to complete the CRISPR reaction, and finally the microsphere signal probe-streptavidin conjugate probe is added to complete the bioorthogonal reaction and then magnetically separated, and the supernatant containing the microsphere signal probe is retained for holographic imaging;

[0009] For the non-nucleic acid target to be detected, the microsphere signal probe-antibody conjugate, the MNP-antigen conjugate, and the non-nucleic acid target to be detected are mixed for immune reaction, and the immune complex and unreacted nanomagnetic particles MNP are removed after magnetic separation, and the supernatant containing the microsphere signal probe is retained for holographic imaging;

[0010] Step 4: After mixing the supernatant containing the microsphere signal probe with PEG-600 in proportion, drip it onto the plasma-cleaned cover glass and spin-coat it, and place it in the sensing area of ​​the holographic microscope for high signal-to-noise ratio holographic imaging. During the holographic imaging process, a function signal generator and a power amplifier are used to emit a signal wave of a specific frequency to excite the piezoelectric effect of the SAW-IDT resonator chip and generate a surface acoustic wave, which propagates and couples to the cover glass to cause the thin layer of liquid on the surface to evaporate and deform, forming a nanolens phenomenon to increase the scattering cross section at the probe imaging location, thereby improving the signal-to-noise ratio of the probe holographic imaging;

[0011] Step 5: Use the trained unsupervised learning target detection algorithm to identify the holographic image and obtain the number of microspheres;

[0012] Step 6: Calculate the content of the target to be detected by using the standard curve of the relationship between the number of microspheres and the content of the target to be detected, so as to achieve quantitative detection of the target.

[0013] Optionally, the microspheres include: polystyrene microspheres, silica microspheres, polybutadiene microspheres, and polyisoprene microspheres.

[0014] Optionally, the particle size of the microspheres ranges from 0.1 to 50 μm.

[0015] Optionally, the biorecognition element includes: complete antigen / antibody, capture antibody / detection antibody, DNA capture probe / DNA detection probe.

[0016] Optionally, the excitation frequency range of the SAW-IDT resonator chip is 0-100 MHz, and the waveform is one of a sine wave, a triangle wave, a sawtooth wave, and a square wave.

[0017] Optionally, the immune response includes a competitive immune response, a double antibody sandwich immune response or a DNA hybridization response.

[0018] Optionally, the particle size of the MNPs ranges from 50 to 5000 nm.

[0019] Optionally, the aperture diameter of the holographic microscope is in the range of 50-1000 μm, and the wavelength of the coherent light source is in the range of 390-780 nm.

[0020] Optionally, the bioorthogonal reaction (SA-biotin) can be any one of the bioorthogonal reactions, such as a click reaction, involving functional groups including azide-alkyne, tetrazine (Tz)-trans-cyclooctene (TCO), dibenzocyclooctyne (DBCO)-azide (Azide), cysteine ​​(Cys)-cyanobenzothiazole (CBT) or others.

[0021] Optionally, the number of multiple crRNAs in the CRISPR reaction ranges from 1 to 10.

[0022] Optionally, the molecular weight of PEG added to form the nanolens ranges from 100-1000 and the concentration ratio ranges from 1-90%.

[0023] Optionally, the targets to be detected include viruses, bacteria, RNA molecules, biomarkers, antibiotic molecules, pesticide molecules, veterinary drug molecules, and biological toxins.

[0024] The present invention provides a digital analysis system based on nano-lens, comprising: a signal generator, a power amplifier, a SAW-IDT resonator chip, a holographic imaging device, a target recognition module, and a result analysis module;

[0025] The signal generator and the power amplifier excite a sine wave signal of a specific frequency and transmit it to the SAW-IDT resonator chip, and the piezoelectric substrate of the SAW-IDT resonator chip generates a surface acoustic wave, which is finally coupled and propagated to a thin layer of liquid to be tested on a disposable cover glass, wherein the liquid to be tested is the supernatant containing the microsphere signal probe as described in any one of the above items;

[0026] The holographic imaging device is used to perform holographic imaging on the supernatant containing the microsphere signal probe;

[0027] The target recognition module is configured to use the trained target detection algorithm to recognize the holographic image and obtain the number of microspheres;

[0028] The result analysis module is configured to calculate the content of the target to be detected by using a standard curve of the relationship between the number of microspheres and the content of the target to be detected, thereby achieving quantitative detection of the target.

[0029] Optionally, the piezoelectric substrate material of the SAW-IDT resonator chip includes: lithium niobate, lithium tantalate, and aluminum nitride.

[0030] Optionally, the holographic imaging device includes: a CMOS camera, a CCD camera, and a smartphone image sensor.

[0031] The beneficial effects of the present invention are:

[0032] (1) The present invention realizes high SNR lensless holographic imaging under ultra-wide FOV. In view of the problem that it is difficult to achieve both ultra-wide FOV and high SNR in traditional digital imaging equipment, the present invention is different from the traditional method of realizing high SNR holographic imaging by combining complex algorithms and precision equipment. It innovatively breaks through the Abbe diffraction limit by using the sound-driven "nano lens" effect to increase the scattering cross section at the probe, thereby improving the contrast of the imaged nanoparticles relative to the background noise, thereby solving the contradiction between the FOV area and SNR in lensless holographic imaging. In addition, the high SNR holographic imaging equipment constructed by the present invention effectively improves the contrast and SNR of the PS microsphere nanoprobe, so that it can be detected relative to the background noise, and can be directly used for target detection without the need for complex digital holographic reconstruction, which is crucial to improving the sensitivity and detection efficiency of the digital detection platform.

