Preparation method of water-soluble CsPbBr3 perovskite quantum dot fluorescence sensing array and application of water-soluble CsPbBr3 perovskite quantum dot fluorescence sensing array in rapid detection and inactivation of food-borne pathogenic bacteria

Abstract

The invention discloses a preparation method of a water-soluble CsPbBr3 perovskite quantum dot fluorescence sensing array and application of the water-soluble CsPbBr3 perovskite quantum dot fluorescence sensing array in rapid detection and inactivation of food-borne pathogenic bacteria. The sensor array is composed of three perovskite quantum dots (respectively named as PQD1, PQD2 and PQD3) with different fluorescence characteristics, wherein the perovskite quantum dots are synthesized at room temperature. The feasibility of constructing the sensing platform is verified through different fluorescence responses after the sensing platform is combined with five food-borne pathogenic bacteria (escherichia coli, staphylococcus aureus, salmonella typhimurium, listeria monocytogenes and pseudomonas aeruginosa). The relative signal intensity change is further analyzed through a machine learning algorithm, and rapid recognition and identification of pathogenic bacteria are achieved. In addition, the sensing array shows high practicability and accuracy in actual sample detection, and pathogenic bacteria can be effectively inactivated after 30 minutes of detection. The invention provides a novel efficient and reliable technical solution for food safety detection.

Description

Technical Field

[0001] The present invention belongs to the technical field of microbial detection, and specifically relates to a method for preparing a water-soluble CsPbBr3 perovskite quantum dot fluorescence sensor array based on a machine learning algorithm and its application in the rapid detection and inactivation of foodborne pathogens. Background Art

[0002] Food safety is closely related to human health and social stability. The presence of foodborne pathogens is one of the main causes of food safety accidents. Foodborne pathogens refer to pathogenic microorganisms that enter the human body through ingestion of food and cause infection or poisoning, such as Escherichia coli, Staphylococcus aureus, and Salmonella typhi. The presence of these pathogens in food is often difficult to detect, but once ingested, they may cause serious illness or even endanger life. Traditional methods for detecting foodborne pathogens mainly include culture, polymerase chain reaction (PCR), and immunological detection methods such as enzyme-linked immunosorbent assay (ELISA). Although these technologies provide high specificity and sensitivity, they require a lot of sample preparation as well as expensive equipment and well-trained personnel for detection. Therefore, it is of great practical significance to develop a new method for detecting foodborne pathogens that is rapid, sensitive, accurate, and easy to operate.

[0003] As an emerging detection technology, fluorescent sensor arrays have shown great potential in the field of biosensing due to their unique optical properties and high sensitivity. It is a detection method that uses fluorescent labeling technology combined with a highly sensitive optical detection system. Fluorescent sensors generate detectable fluorescent signals through the combination of fluorescent markers and target molecules. Sensor arrays can detect multiple target molecules simultaneously by integrating multiple fluorescent sensors with different specificities. This method not only improves the sensitivity and selectivity of detection, but also significantly shortens the detection time, making rapid screening possible.

[0004] In recent years, with the rapid development of nanotechnology, quantum dots, as a new type of nanomaterial, have shown broad application prospects in the biomedical field. Perovskite quantum dots (PQDs) are semiconductor nanomaterials with an ABX3 structure, where A is usually a large organic or inorganic cation, B is a smaller metal cation, and X is a halogen anion. The uniqueness of PQDs lies in their electronic structure and band gap that can be adjusted by chemical composition, which provides a broad design space for their application in the optoelectronic field. In particular, the high quantum yield, narrow emission peak width, and luminescence wavelength that can be regulated by size and composition exhibited by PQDs make it possible to use them in photoelectric conversion devices such as light-emitting diodes (LEDs) and solar cells. In the field of LEDs, the high color purity and tunable spectral characteristics of PQDs make it possible to achieve richer and more accurate color reproduction, which is of great significance for the development of high-definition display screens and lighting technologies. In the field of solar energy, the high light absorption coefficient and long carrier diffusion length of PQDs give it a significant advantage in improving the photoelectric conversion efficiency of solar cells. In addition, the solution processing characteristics and cost-effectiveness of PQDs also facilitate large-scale production and application. However, although the application of PQDs in solar cells, LEDs and other fields has made significant progress, its application research in biosensing and detection is relatively limited. This is mainly limited by its poor biocompatibility and instability in polar solutions. All-inorganic lead halide cesium CsPbX3 (X = Cl, Br, I) PQDs can achieve full-spectrum luminescence and high fluorescence quantum efficiency in the visible spectrum due to their similar diameters, which makes them have great application potential in high-sensitivity, visual rapid detection.

[0005] Machine learning is a technology that enables computer systems to learn and improve using data through algorithms. In the field of biosensing, machine learning algorithms are used to analyze complex data generated by sensors to improve the accuracy and efficiency of detection. As a highly sensitive detection tool, fluorescent sensor arrays can generate a large amount of fluorescent signal data. Machine learning algorithms can extract features from these data and establish a quantitative relationship model between pathogen concentration and fluorescent signal. This process usually involves multiple steps such as data preprocessing, feature selection, model training and verification to ensure the accuracy and generalization ability of the model. In addition, machine learning algorithms have shown their powerful ability in classifying and identifying the fluorescent features of different pathogens. By building a multi-classification model, machine learning algorithms can identify and distinguish the unique fluorescent features of multiple pathogens and realize the simultaneous detection of multiple pathogens. This multi-task learning method not only improves the efficiency of detection, but also enhances the comprehensiveness and accuracy of detection. In practical applications, machine learning algorithms can also be combined with mobile devices such as smartphones to achieve real-time monitoring and analysis of fluorescent signal data through corresponding applications. This integrated detection platform provides users with a portable, fast and accurate detection method, greatly expanding the application scope of biosensor technology. Summary of the invention

[0006] The object of the present invention is to provide a method for preparing a water-soluble CsPbBr3 perovskite quantum dot fluorescence sensor array and its application in the rapid detection and inactivation of foodborne pathogens.