[0033] (2) The image processing speed of the present invention is faster and the training time is shorter, and a large number of manually labeled data sets are not required. In order to solve the problems that supervised deep learning target detection algorithms rely on artificial data sets, have long training times, and have poor adaptability, the present invention constructs a lightweight unsupervised learning image processing algorithm through geometric transformation and weight assignment, and innovatively applies it to digital holographic image processing. Without the need for long-term training with a large number of complex data sets with artificially marked real labels, accurate detection of the target to be measured can be achieved, further improving the image processing efficiency of the digital detection platform, reducing the requirements for training equipment, and saving a lot of labor costs.

[0034] (3) The efficiency of the biochemical reaction of the present invention is significantly improved, and the detection sensitivity is further improved. In order to solve the problem of low sensing efficiency of the amplification-free CRISPR system combined with the PS microsphere probe, the present invention innovatively combines the multi-crRNA activity-enhanced CRISPR system with the PS microsphere nanoprobe for digital detection, shortening the amplification-free detection time of foodborne pathogens from 1.5 hours in previous work to 1 hour (compared to patent CN117551743A), and the detection can be completed without complicated pre-treatment and pre-amplification methods. In addition, the small-particle PS microsphere probe has higher detection sensitivity and smaller steric hindrance, and cooperates with the high SNR holographic microscope constructed in the present invention to more sensitively monitor the changes in the concentration of the target to be detected, thereby increasing the detection sensitivity of Salmonella typhimurium by about 2 times compared with previous work. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0036] Figure 1 The figure is a flow chart of the biochemical reaction principle for detecting foodborne pathogens in an embodiment of the present invention.

[0037] Figure 2 This is a schematic diagram of a high SNR holographic microscope using a SAW-IDT-driven nanolens built in an embodiment of the present invention.

[0038] Figure 3 This is a flow chart of unsupervised learning holographic image processing based on geometric self-transformation in an embodiment of the present invention.

[0039] Figure 4 Schematic diagram of the structure of the SAW-IDT resonator chip in an embodiment of the present invention.

[0040] Figure 5This is a comparison chart of the field of view area of ​​the holographic microscope constructed in an embodiment of the present invention and the traditional optical microscope.

[0041] Figure 6 This is a diagram for the high SNR holographic microscope imaging limit test and algorithm recognition accuracy verification built in an embodiment of the present invention.

[0042] Figure 7 This is a diagram showing the optimization results of key parameters for nanolens formation and biochemical reaction in Example 1.

[0043] Figure 8 This is a graph showing the linear range results of the method of the present invention in Example 1 for detecting Salmonella typhimurium in chicken samples.

[0044] Fig. 9 This is the SEM characterization image of the MNP-phage conjugate capturing live Salmonella typhimurium and the PS-ssDNA-MNP complex in Example 1.

[0045] Fig.10 This is a graph showing the linear range results of the method of the present invention in Example 2 for detecting Staphylococcus aureus in lettuce samples.

[0046] Fig.11 This is a graph showing the linear range results of the method of the present invention in Example 3 for detecting interleukin-6 in serum samples.

[0047] Fig.12 This is a graph showing the linear range results of detecting chloramphenicol in milk samples using the method of the present invention in Example 4. DETAILED DESCRIPTION

[0048] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0049] Example 1: Quantitative detection of Salmonella typhimurium in chicken samples

[0050] The biochemical reaction detection principle involved in this embodiment is as follows Figure 1 As shown, the biochemical reaction process can be roughly divided into two steps: (1) biosensing based on the CRISPR / Cas12a system; (2) artificial intelligence signal transduction.

[0051] First, the phage corresponding to Salmonella typhimurium is coupled to MNP, and the MNP-phage conjugate is added to the test object. The phage specifically captures the live Salmonella typhimurium bacteria. After magnetic separation, the MNP-phage-Salmonella typhimurium live bacteria are retained and the bacteria are lysed to release DNA, completing the enrichment and extraction of target DNA. The specifically designed multiple crRNAs can specifically bind to the target DNA of Salmonella typhimurium, share a target nucleic acid substrate while forming more ribonucleoproteins, and activate more Cas12a enzymes to enhance trans-cutting activity. Among them, the activity-enhanced Cas12a enzyme triggers the CRISPR reaction, thereby randomly trans-cutting the MNP-ssDNA-biotin conjugate multiple times, so that the ssDNA single-strand break hinders the coupling with PS-SA to form a PS-SA-biotin-ssDNA-MNP complex, and changes the number of PS microspheres in the supernatant after magnetic separation. In short, when the target nucleic acid concentration is higher, the more the number of activated Cas12a enzymes is, the more MNP-ssDNA-biotin is cut off, and the number of microspheres in the supernatant after magnetic separation is greater; on the contrary, the lower the target nucleic acid concentration is, the fewer the number of activated Cas12a enzymes is, the less MNP-ssDNA-biotin is cut off, forming more PS-ssDNA-MNP complexes, and the number of microspheres in the supernatant after magnetic separation is less. Therefore, the number of microsphere signal probes in the supernatant is negatively correlated with the concentration content of the bacteria to be tested. These supernatants containing PS microsphere probes are retained for high SNR holographic imaging and unsupervised learning convolutional neural network (UL-CNN) recognition counting, completing artificial intelligence signal transduction, feedback of PS microsphere quantity and calculation of target content to be tested, and accurate quantitative detection.

[0052] The detailed implementation steps of this embodiment are as follows:

[0053] (1) Preparation of PS microsphere-streptavidin conjugate (PS-SA), nanomagnetic particle-biotinylated single-stranded "bridge" probe conjugate (MNP-ssDNA-biotin) and MNP-phage conjugate.

[0054] Preparation of PS-SA conjugate: First, 1 mg of 3 μm PS microsphere solution was added to a 1.5 mL centrifuge tube and washed 3 times with MEST buffer. EDC (15 μL, 5 mg / mL) and NHS (15 μL, 5 mg / mL) were then added to the PS microsphere solution to activate the carboxyl groups on the surface of the microspheres, and the mixture was diluted to 1 mL with MES buffer and slowly rotated at room temperature for 15 minutes. The activated PS microsphere solution was washed 3 times with PBST buffer (1 mL) and centrifuged at 6000 r / min for 5 minutes. Subsequently, SA (30 μg) was added to the centrifuge tube containing the activated PS microspheres and diluted to 500 μL with PBS buffer, and the mixture was slowly rotated at 37 ° C for 30 minutes to allow the reaction to be sufficient. After washing, PBST buffer (1 mL, containing 1% BSA, pH 7.4) was added to mix with the PS-SA conjugate to block the residual active sites for 30 minutes. Finally, the blocked PS-SA conjugate was washed three times with PBST buffer (1 mL), resuspended with PBST (1 mL, containing 0.5% BSA, 0.02% NaN3), and stored at 4°C.

[0055] Preparation of MNP-ssDNA-biotin conjugate: First, add the newly prepared EDC (5 μL, 5 mg / mL) and NHS solution (5 μL, 5 mg / mL) to the biotin-ssDNA-COOH (10 μL, 50 μM) solution and mix and react at room temperature for 15 minutes to activate the carboxyl groups on the ssDNA probe. Subsequently, add the activated biotin-ssDNA-COOH to 1 μm MNP1000-NH2 (1 mL, 1 mg / mL) and mix at room temperature for 3 hours to allow it to react fully to form the MNP-ssDNA-biotin conjugate. Finally, remove the supernatant containing excess biotin-ssDNA-COOH by magnetic separation, and wash repeatedly with PBST buffer (1 mL) 3 times, and resuspend the conjugate in PBS buffer (1 mL) and store at 4°C.

[0056] The preparation process of MNP-phage conjugates is similar to that of MNP-ssDNA-biotin conjugates, except that the amounts of EDC and NHS used are 20 μL (10 mg / mL) and 10 μL (10 mg / mL), respectively. The conjugation process requires 3 hours at room temperature and is stored at 4°C.

[0057] (2) Phage-mediated extraction of Salmonella typhimurium DNA.

[0058] For MNP-phage-mediated S. Typhimurium DNA extraction, MNP-phage conjugates (100 μL, 250 μg / mL) were added to pre-treated chicken samples containing S. Typhimurium bacteria (100 μL) and reacted at 37°C for 15 min followed by magnetic separation. The MNP-phage-S. Typhimurium bacterial complex was placed at room temperature for 1.5 h to completely lyse and extract bacterial DNA. Finally, bacterial DNA was resuspended in ultrapure water (50 μL), stored at -20°C, and quantified using NanoPhotometer.

[0059] (3) Salmonella Typhimurium Detection Process

[0060] First, Cas12a (2 μL, 1 μM), crRNA1 (2 μL, 1 μM), crRNA2 (2 μL, 1 μM) and NE buffer 2.1 (4 μL, 1×) were mixed in a 200 μL centrifuge tube and shaken at 37°C for 15 minutes to complete the assembly of multiple crRNAs and Cas12a. The above steps were carried out simultaneously with phage-mediated DNA extraction.

[0061] Subsequently, the assembled Cas12a (9 μL), MNP-ssDNA conjugate (9 μL, 80 μg / mL), Salmonella typhimurium DNA (9 μL), and NE buffer 2.1 (3 μL, 10×) were mixed in a centrifuge tube and gently shaken at 37°C for 60 minutes. After magnetic separation, the mixture was washed three times with deionized water, and the magnetic separation substrate was retained.

[0062] Finally, PS-SA conjugate (20 μL, 100 μg / mL) was added to the centrifuge tube and mixed with the above-mentioned MNP conjugate for 10 minutes. After magnetic separation, the supernatant (15 μL) was transferred to a new centrifuge tube, and PEG-600 (15 μL) was added and mixed evenly for high SNR holographic imaging.