[0007] To achieve the above-mentioned purpose and other related purposes, the technical solution provided by the present invention is: a method for preparing water-soluble CsPbBr3 perovskite quantum dots, comprising the following steps: Step 1: Prepare the precursor solution Dispersing cesium bromide in hydrobromic acid to obtain a cesium bromide mother solution, dispersing lead bromide in hydrobromic acid to obtain a lead bromide mother solution; mixing the cesium bromide mother solution and the lead bromide mother solution, and then adding N, N-dimethylformamide to the mixture; then adding perfluorooctyltriethoxysilane, and reacting to obtain a precursor solution; Step 2: Preparation of composite emulsion The invention relates to the preparation of at least one of the following composite emulsions: Preparation of PQD1 composite emulsion: adding oleylamine and water to the precursor solution, and obtaining a composite emulsion containing PQD1 after reaction; Preparation of PQD2 composite emulsion: adding oleylamine, hydrochloric acid and water to the precursor solution, and obtaining a composite emulsion containing PQD2 after reaction; Preparation of PQD3 composite emulsion: adding oleylamine, hydrochloric acid and water to the precursor solution, and obtaining a composite emulsion containing PQD3 after reaction, wherein the volume of hydrochloric acid used in step 4 is 0.8-1.5 times the volume of hydrochloric acid used in step 3; Step 3: Centrifugation purification: At least one of the composite emulsion containing PQD1, the composite emulsion containing PQD2 and the composite emulsion containing PQD3 is centrifuged to discard the precipitate and take the supernatant; the supernatant is centrifuged again to remove the unreacted precursor, and the precipitate is redispersed in water to obtain an aqueous solution containing PQD1, an aqueous solution containing PQD2 and an aqueous solution containing PQD3.

[0008] The preferred technical solution is: in the mixture, the mass ratio of cesium bromide mother liquor to lead bromide is 0.8-1.2:1, the first centrifugation rate is 5-8000 rpm, and the second centrifugation rate is 10000-15000 rpm.

[0009] To achieve the above-mentioned purpose and other related purposes, the technical solution provided by the present invention is: a method for detecting foodborne pathogens based on a water-soluble CsPbBr3 perovskite quantum dot fluorescence sensor array based on a machine learning algorithm, comprising the following steps: S1. Inoculate the bacteria onto LB solid medium and culture at 35-38°C overnight; then select a single colony from the cultured colonies, wash it with ultrapure water and resuspend it; then measure the OD value of the bacterial resuspension; according to the pre-determined CFU-OD 600 For the standard curve, the bacterial resuspension was diluted with ultrapure water to different concentrations: 1.0×10 3 CFU / mL, 1.0×10 4 CFU / mL, 1.0×10 5 CFU / mL, 1.0×10 6 CFU / mL, 1.0×10 7 CFU / mL; Reserve; S2, the concentration is 1.0×10 3 CFU / mL of bacterial solution, respectively, combined with the aqueous solution containing PQD1, the aqueous solution containing PQD2, and the aqueous solution containing PQD3 as described in claim 1 or 2, and observed the changes in the fluorescence intensity of the material and the differences in the changes in fluorescence intensity after combining with different bacterial species to verify the feasibility of constructing a fluorescent sensor array; S3, dropping bacterial resuspensions of different concentrations and types into the fluorescent sensor array composed of three perovskite quantum dot arrangements, and incubating under standard conditions; then, obtaining the relative signal intensity change value of the image through the Color Grab app on the smartphone; S4. After collecting the relative signal intensity change values, the pattern recognition method is used to analyze the relative signal intensity change values; in the training stage, 70% of the available data is divided into a training set through a random sampling procedure, and the remaining 30% is used as a test set; then, the data set is analyzed using the built-in machine learning algorithm of MATLAB to achieve classification and identification of bacteria; the algorithms used include: decision tree, support vector machine (SVM), k-nearest neighbor (KNN), linear discriminant analysis (LDA) and naive Bayes method.

[0010] A preferred technical solution is that the bacterial species include at least one of the following bacterial species: Escherichia coli, Staphylococcus aureus, Salmonella typhimurium, Listeria monocytogenes, and Pseudomonas aeruginosa.

[0011] The preferred technical solution is: the relative signal intensity change value is calculated according to the following formula: ΔRGB = (R0− R) / R0+ (G0− G) / G0+ (B0− B) / B0, where R0, G0, B0 are the initial red, green and blue values ​​of the image when no bacteria are added, and R, G and B are the red, green and blue values ​​of the image extracted after adding bacteria.

[0012] To achieve the above-mentioned purpose and other related purposes, the technical solution provided by the present invention is: an application of a water-soluble CsPbBr3 perovskite quantum dot fluorescence sensor array based on a machine learning algorithm in bacterial inactivation, comprising the following steps: a. Observe the sterilization effect by plate counting: a bacterial suspension with an OD value of 0.2 was placed in a sterile centrifuge tube, and the bacterial suspension was randomly divided into 4 groups: control group, PQD1 group, PQD2 group and PQD3 group; a PQDs solution with a concentration of 12.5 mg / mL was added to each centrifuge tube, and incubated at room temperature for 30 minutes; after the incubation, the mixed solution was centrifuged at 3000-5000 rpm for 1-5 minutes, and the precipitate was washed several times to finally obtain a bacterial resuspension; then it was diluted 10,000 times, 100 μL was evenly spread on the plate, and cultured at 37°C until a single colony grew, and the colonies were counted; b. Observation of sterilization effect by live-dead cell staining: The bactericidal performance of PQDs was observed using a calcein AM / PI staining kit; in brief, after incubation with the bacterial suspension for 30 min, the sample was washed several times with ultrapure water and a bacterial resuspension was prepared. Subsequently, calcein-AM (1.5 μL / mL) and PI (1.5 μL / mL) were added to the sample, and then incubated at 37 °C in the dark for 45 min. The prepared sample was observed under an inverted fluorescence microscope.

[0013] The preferred technical solution is: after incubating with the bacterial suspension for 30 minutes, the sample is washed with ultrapure water several times to prepare a bacterial resuspension. Subsequently, calcein-AM 1.5 μL / mL and PI 1.5 μL / mL are added to the sample, and then incubated at 37°C in the dark for 45 minutes.

[0014] The preferred technical solution is that the bacterial species of the bacterial suspension include at least one of the following species: Escherichia coli, Staphylococcus aureus, Salmonella typhimurium, Listeria monocytogenes, and Pseudomonas aeruginosa.

[0015] Due to the application of the above technical solution, the present invention has the following advantages compared with the prior art: 1. The present invention is different from the traditional synthesis method of perovskite quantum dots, that is, the preparation of quantum dots is carried out at high temperature and in an inert gas environment. Instead, the surface modification strategy is adopted to synthesize CsPbBr3PQDs with three different fluorescent properties through simple steps at room temperature to form a fluorescent sensing array, which greatly reduces the production cost and has a good economic effect.

[0016] 2. The fluorescent sensor array constructed by the present invention can accurately identify and recognize the target bacteria after incubation with the target bacteria for 5 minutes. Compared with traditional bacterial detection technology, it significantly shortens the detection time and simplifies the operation process.

[0017] 3. The present invention applies machine learning algorithms to analyze the relative signal intensity changes (ΔRGB) after the fluorescent sensor array interacts with bacteria, thereby improving the efficiency of detection and enhancing the comprehensiveness and accuracy of detection.

[0018] 4. The fluorescent sensor array constructed by the present invention has bactericidal properties and can inactivate pathogens within 30 minutes after detection. It has important practical significance and broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 TEM images and particle size statistics of three types of perovskite quantum dots in Example 1.

[0020] Figure 2 These are the XPS graphs, FT-IR graphs and Zeta potential graphs of the three perovskite quantum dots in Example 1.

[0021] Figure 3 This is the fluorescence emission spectra of the three perovskite quantum dots in Example 1 at an excitation wavelength of 280-380 nm.

[0022] Figure 4 This is a graph showing the changes in fluorescence intensity after the three types of perovskite quantum dots in Example 2 are combined with five foodborne pathogens of different concentrations.

[0023] Figure 5 The three perovskite quantum dots in Example 3 are respectively and the concentration is 1.0 × 10 3 Relative signal intensity change (ΔRGB) graph of five foodborne pathogens after interaction at CFU / mL; performance indicator graph of five machine learning algorithms; confusion matrix graph and parallel coordinate graph of SVM algorithm.

[0024] Figure 6 The characteristic space map of the five bacteria drawn by machine learning in Example 3; the standardized score map for identifying the five bacterial samples; the ring cluster heat map of the five bacteria; and the radar map of the five bacteria. 3 The test was performed at a CFU / mL concentration.

[0025] Figure 7 For different concentrations of Escherichia coli in Example 4 ( E. coli ) of a typical score plot; different concentrations of Escherichia coli ( E. coli ) and factor 1; Different concentrations of Escherichia coli ( E. coli ) of the circular cluster heat map; different concentrations of Escherichia coli ( E. coli ). Escherichia coli ( E. coli ) are both within 1.0 × 10 3 CFU / mL -1.0 ×10 7 The test was performed at a CFU / mL concentration.

[0026] Figure 8 For different concentrations of Staphylococcus aureus in Example 4 ( S. aureus ) of typical scores; different concentrations of Staphylococcus aureus ( S. aureus ) and factor 1; Different concentrations of Staphylococcus aureus ( S. aureus ) of the circular cluster heat map; different concentrations of Staphylococcus aureus ( S. aureus ). Staphylococcus aureus ( S. aureus ) are both within 1.0 × 10 3 CFU / mL -1.0 × 10 7 The test was performed at a CFU / mL concentration.

[0027] Fig. 9 For different concentrations of Salmonella typhimurium in Example 4 ( S. typhimurium ) of different concentrations of Salmonella typhimurium ( S. typhimurium ) and factor 1; Different concentrations of Salmonella typhimurium ( S. typhimurium ) of the circular cluster heat map; different concentrations of Salmonella typhimurium ( S. typhimurium ) radar chart of Salmonella Typhimurium ( S. typhimurium ) are both within 1.0 × 10 3 CFU / mL -1.0 × 10 7 The test was performed at a CFU / mL concentration.

[0028] Fig.10 The different concentrations of Listeria monocytogenes in Example 4 ( L. monocytogenes ) of different concentrations of Listeria monocytogenes ( L. monocytogenes ) and factor 1; Different concentrations of Listeria monocytogenes ( L. monocytogenes ) of the circular cluster heat map; different concentrations of Listeria monocytogenes ( L. monocytogenes ). Listeria monocytogenes ( L. monocytogenes ) are both within 1.0 × 10 3 CFU / mL-1.0 × 10 7 The test was performed at a CFU / mL concentration.

[0029] Fig.11 For different concentrations of Pseudomonas aeruginosa in Example 4 ( P. aeruginosa ) of different concentrations of Pseudomonas aeruginosa ( P. aeruginosa ) and factor 1; Different concentrations of Pseudomonas aeruginosa ( P. aeruginosa ) of the circular cluster heat map; different concentrations of Pseudomonas aeruginosa ( P. aeruginosa ). Pseudomonas aeruginosa ( P. aeruginosa ) are both within 1.0 × 10 3 CFU / mL -1.0 × 10 7 The test was performed at a CFU / mL concentration.

[0030] Fig.12 These are the detection limits of the fluorescent sensor array in Example 4 for five foodborne pathogens.

[0031] Fig.13 The fluorescence sensing array in Example 5 distinguishes two groups of binary mixed bacteria ( L. monocytogenes and E. coli Mixed group, S. aureus and S. typhimurium Mixed group) and a three-component mixed bacteria group ( L. monocytogenes , S. aureus and P. aeruginosa The total concentration of mixed bacteria was 1.0 × 10 3 The test was performed at CFU / mL.