[0063] (4) Acoustically driven nanolens high SNR holographic imaging

[0064] Before holographic imaging, the disposable quartz cover glass was first cleaned with ethanol and deionized water, and ultrasonically treated for 15 minutes to remove surface impurities, then transferred to an oven to dry the moisture at 60°C, and finally the cover glass was cleaned and treated for 5 minutes using a vacuum plasma surface treatment system to make its surface hydrophilic to complete the preparation. During holographic imaging, a droplet sample (~10μL) was added to the surface of the cover glass and allowed to settle for 2 minutes in advance. The droplet was then evenly applied to the surface of the cover glass and coupled to the SAW-IDT chip through gel and placed directly above the CMOS sensing area. The function generator sets the sine wave generation frequency to 9.71Mhz, and the SAW waveform characteristics can be controlled by adjusting the amplitude voltage. The sine wave signal of a specific frequency is amplified by a power amplifier and connected to the SAW-IDT resonator through an SMA interface to excite the inverse piezoelectric effect to generate SAW, which is further propagated to the thin layer of liquid on the quartz cover glass to form a nano-microlens at the PS microsphere to enhance the local spatial resolution. The glass slide was adjusted to the appropriate imaging area by adjusting the micro-displacement device, and the imaging area was selected according to Poisson distribution sampling to record the original hologram (4912 pixels × 3684 pixels). In addition, each test sample (~30 μL) was measured three times (~10 μL) to capture three original holograms for calculating the relative standard deviation.

[0065] (5) UL-CNN unsupervised learning image processing algorithm training

[0066] In this embodiment, the UL-CNN unsupervised learning image processing algorithm training uses a conventionally configured portable computer (i5-11400H / GTX 1650), wherein the key software library versions called are Python 3.8.0, Deeptrack1.5.6, TensorFlow 2.10.0, Nvidia-SMI 460.89, CUDA 11.2, Cupy 8.3.0, Cudnn 7.6.5, and Tensorboard2.10.1. The dataset image and the predicted image are bundled and saved as npy format files in advance. For UL-CNN algorithm training, a holographic image with multiple target PS microspheres is selected as a training dataset image (512 pixels × 512 pixels), and a feature image of a single PS microsphere (80 pixels × 80 pixels) is defined as a training feature sample. The UL-CNN algorithm automatically performs unsupervised training according to the defined training feature samples. Among them, the number of mini-batches used for training is set to 5000 (Mini-batches, 8 samples per batch), and these feature images used for training are enhanced by affine transformation (rotation, translation, scaling, flipping, etc.), noise addition, Gaussian blur and multiplication operations to create an adaptive training data set. In addition, the key parameters set for UL-CNN algorithm training include: the number of iterations (Epoch) is 30 times, the image input batch (Batchsize) is 8, the loss function is set to the loss between the inverse transformation average prediction value and its batch average, the initial learning rate (lr0) and the cycle learning rate (lrf) are set to 0.001, and the training is set to Adam as the optimizer. The entire training time is about 3 minutes, and the system running memory is less than 1GB. At the same time, the UL-CNN algorithm automatically saves the best learning model and predicts all PS microsphere images, finally marks the predicted target, and outputs the PS microsphere quantity information to complete the image processing.

[0067] (6) Quantitative detection of Salmonella Typhimurium in chicken samples

[0068] like Figure 7As shown, the key parameters of nanolens formation and biochemical reaction in this embodiment are optimized, the number of PS microsphere signal probes in the biosensing platform is calibrated, and the nanolens formation conditions, as well as the important parameters in the bioorthogonal reaction and CRISPR-Cas12a system are systematically optimized, including PEG-600 concentration, MNP-ssDNA concentration, Cas12a to crRNA ratio, crRNA concentration and cutting time. The experimental results show that the optimal ratio concentration of PEG-600 is 50%, the optimal concentration of MNP-ssDNA is 80μg / mL, the optimal concentration ratio of Cas12a to multi-crRNAs is set to 1:2, the optimal concentration of Cas12a is 1μM, and the optimal cutting time is 60 minutes.

[0069] The results of the detection of Salmonella typhimurium under the optimized conditions are as follows Figure 8 As shown in (A), after the image processing flow, UL-CNN will mark the identified microsphere image and feedback the number of microspheres in the identified supernatant. The number of 3μmPS microsphere signal probes in the supernatant is used as the vertical coordinate, and the logarithm of the concentration of Salmonella typhimurium is used as the horizontal coordinate. The relationship between the two can be used for the quantitative detection of Salmonella typhimurium.

[0070] The experimental results show that the method of this embodiment can 7 In the range of CFU / mL, there is a good linear relationship between the number of 3 μm microspheres in the supernatant and the concentration of Salmonella typhimurium, and the linear equation is Y=196.9X+109.3 (X is the logarithmic value of the concentration of Salmonella typhimurium bacteria, Y is the number of microspheres in the supernatant), R 2 =0.99. The detection limit LOD was 38 CFU / mL (S=10.66, M=0.85).