[0032] Fig.14 The fluorescence sensor array in Example 6 distinguishes seven interfering substances (Na + , K + HCl - , glucose, chitosan, L-arginine and L-tyrosine) and five target pathogens; the relationship between the relative signal intensity change (ΔRGB) and the number of repeated tests (n=10); the change of ΔRGB value detected by fluorescence sensor array during 15 days of storage.

[0033] Fig.15 The typical scoring diagram of the fluorescence sensor array in Example 7 for distinguishing a single bacterial unknown sample; the typical scoring diagram for distinguishing a mixed unknown sample of two strains; and the typical scoring diagram for distinguishing a mixed unknown sample of three strains. The total concentration of each single bacterium and mixed bacteria was 1.0 × 10 3 The test was performed at CFU / mL.

[0034] Fig.16 This is a typical scoring diagram of the fluorescence sensor array in Example 8 for distinguishing unknown samples of single bacteria in real sample tap water. Each bacterium is scored at 1.0 × 10 3 The test was performed at a CFU / mL concentration.

[0035] Fig.17 For Example 9 E. coli and S. aureus Plate images after interaction with PQD1, PQD2, and PQD3; Quantitative analysis of the antibacterial effect of PQDs observed using the plate counting method; E. coli and S. aureus Scanning electron microscopy (SEM) images of morphological changes after interaction with PQD1; E. coli and S. aureus Calcein-AM / PI staining images. Statistical analysis was performed using one-way ANOVA and Tukey post hoc test. ***P<0.001.

[0036] Fig.18 This is the flow chart of the decision tree algorithm.

[0037] Fig.19 This is the flow chart of the support vector machine (SVM) algorithm.

[0038] Fig. 20 This is the flow chart of the K-nearest neighbor (KNN) algorithm.

[0039] Fig.21 This is the flowchart of the linear discriminant analysis (LDA) algorithm.

[0040] Fig. 22This is the flow chart of the Naive Bayes algorithm. DETAILED DESCRIPTION

[0041] The following is a description of the implementation of the present invention by means of specific embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification.

[0042] See also Figure 1-22 . It should be noted that the structures, proportions, sizes, etc. illustrated in the drawings in this specification are only used to match the contents disclosed in the specification for people familiar with this technology to understand and read, and are not used to limit the limiting conditions for the implementation of the present invention, so they have no substantial technical significance. Any modification of the structure, change in the proportional relationship, or adjustment of the size should still fall within the scope of the technical content disclosed by the present invention without affecting the effects and purposes that can be achieved by the present invention. At the same time, the terms such as "upper", "lower", "left", "right", "middle" and "one" quoted in this specification are only for the convenience of description, and are not used to limit the scope of the implementation of the present invention. Changes or adjustments in their relative relationships should also be regarded as the scope of the implementation of the present invention without substantially changing the technical content.

[0043] Unless otherwise specified, the reagents or materials described in the following examples are commercially available.

[0044] Example 1: Preparation and characterization of three water-soluble CsPbBr3 perovskite quantum dots with different fluorescence properties (1) Preparation of water-soluble CsPbBr3 perovskite quantum dots: 0.2 g of cesium bromide (CsBr) and 0.2 g of lead bromide (PbBr2) were dissolved in 1 mL of hydrobromic acid (HBr) solution as mother liquors; 0.125 mL of CsBr mother liquor and 0.125 mL of PbBr2 mother liquor were mixed, and 1 mL of N, N-dimethylformamide (DMF) was added to the mixture to prepare a clear solution. Next, 0.25 mL of perfluorooctyltriethoxysilane (C6F 13CH2CH2Si(OCH2CH3)3, PFOTES) was added to the clear solution, and the solution was stirred in an ice-water bath for about 1 hour until the emulsion solution turned green. 0.25mL of oleylamine (OLA) and 60mL of water were added to the above green emulsion, and incubated at room temperature for about 2 hours to obtain a composite emulsion containing PQD1; 0.25mL of oleylamine (OLA), 0.25mL of concentrated hydrochloric acid (HCl), and 60mL of water were added to the above green emulsion, and incubated at room temperature for about 2 hours to obtain a composite emulsion containing PQD2; 0.25mL of oleylamine (OLA), 0.5mL of concentrated hydrochloric acid (HCl), and 60mL of water were added to the above green emulsion, and incubated at room temperature for about 2 hours to obtain a composite emulsion containing PQD3. The composite emulsion was then centrifuged at 6000rpm for 6 minutes, the precipitate was discarded, and the supernatant was taken out as a PQDs aqueous solution. The supernatant can be further purified by high-speed centrifugation at 12,000 rpm for 15 min twice to remove unreacted precursors, and the precipitate can be redispersed in water to obtain an aqueous solution containing PQD1, PQD2 and PQD3.

[0045] (2) Characterization of synthesized PQD1, PQD2, and PQD3: The morphology of PQDs was observed by transmission electron microscopy (TEM). Figure 1 As shown, the average sizes of PQD1, PQD2 and PQD3 are 3.36nm, 2.92nm and 3.41nm, respectively. The XPS spectrum used to identify and characterize the elemental composition and chemical bonds of PQDs shows the main peaks of C1s, N1s, O1s, F1s, Si2p, Cs3d, Pb4f, Br3d ( Figure 2 AC) In particular, the XPS spectra of PQD2 and PQD3 also showed a main peak of Cl2p, which indicated that the halogen ion exchange reaction was successfully carried out during the synthesis. At the same time, the molecular structure, chemical bonds and functional groups of the products were characterized by Fourier transform infrared spectroscopy (FTIR). Figure 2 As shown in D, at 1468cm -1 The absorption peak at 1578cm indicates the presence of CF bond. -1 The peaks at 2850cm are attributed to the stretching vibration of NH bonds. -1 and 2918cm -1 The peak at is due to the symmetric stretching vibration and antisymmetric stretching vibration of -CH2-. The charge information of PQD1, PQD2 and PQD3 was analyzed using Zeta potential. Figure 2As shown in E, the surface bands of the three PQDs are all positive (+78.3mV, +75.7mV and +64.9mV respectively). These characterization results all indicate the successful synthesis of water-soluble PQDs. In addition, the fluorescence emission spectra of the three different PQDs were measured in the excitation wavelength range of 280-380nm. The results are shown in Figure 3 As shown, the excitation wavelengths corresponding to the maximum fluorescence emission intensity of PQD1, PQD2 and PQD3 are 360nm, 300nm, and 300nm, respectively.

[0046] Example 2: Feasibility test of constructing a fluorescence sensor array (1) Experimental method: Five types of bacteria grown in solid Luria-Bertani (LB) medium were selected ( E. coli , S. aureus , S. typhimurium , L. monocytogenes and P. aeruginosa ) and washed twice with ultrapure water. Then the OD value of the bacterial resuspension was measured using a biospectrometer. 600 For the standard curve, the bacterial resuspension was diluted with ultrapure water to different concentrations (1.0 × 10 3 CFU / mL, 1.0×10 4 CFU / mL, 1.0×10 5 CFU / mL, 1.0×10 6 CFU / mL, 1.0×10 7 CFU / mL). Then, the five pathogen resuspensions with different concentrations were incubated with three PQDs (i.e., PQD1, PQD2, and PQD3) for 5 minutes, and the fluorescence changes were measured using a fluorescence spectrophotometer.

[0047] (2) Experimental results: When these three PQDs were combined with five types of bacteria, fluorescence quenching occurred to varying degrees as the bacterial concentration gradually increased ( Figure 4 ).by E. coli For example, when it is combined with PQDs, E. coli The concentration was 1.0×10 3 CFU / mL to 1.0×10 7 The fluorescence quenching effect of PQDs gradually increased with the gradual increase of CFU / mL. E. coliThe fluorescence quenching effect of different PQDs showed obvious differences. In addition, this difference also existed in four other foodborne pathogens. The difference in fluorescence quenching between different bacteria and different PQDs provides a theoretical basis for the construction of sensors based on fluorescence arrays, which helps to achieve rapid, sensitive and wide-range identification of bacterial strains.

[0048] Example 3: Performance test of the fluorescence sensor array to distinguish five types of bacteria (1) Experimental method: Five types of bacteria grown in solid Luria-Bertani (LB) medium were selected ( E. coli , S. aureus , S. typhimurium , L. monocytogenes and P. aeruginosa ) and washed twice with ultrapure water. Then the OD value of the bacterial resuspension was measured with a biospectrometer. According to the pre-determined CFU-OD 600 standard curve, the bacterial resuspension was diluted with ultrapure water to 1.0×10 3 CFU / mL. The five bacterial resuspensions were incubated with the three sensing units (i.e., three PQDs) for 5 minutes. The relative signal intensity changes (ΔRGB) of the three PQDs after interaction with the bacteria were then observed and recorded using the Color Grab application on a smartphone. After the ΔRGB values ​​were collected, pattern recognition methods (including multiple machine learning algorithms) were used to conduct in-depth analysis of the ΔRGB values. In the training phase, 70% of the available data were divided into a training set and the remaining 30% were used as a test set by a random sampling procedure. For each bacterium and the three PQDs alone, the experiment was systematically repeated 12 times to construct a comprehensive training data set containing 180 samples (3 PQDs × 5 bacteria × 12 repetitions). Subsequently, the data set was analyzed using the built-in machine learning algorithms in MATLAB to achieve classification and recognition of the bacteria. The algorithms used included decision trees, support vector machines (SVM), k-nearest neighbors (KNN), linear discriminant analysis (LDA), and naive Bayes methods.

[0049] (2) Experimental results: Figure 5As shown in A, the interaction between different bacterial strains and different PQDs resulted in significant differences in ΔRGB values. This difference may be due to the different binding abilities of bacteria to PQDs, which is affected by factors such as the size of the bacteria, the size of the surface charge, and the surface chemical composition. In addition, we evaluated the superiority of these machine learning algorithms by using four performance indicators, including positive predictive value (PPV), true positive rate (TPR), F1 score, and prediction accuracy. The F1 score is defined as F1 score = 2×PPV×TPR / (PPV+TPR). It is worth noting that all performance indicators of the five machine learning methods showed excellent results, such as Figure 5 B. After comprehensive consideration, the support vector machine (SVM) algorithm was selected as the preferred method for subsequent data analysis. Figure 5 The results of the C confusion matrix verified the high accuracy of the SVM algorithm in data processing, with both the true positive rate (TPR) and the precision rate (PPV) reaching 100%. This remarkable performance confirms the effectiveness of machine learning technology in the task of bacterial identification. In order to further illustrate the relationship between the three key factors identified by the machine algorithm, a parallel coordinate plot ( Figure 5 D) Visualization method. This method effectively reveals the inherent trends and connections in multivariate data and provides an intuitive perspective for understanding the complex dynamics among multiple variables. Figure 6 A shows the recognition of five bacteria by the fluorescent sensor array in the three-dimensional feature space, where each data point corresponds to the relative signal intensity change after the interaction between a specific bacterium and three PQDs. The results show that the five bacteria formed independent clusters with clear boundaries between them, and the recognition accuracy reached 100%. Three typical factors were generated from the data set, accounting for 88.6%, 9.3% and 2.1% of the total variance, respectively. Using the two most significant discriminant factors, the bacterial identification map was generated ( Figure 6 B), which clearly distinguishes the five bacteria without any overlap and maintains 100% identification accuracy. This result highlights the high discrimination ability of the PQDs-based sensor array in bacterial identification. In addition, the heat map ( Figure 6 C) and radar chart ( Figure 6 D), further verifying the reliability and accuracy of machine learning algorithms combined with sensor arrays in bacterial identification.