[0071] also, Figure 8 (B) shows the results of the “gold standard” qPCR method for the detection of Salmonella Typhimurium, where the Ct value is related to the concentration of Salmonella Typhimurium at 10 3 -10 7 There is a good linear relationship within the CFU / mL concentration range (R 2 =0.99). Compared with the qPCR detection method, the detection sensitivity of this embodiment for Salmonella typhimurium is increased by about 26.3 times, and the linear range is expanded by more than 1 order of magnitude, which has the advantages of bacterial activity identification, faster detection speed, higher sensitivity, and lower cost.

[0072] also, Fig. 9 (A) and Fig. 9(B) Scanning electron microscopy (SEM) images of live Salmonella typhimurium captured by MNP-phage conjugate and PS-ssDNA-MNP complex are shown, respectively, further demonstrating the successful occurrence of phage-mediated live bacterial nucleic acid extraction and bioorthogonal reaction in this example.

[0073] As shown in Table 1, this is the result of detecting Salmonella typhimurium in chicken samples using the standard addition method. The experimental results show that the method of this embodiment has good accuracy and recovery rate for detecting Salmonella typhimurium in chicken samples.

[0074] Table 1 Results of the detection of Salmonella typhimurium in chicken samples using the standard addition method

[0075]

[0076] The experimental method used in this example is compared with the existing holographic microscope method and qPCR method. The comparison results are shown in Table 2. The comparison results show that the detection method of this example has obvious advantages. First, the detection method of this example can be completed at room temperature without complex amplification means, while the "gold standard" qPCR method requires high-temperature cyclic amplification. In addition, the detection sensitivity of the method of this example (~38 CFU / mL) is higher than that of the other two methods (traditional holographic microscope method: ~72 CFU / mL; qPCR method: 10 3 CFU / mL) were significantly improved, and the time for biochemical reactions was further shortened to 1 hour, which was due to the higher biochemical reaction efficiency of the PS microsphere signal probe with a smaller particle size and the enhanced activity of the CRISPR reaction. Finally, the detection method used in this embodiment does not require fluorescent probes and fluorescent microscopes. The imaging limit of the PS microsphere probe based on the high SNR lensless holographic microscope of the nanolens strategy is as low as 99nm, which provides the possibility for further improvement of the sensitivity of the subsequent digital detection platform.

[0077] Table 2 Performance comparison of the detection method of this embodiment with the traditional holographic microscope method and qPCR method

[0078]

[0079] The performance of the unsupervised learning holographic image processing algorithm constructed in this embodiment is compared with the existing YOLOv7 algorithm. The comparison results are shown in Table 3. The comparison results show that the detection method of this embodiment has obvious advantages. First, the unsupervised learning image processing algorithm based on geometric self-transformation constructed in this embodiment has low hardware requirements for training equipment, GPU bandwidth, etc., and does not require a large number of manually labeled data sets, which greatly reduces labor costs. Secondly, compared with supervised deep learning, the unsupervised CNN network constructed in this embodiment has greatly shortened the number of training rounds and training time requirements, and only a few minutes of short rounds of training are required to complete target detection. Finally, compared with the previous working method, the image processing algorithm constructed in this embodiment has higher target recognition accuracy and faster recognition speed, which is of great significance for improving the performance of digital detection platforms.

[0080] Table 3 Performance comparison between the image processing method in this embodiment and the image processing method in previous work

[0081]

[0082]

[0083] Example 2: Quantitative detection of Staphylococcus aureus in lettuce samples.

[0084] For the quantitative detection of Staphylococcus aureus in lettuce samples, the preparation strategies of PS-SA, MNP-ssDNA-biotin and MNP-phage conjugates were the same as those in Example 1, except that the particle size of the PS microsphere probe in Example 2 was changed from 3 μm to 1 μm.

[0085] For phage-mediated DNA extraction, the difference is that the phage needs to be replaced with the phage corresponding to Staphylococcus aureus. The operation process and extraction strategy are the same as those in Example 1, and the pretreatment method of lettuce is carried out according to the national standard method.

[0086] For the Staphylococcus aureus detection process, refer to the Salmonella Typhimurium detection process in Example 1, except that the multiple crRNAs corresponding to Salmonella Typhimurium in Example 1 need to be replaced with multiple crRNAs corresponding to Staphylococcus aureus. The specific detection process and the acoustic-driven nanolens high SNR holographic imaging steps are consistent with Example 1.

[0087] For unsupervised learning holographic image processing, the UL-CNN model trained in Example 1 will be migrated and saved to the computer, and can be used for detection and identification of PS microsphere signal probes of different particle sizes. The host computer will convert the number of identified PS microsphere probes into the concentration of the target to be measured.

[0088] According to the above operation process, the linear range of Staphylococcus aureus detection in lettuce samples is as follows Fig.10 As shown, the results show that the method of this embodiment is effective in 10 1 -10 6 In the range of CFU / mL, there is a good linear relationship between the number of 1 μm PS microspheres in the supernatant and the concentration of Staphylococcus aureus. The linear equation is Y=343.3X+125.4 (X is the logarithmic value of the concentration of Staphylococcus aureus, Y is the number of microspheres in the supernatant), R 2 =0.98.