[0050] Example 4: Quantitative analysis test of five bacteria using fluorescent sensor array (1) Experimental method: Five types of bacteria grown in solid Luria-Bertani (LB) medium were selected ( E. coli , S. aureus , S. typhimurium , L. monocytogenes and P. aeruginosa ) and washed twice with ultrapure water. Then the OD value of the bacterial resuspension was measured using a biospectrometer. 600 For the standard curve, the bacterial resuspension was diluted with ultrapure water to different concentrations (1.0 × 10 3 CFU / mL, 1.0×10 4 CFU / mL, 1.0×10 5 CFU / mL, 1.0×10 6 CFU / mL, 1.0×10 7 CFU / mL). Each bacterial resuspension at different concentrations was reacted with the sensor array, and the ColorGrab application on the smartphone was used to collect the relative signal intensity change (ΔRGB) data of the fluorescent sensor array to different concentrations of bacteria to form a data set, and then the machine learning algorithm was used to analyze the test data set.

[0051] (2) Experimental results: Figure 7 As shown, E. coli For example, the sensor array can accurately distinguish samples of different concentrations. Further, the correlation analysis between Factor 1 and bacterial concentration revealed that the change in sensor array signal intensity is related to E. coli There is a significant linear relationship between the concentrations. In addition, the application of heat maps and radar maps further highlights the excellent ability of the fluorescent sensor array in distinguishing different concentrations of E. coli. Similarly, the sensor array also showed consistent high resolution in identifying different concentrations of four other foodborne pathogens ( Figures 8 to 11 ). Fig.12 The detection limits of the sensor array for these five bacteria are summarized in Figure 3. These results show that the fluorescent sensor array can not only identify bacterial species, but also has the ability to conduct quantitative analysis.

[0052] Example 5: Performance test of fluorescent sensor array to distinguish mixed bacteria (1) Experimental method: Five types of bacteria grown in solid Luria-Bertani (LB) medium were selected ( E. coli , S. aureus , S. typhimurium , L. monocytogenes and P. aeruginosa ) and washed twice with ultrapure water to obtain a bacterial suspension for later use. Two double-strain mixed sample groups were prepared, namely Listeria ( L. monocytogenes ) and Escherichia coli ( E. coli ) and Staphylococcus aureus ( S. aureus ) and Salmonella typhimurium ( S. typhimurium) and a set of three mixed samples were prepared, including Listeria monocytogenes ( L. monocytogenes ), Staphylococcus aureus ( S. aureus ) and Pseudomonas aeruginosa ( P. aeruginosa ). For the double-strain mixed samples, seven different mixing ratios were designed, namely L. monocytogenes and E. coli 0:1 to 1:0, S. aureus and S. typhimurium The ratio of the three strains was 0:1 to 1:0. Nine different ratio combinations were designed for the three-strain mixed sample, covering different situations from a single strain to an equal ratio of three strains. The OD value of the mixed bacterial resuspension was then measured using a biospectrometer. According to the pre-determined CFU-OD 600 For the standard curve, the mixed bacterial resuspension was diluted to 1.0 × 10 3 CFU / mL. The mixed bacterial resuspensions of different types were respectively exposed to the sensor array, and the Color Grab application on the smartphone was used to collect the relative signal intensity change (ΔRGB) data of the fluorescent sensor array to different types of mixed bacteria to form a data set, and then the machine learning algorithm was used to analyze the test data set.

[0053] (2) Experimental results: Fig.13 As shown in the figure, the fluorescent sensor array maintained excellent performance in the recognition of binary mixed bacteria and ternary mixed bacteria. Through the analysis of confusion matrix and parallel coordinates, the SVM algorithm achieved 100% recognition accuracy. The above results confirm the excellent performance of the fluorescent sensor array constructed based on water-soluble PQDs in detecting single and mixed foodborne pathogens.

[0054] Example 6: Study on the anti-interference, stability and repeatability of the fluorescence sensor array (1) Experimental method: NaOH, KOH, HCl, glucose, chitosan, L-arginine and L-tyrosine were selected as interfering substances and prepared into solutions with a concentration of 300 µM. Different types of interfering substance solutions were respectively applied to the sensor array, and the Color Grab application on the smartphone was used to collect the relative signal intensity change (ΔRGB) data of the fluorescence sensor array to different types of interfering substances to form a data set, and then the machine learning algorithm was used to analyze the test data set. In addition, the fluorescence sensor array was subjected to a concentration of 1.0×10 3 The ΔRGB response values ​​of five foodborne bacteria with a concentration of 1.0×10 3CFU / mL E. coli The solution was used as the test sample and was measured 10 times in a row to verify the repeatability of the sensor array. At the same time, the sensor array was used to test the sample every other day for 15 consecutive days to evaluate its stability.

[0055] (2) Experimental results: Fig.14 As shown, the fluorescent sensor array can effectively distinguish interferents from target bacteria, showing good anti-interference properties. After ten consecutive repeated measurements, the signal intensity of the sensing platform changed slightly, indicating that it has excellent repeatability. After 15 days of storage, the ΔRGB value remained stable, confirming the long-term stability of the sensor array. In summary, the sensor array has high selectivity, repeatability and stability for target bacteria detection.

[0056] Example 7: Detection of blind samples by fluorescence sensing array (1) Experimental method: Five types of bacteria grown in solid Luria-Bertani (LB) medium were selected ( E. coli , S. aureus , S. typhimurium , L. monocytogenes and P. aeruginosa ) and washed twice with ultrapure water to obtain a bacterial suspension for later use. 3 In addition to the single bacterial sample with CFU / mL, a dual-strain mixed sample group was prepared, which was Listeria ( L. monocytogenes ) and Escherichia coli ( E. coli ) and a set of three mixed samples were prepared, including Listeria monocytogenes ( L. monocytogenes ), Staphylococcus aureus ( S. aureus ) and Pseudomonas aeruginosa ( P. aeruginosa ). For the two-strain mixed sample, three different mixing ratios were designed, which were 0.2:0.8, 0.5:0.5, and 0.8:0.2. Three different ratio combinations were also designed for the three-strain mixed sample, which were 0.2:0.2:0.6, 0.2:0.6:0.2, and 0.6:0.2:0.2. The OD value of the mixed bacterial resuspension was then measured using a biospectrometer. According to the pre-determined CFU-OD 600 For the standard curve, the mixed bacterial suspension was diluted to 1.0 × 10 3 Then all samples were randomly shuffled and tested using a fluorescent sensor array. The Color Grab application on a smartphone was used to collect the relative signal intensity change (ΔRGB) data of the fluorescent sensor array for different samples to form a data set, and then a machine learning algorithm was used to analyze the test data set.

[0057] (2) Experimental results: The fluorescence sensing array successfully identified all 62 blind samples, including 26 unknown single bacterial samples, 24 unknown binary mixed bacterial samples, and 12 unknown ternary mixed bacterial samples. The specific results are shown in Fig.15 .

[0058] Example 8: Detection of real samples by fluorescence sensing array (1) Experimental method: Tap water was used as the actual sample, and the detection accuracy of the fluorescence sensor array was evaluated by the standard addition method. The tap water was pretreated before detection, that is, the water sample was filtered using a 0.45 μM sterile filter membrane to remove suspended particles and large microbial aggregates that may interfere with fluorescence detection. Five types of bacteria grown in solid Luria-Bertani (LB) medium ( E. coli , S. aureus , S. typhimurium , L. monocytogenes and P. aeruginosa ) and washed twice with pre-treated tap water. Then the OD value of the bacterial resuspension was measured using a biospectrometer. 600 For the standard curve, the bacterial suspension was diluted to 1.0 × 10 3 Then all samples were randomly shuffled and tested using a fluorescent sensor array. The Color Grab application on a smartphone was used to collect the relative signal intensity change (ΔRGB) data of the fluorescent sensor array for different samples to form a data set, and then a machine learning algorithm was used to analyze the test data set.

[0059] (2) Experimental results: Even in the complex medium of tap water, the sensor array still accurately distinguished all samples. The results are as follows: Fig.16 It is demonstrated that the constructed fluorescence sensing array is suitable for identifying these five types of bacteria in real samples.

[0060] Example 9: Study on the bactericidal performance of the fluorescence sensor array (1) Experimental method: Select the cells grown in solid Luria-Bertani (LB) medium. E. coli and S. aureus The single colony was isolated and washed twice with ultrapure water. The OD value of the bacterial resuspension was then measured using a biospectrometer. 600Standard curve, the bacterial resuspension was diluted with ultrapure water to an OD value of 0.2. The suspension was randomly divided into 4 groups: control group, PQD1 group, PQD2 group and PQD3 group. PQDs solution with a concentration of 12.5 mg / mL was added to each centrifuge tube of the experimental group and incubated for 30 minutes at room temperature. After the incubation, the mixed solution was centrifuged at 4000 rpm for 3 minutes, and the precipitate was washed several times to finally obtain the bacterial resuspension. Then it was diluted 10000 times and 100 μL was evenly spread on the plate, placed at 37 ° C to grow a single colony, and the colonies were counted. The bactericidal performance of PQDs was observed using the Calcein AM / PI live cell staining kit. In short, after incubation with the bacterial suspension for 30 minutes, the sample was washed several times with ultrapure water and the bacterial resuspension was prepared. Subsequently, calcein-AM (1.5 µL / mL) and PI (1.5 µL / mL) were added to the samples, and then incubated at 37 °C in the dark for 45 min. The prepared samples were observed under an inverted fluorescence microscope.

[0061] (2) Experimental results: Fig.17 As shown in A, after being acted upon by PQDs solution, E. coli and S. aureus The colony forming units (CFU) of the samples were significantly lower than those of the blank control group, indicating that PQDs had a significant antibacterial effect. Fig.17 As shown in B, PQD1, PQD2 and PQD3 E. coli The antibacterial rates were 99.94%, 99.93% and 99.67% respectively. S. aureus The antibacterial rates of PQDs solution were 99.27%, 99.37% and 99.33% respectively, which verified that PQDs solution had high antibacterial activity against both bacteria. E. coli and S. aureus The morphological observation results are as follows Fig.17 C. The bacterial morphology of the control group was intact, but after being treated with PQD1, the bacterial cells shrank and ruptured, and the cell contents overflowed, revealing the destructive effect of PQDs on the bacterial cell structure. In addition, the bacteria treated with PQD1 were stained using the Calcein AM / PI live cell staining kit. Fig.17 D shows the staining results, in which the significant enhancement of the red fluorescence signal and the disappearance of the green fluorescence signal indicate that the integrity of the bacterial cell membrane is severely damaged and the bacterial survival rate is reduced after PQD1 treatment. The above results all show that the constructed fluorescent sensing array has excellent bactericidal performance.

[0062] The present invention has successfully developed a fluorescent sensor array that integrates rapid detection and sterilization functions. The fluorescent sensor array is composed of three perovskite quantum dots with different fluorescent properties. It can not only quickly and accurately identify foodborne pathogens, but also inactivate pathogens 30 minutes after detection. The present invention provides a new and efficient tool for food safety detection, which has important application prospects and commercial potential.

[0063] The machine learning algorithm used in the present invention is based on MATLAB software, and the required hardware conditions only need to meet the 64-bit operating system of Windows 10, 16GB of memory, and a graphics card of Nvidia GTX 1050Ti. Using machine learning algorithms to detect foodborne pathogens is an innovative and practical method that can provide more comprehensive and accurate protection for food safety. With the continuous development and optimization of technology, it is believed that this method will be more widely used and promoted in the future.

[0064] The algorithm database of the present invention is obtained by: (1) 0.5 mL of five bacterial suspensions from the same batch were incubated with 0.5 mL of three perovskite quantum dot solutions for 5 min respectively. The Color Grab application on a smartphone was used to collect the relative signal intensity change (ΔRGB) data of the fluorescence sensor array for different samples to form a data set.