[0089] Example 3: Quantitative detection of interleukin-6 (IL-6) in serum samples

[0090] (1) Preparation of PS-IL-6Ab1 and MNP-IL-6Ab2 conjugates

[0091] Take 2 mg of carboxyl-modified PS (1 μm, 100 mg / mL), wash twice with MES buffer (pH=6.0), resuspend with MES buffer, add 10 μL EDC (10 mg / mL) and 10 μL NHS (10 mg / mL), mix well and activate at room temperature for 15 minutes. After activation, wash twice with PBS buffer, resuspend with PBS buffer, add 0.2 mg IL-6Ab1, react at room temperature for 3 hours; after the reaction, block with 1% BSA solution for 30 minutes, wash twice with PBST buffer after blocking, resuspend the PS-IL-6Ab1 conjugate with PBS buffer, and store at 4°C for use.

[0092] Take 2 mg of carboxyl-modified MNP (150 nm, 10 mg / mL), wash twice with MES buffer (pH = 6.0), resuspend with MES buffer, add 10 μL EDC (10 mg / mL) and 15 μL NHS (10 mg / mL), mix well and activate at room temperature for 15 minutes. After activation, wash twice with PBS buffer, resuspend with PBS buffer, add 0.2 mg IL-6Ab2, react at room temperature for 2-4 hours, and resuspend the MNP-IL-6Ab2 conjugate with PBS buffer and store at 4°C for use.

[0093] (2) Double antibody sandwich immunoreaction and magnetic separation

[0094] The PS-IL-6Ab1 conjugate, MNP-IL-6Ab2 conjugate and serum sample containing IL-6 were added to a 1.5 mL centrifuge tube at the same time for immune reaction at 37°C for 10-30 minutes. The centrifuge tube was then placed in a magnetic separation rack for 1 minute to complete magnetic separation, and washed three times with PBST buffer to remove the immune complex after magnetic separation. The unreacted PS-IL-6Ab1 conjugate in the supernatant was retained and stored at 4°C for high SNR holographic imaging.

[0095] For unsupervised learning holographic image processing, the image processing process of Example 3 is basically the same as that of Example 1. The UL-CNN model trained in Example 1 will be migrated and saved to the computer, and can be used for detection and identification of PS microsphere signal probes of different particle sizes. The host computer will convert the number of identified PS microsphere probes into the concentration of the target to be measured.

[0096] According to the above operation procedure, the linear range of IL-6 detection in serum samples is as follows Fig.11 As shown, the results show that the method of this embodiment is effective in 10 1 -10 6 In the range of pg / mL, there is a good linear relationship between the number of 1 μm PS microspheres in the supernatant and the IL-6 concentration. The linear equation is Y=-370.3X+2511 (X is the logarithm of the IL-6 concentration, Y is the number of microspheres in the supernatant), R 2 =0.99.

[0097] Example 4: Quantitative detection of chloramphenicol (CAP) in milk samples.

[0098] For the quantitative detection of CAP in milk samples, the preparation method of PS-CAP Ab and MNP-CAP Ag conjugates was basically the same as in Example 3, except that Ab1 and Ab2 corresponding to IL-6 were replaced with Ab and Ag corresponding to CAP, respectively, and the particle sizes of the PS microsphere signal probe and the MNP magnetic separation carrier were 1 μm and 150 nm, respectively.

[0099] For the detection of CAP, the double antibody sandwich immune reaction in Example 3 was changed to a competitive immune reaction, and the detection process was basically the same as that in Example 3. After the immune reaction was completed, the immune complex after magnetic separation was removed, and the unreacted PS-CAP Ab conjugate in the supernatant was retained and stored at 4°C for high SNR holographic imaging.

[0100] For unsupervised learning holographic image processing, the image processing process of Example 4 is similar to that of Example 1. The UL-CNN model trained in Example 1 will be migrated and saved to the computer, and can be used for detection and identification of PS microsphere signal probes of different particle sizes. The host computer will convert the number of identified PS microsphere probes into the concentration of the target to be measured.

[0101] According to the above operation process, the linear range of CAP detection in milk samples is as follows Fig.12 As shown, the results show that the method of this embodiment is effective in 10 1.5 -10 6 In the range of pg / mL, there is a good linear relationship between the number of 1 μm PS microspheres in the supernatant and the concentration of CAP, and the linear equation is Y=315.9X–31.9 (X is the logarithm of CAP concentration, Y is the number of microspheres in the supernatant), R 2 =0.97.

[0102] Embodiment 5:

[0103] This embodiment provides a digital analysis system based on nanolenses, such as Figure 2 As shown, it includes: an adjustable frequency signal generator (0-100Mhz), a signal power amplifier, a SAW-IDT resonator chip, a holographic imaging device, a target recognition module, and a result analysis module; the holographic imaging device includes a 3D printing optical machine housing (black light-proof resin), a partially coherent illumination light source (620nm), two micro-displacement controllers (x, z axis, 10-15mm), a DC adjustable power supply (0-24V), a CMOS sensor and other key components. Among them, the adjustable frequency signal generator and the signal power amplifier can excite a sine wave signal of a specific frequency and transmit it to the SAW-IDT chip. The piezoelectric substrate of the chip will generate SAW under high-frequency signals and propagate on the surface, and finally couple and propagate to the thin layer of liquid on the disposable cover glass, causing the thin layer of liquid at the PS microsphere to deform and evaporate water, resulting in a nanolens phenomenon. In addition, incoherent LED light is modulated by an optical pinhole to become partially coherent light, which is used for high SNR holographic imaging illumination. The lightweight UL-CNN unsupervised learning algorithm recognizes and decodes the nanolens hologram captured by the host computer, and calculates the quantity information of the PS microsphere probes into the concentration information of the target to be measured through the standard curve, thereby achieving quantitative detection.

[0104] See also Figure 4 The above is a schematic diagram of the SAW-IDT resonator chip structure in this embodiment. Figure 4 (A) is a 5-inch photolithography mask used in the production of SAW-IDT resonator chips. Figure 4 (B) is a physical picture of the SAW-IDT resonator chip. Figure 4(C) is a structural characterization diagram of the interdigital electrodes of the SAW-IDT resonator chip under an optical microscope. In this embodiment, the piezoelectric substrate material for making the SAW-IDT resonator chip is lithium niobate crystal with a thickness of 1 mm and a cutting direction of 128°yx. The material of the interdigital plated electrode is double-layer Ti / Au, the number of electrode pairs is 20-40 pairs, the designed excitation frequency is 9.71 MHz, and the wavelength is 411 μm. It can be clearly observed under an optical microscope that the metal electrodes of the SAW-IDT resonator are arranged in a regular staggered distribution, and the morphology is complete and clear.

[0105] See also Figure 5 The above is a comparison chart of the field of view area of ​​the holographic microscope built in this embodiment and the traditional optical microscope. The FOV area of ​​the high SNR holographic microscope built in this embodiment has obvious advantages over the traditional optical microscope, wherein the effective FOV area is 28.275mm 2 ,like Figure 5 As shown in (A), compared with an optical microscope equipped with 4×, 10×, and 20× objectives, the FOV area is increased by approximately 7 times, 39 times, and 221 times, respectively. Figure 5 (C)-(D) It is worth noting that the larger FOV area indicates that under the same imaging conditions, the holographic microscope constructed in this embodiment can capture a larger number of PS microspheres. In addition, the combination of a new optical strategy can achieve high SNR holographic imaging of smaller PS microsphere signal probes, thereby further improving the accuracy and sensitivity of the digital detection platform.

[0106] The target recognition module of this embodiment uses the UL-CNN unsupervised learning algorithm to process the holographic image. The unsupervised learning holographic image processing flow chart based on geometric self-transformation is as follows: Figure 3As shown in the figure, the inherent geometric characteristics of the detected object are used to realize target detection, where the lightweight feature extraction CNN includes convolutional layers activated by ReLU function (3, 3×3×32), pooling layer (1, 2×2), additional convolutional layers activated by ReLU function (8, 3×3×32) and convolutional layer without activation function (1, 1×1×3). The target microsphere image (n×m pixel value, c color channels) is input into CNN after geometric transformation such as rotation, translation and affine. Each geometric transformation image will generate a vector field map (n / 2×m / 2, 2) and a weight map (n / 2×m / 2, 1), which represent the distance from the pixel itself to the target object and the prediction contribution, respectively. The predicted target position is obtained by multiplying and summing the two feature tensors, and the intersection of the symmetry axes before and after the geometric transformation is the predicted target center. Each prediction result can be used for retraining of UL-CNN. Finally, UL-CNN will mark the target microspheres detected and identified with an “×”, complete the probe counting task, and associate it with the content of the target object, and output the content of the target object to be tested.

[0107] See also Figure 6 The above is a diagram of the imaging limit test of the high SNR holographic microscope built in this embodiment and the verification of the algorithm recognition accuracy. In order to evaluate the imaging limit of the high SNR holographic microscope built in this embodiment, the imaging limit of 5 PS microspheres with different particle sizes (99nm, 510nm, 907nm, 3μm and 5.98μm) was tested.

[0108] The experimental results show that when the SAW is turned on, the high SNR holographic microscope can produce a local spatial resolution enhancement effect, thereby breaking through the resolution limit and achieving high SNR holographic imaging. However, when the SAW is turned off, the local spatial resolution enhancement effect disappears, and the high SNR holographic microscope cannot capture the holographic image of PS microspheres with a particle size less than 5.98μm. Figure 6 (A) shown.

[0109] The captured small-size PS microsphere hologram based on the local spatial resolution enhancement effect has a clear morphology, and the imaging effect is basically not affected by the twin image generated during the holographic imaging process. Therefore, there is no need for a holographic reconstruction step, and the target detection can be performed directly through the algorithm. In addition, compared with other target detection algorithms, such as ImageJ software and machine vision (CV) image processing algorithms, the UL-CNN network constructed in this embodiment can achieve efficient and accurate target detection for holographic images of different numbers and SNRs, especially for the processing of low SNR holographic images. It has obvious advantages, such as Figure 6 (B) shown.

[0110] Some steps in the embodiments of the present invention may be implemented using software, and the corresponding software program may be stored in a readable storage medium, such as a CD or a hard disk.

[0111] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A digital analysis method based on nanolenses, characterized in that: The method comprises: Step 1: For the target bacteria to be detected, the bacteriophage corresponding to the target bacteria to be detected is coupled to the surface of the nanomagnetic particle MNP to form an MNP-bacteriophage conjugate; for the non-nucleic acid target to be detected, the biological recognition element that recognizes the non-nucleic acid target is coupled to the surface of the nanomagnetic particle to form an MNP-biological recognition element conjugate; Step 2: adding the MNP-bacteriophage conjugate to the target bacteria to be tested for reaction, the bacteriophage specifically captures the active target bacteria, and then magnetically separates and retains the MNP-bacteriophage-target bacteria complex, and after standing to lyse and release DNA, collects the reaction solution for later use; Step 3: For the target bacteria to be tested, the target bacterial DNA extracted in step 2 is mixed with the Cas12a enzyme assembled with multiple crRNAs to enhance its trans-cutting ability with activity, and then the reaction solution of activating the Cas12a enzyme is added to the MNP-ssDNA-biotin conjugate to trans-cut the ssDNA probe to complete the CRISPR reaction, and finally the microsphere signal probe-streptavidin conjugate probe is added to complete the bioorthogonal reaction and then magnetically separated, and the supernatant containing the microsphere signal probe is retained for holographic imaging; For the non-nucleic acid target to be detected, the microsphere signal probe-antibody conjugate, the MNP-antigen conjugate, and the non-nucleic acid target to be detected are mixed for immune reaction, and the immune complex and unreacted nanomagnetic particles MNP are removed after magnetic separation, and the supernatant containing the microsphere signal probe is retained for holographic imaging; Step 4: After mixing the supernatant containing the microsphere signal probe with PEG-600 in proportion, drip it onto the plasma-cleaned cover glass and spin-coat it, and place it in the sensing area of ​​the holographic microscope for high signal-to-noise ratio holographic imaging. During the holographic imaging process, a function signal generator and a power amplifier are used to emit a signal wave of a specific frequency to excite the piezoelectric effect of the SAW-IDT resonator chip and generate a surface acoustic wave, which propagates and couples to the cover glass to cause the thin layer of liquid on the surface to evaporate and deform, forming a nanolens phenomenon to increase the scattering cross section at the probe imaging location, thereby improving the signal-to-noise ratio of the probe holographic imaging; Step 5: Use the trained unsupervised learning target detection algorithm to identify the holographic image and obtain the number of microspheres; Step 6: Calculate the content of the target to be detected by using the standard curve of the relationship between the number of microspheres and the content of the target to be detected, so as to achieve quantitative detection of the target.

2. The digital analysis method according to claim 1, characterized in that: The microspheres include: polystyrene microspheres, silicon dioxide microspheres, polybutadiene microspheres, and polyisoprene microspheres.

3. The digital analysis method according to claim 1, characterized in that: The targets to be detected include viruses, bacteria, RNA molecules, biomarkers, antibiotic molecules, pesticide molecules, veterinary drug molecules, and biological toxins.

4. The digital analysis method according to claim 1, characterized in that: The particle size of the microspheres ranges from 0.1 to 50 μm.

5. The digital analysis method according to claim 1, characterized in that: The biological recognition elements include: complete antigen / antibody, capture antibody / detection antibody, DNA capture probe / DNA detection probe.

6. The digital analysis method according to claim 1, characterized in that: The excitation frequency range of the SAW-IDT resonator chip is 0-100 MHz, and the waveform is one of a sine wave, a triangle wave, a sawtooth wave, and a square wave.

7. The digital analysis method according to claim 1, characterized in that: The immune reaction includes competitive immune reaction, double antibody sandwich immune reaction or DNA hybridization reaction.

8. A digital analysis system based on nanolenses, characterized in that: include: Signal generator, power amplifier, SAW-IDT resonator chip, holographic imaging device, target recognition module, result analysis module; The signal generator and the power amplifier excite a sine wave signal of a specific frequency and transmit it to the SAW-IDT resonator chip, and the piezoelectric substrate of the SAW-IDT resonator chip generates a surface acoustic wave, which is finally coupled and propagated to a thin layer of liquid to be tested on a disposable cover glass, wherein the liquid to be tested is the supernatant containing the microsphere signal probe according to any one of claims 1 to 7; The holographic imaging device is used to perform holographic imaging on the supernatant containing the microsphere signal probe; The target recognition module is configured to use a trained target detection algorithm to recognize the holographic image and obtain the number of microspheres; The result analysis module is configured to calculate the content of the target to be detected by using a standard curve of the relationship between the number of microspheres and the content of the target to be detected, thereby achieving quantitative detection of the target.

9. The digital analysis system according to claim 8, characterized in that: The piezoelectric substrate material of the SAW-IDT resonator chip includes: lithium niobate, lithium tantalate, and aluminum nitride.

10. The digital analysis system according to claim 8, characterized in that: The holographic imaging device includes: a CMOS camera, a CCD camera, and a smart phone image sensor.

Images