[0065] (2) The relative signal intensity change (ΔRGB) was calculated according to the following formula: ΔRGB = (R0−R) / R0+(G0−G) / G0+(B0−B) / B0, where R0, G0, B0 are the initial red, green, and blue values ​​of the image without the addition of bacteria, and R, G, and B are the red, green, and blue values ​​of the image extracted after the addition of bacteria.

[0066] (3) After the data set is collected, the built-in machine learning algorithm in MATLAB is used to analyze the data set to achieve classification and identification of bacteria. The algorithms used include decision tree, support vector machine (SVM), k-nearest neighbor (KNN), linear discriminant analysis (LDA) and naive Bayes method. The flowcharts of the relevant algorithms are shown in the figure below. Figure 18-22 shown.

[0067] A method for rapid detection of foodborne pathogens based on water-soluble CsPbBr3 perovskite quantum dots and machine learning algorithms is obtained by: (1) For the individual effects of each bacterium with the three PQDs, the experiment was systematically repeated 12 times, and a comprehensive training dataset containing 180 samples was constructed (3 PQDs × 5 bacteria × 12 repetitions).

[0068] (2) After collecting the ΔRGB values, pattern recognition methods (including various machine learning algorithms) were used to conduct in-depth analysis of the ΔRGB values. During the training phase, 70% of the available data was divided into a training set through a random sampling procedure, and the remaining 30% was used as a test set.

[0069] (3) All five algorithms, including decision tree, support vector machine (SVM), k-nearest neighbor (KNN), linear discriminant analysis (LDA) and naive Bayes method, were able to successfully distinguish bacteria.

[0070] (4) The support vector machine (SVM) among the five algorithms enhances the generalization ability of the model by maximizing the interval of the decision boundary, thereby providing higher prediction accuracy on unknown data. Secondly, the kernel technique of SVM enables the algorithm to effectively handle nonlinear problems by implicitly mapping the input data to a high-dimensional space to find the optimal linear segmentation plane. In addition, SVM also has good scalability and can handle multi-class classification problems through different strategies, making it perform well in a variety of application scenarios.

[0071] The above description is only used to explain the preferred embodiments of the present invention, and is not intended to limit the present invention in any form. Therefore, any modifications or changes made to the present invention under the same inventive spirit should still be included in the scope of protection intended by the present invention.

Claims

1. A method for preparing water-soluble CsPbBr3 perovskite quantum dots, characterized in that: The following steps are involved: Step 1: Prepare the precursor solution Dispersing cesium bromide in hydrobromic acid to obtain a cesium bromide mother solution, dispersing lead bromide in hydrobromic acid to obtain a lead bromide mother solution; mixing the cesium bromide mother solution and the lead bromide mother solution, and then adding N, N-dimethylformamide to the mixture; then adding perfluorooctyltriethoxysilane, and reacting to obtain a precursor solution; Step 2: Preparation of composite emulsion The invention relates to the preparation of at least one of the following composite emulsions: Preparation of PQD1 composite emulsion: adding oleylamine and water to the precursor solution, and obtaining a composite emulsion containing PQD1 after reaction; Preparation of PQD2 composite emulsion: adding oleylamine, hydrochloric acid and water to the precursor solution, and obtaining a composite emulsion containing PQD2 after reaction; Preparation of PQD3 composite emulsion: adding oleylamine, hydrochloric acid and water to the precursor solution, and obtaining a composite emulsion containing PQD3 after reaction, wherein the volume of hydrochloric acid used in step 4 is 0.8-1.5 times the volume of hydrochloric acid used in step 3; Step 3: Centrifugation purification: At least one of the composite emulsion containing PQD1, the composite emulsion containing PQD2 and the composite emulsion containing PQD3 is centrifuged to discard the precipitate and take the supernatant; the supernatant is centrifuged again to remove the unreacted precursor, and the precipitate is redispersed in water to obtain an aqueous solution containing PQD1, an aqueous solution containing PQD2 and an aqueous solution containing PQD3.

2. The method for preparing water-soluble CsPbBr3 perovskite quantum dots according to claim 1, characterized in that: In the mixture, the mass ratio of cesium bromide mother liquor to lead bromide is 0.8-1.2:1, the speed of the first centrifugation is 5-8000 rpm, and the speed of the second centrifugation is 10000-15000 rpm.

3. A method for detecting foodborne pathogens using a water-soluble CsPbBr3 perovskite quantum dot fluorescence sensor array based on a machine learning algorithm, characterized in that: The following steps are involved: S1. Inoculate the bacteria onto LB solid medium and culture at 35-38°C overnight; then select a single colony from the cultured colonies, wash it with ultrapure water and resuspend it; then measure the OD value of the bacterial resuspension; according to the pre-determined CFU-OD 600 For the standard curve, the bacterial resuspension was diluted with ultrapure water to different concentrations: 1.0×10 3 CFU / mL, 1.0×10 4 CFU / mL, 1.0×10 5 CFU / mL, 1.0×10 6 CFU / mL, 1.0×10 7 CFU / mL; Reserve; S2, the concentration is 1.0×10 3 CFU / mL of bacterial solution, respectively, combined with the aqueous solution containing PQD1, the aqueous solution containing PQD2, and the aqueous solution containing PQD3 as described in claim 1 or 2, and observed the changes in the fluorescence intensity of the material and the differences in the changes in fluorescence intensity after combining with different bacterial species to verify the feasibility of constructing a fluorescent sensor array; S3, dropping bacterial resuspensions of different concentrations and types into the fluorescent sensor array composed of three perovskite quantum dot arrangements, and incubating under standard conditions; then, obtaining the relative signal intensity change value of the image through the Color Grab app on the smartphone; S4. After collecting the relative signal strength change values, the relative signal strength change values ​​are analyzed using a pattern recognition method; in the training phase, 70% of the available data is divided into a training set through a random sampling procedure, and the remaining 30% is used as a test set; Subsequently, the data set was analyzed using MATLAB's built-in machine learning algorithms to achieve classification and identification of bacteria; the algorithms used included: decision tree, support vector machine (SVM), k-nearest neighbor (KNN), linear discriminant analysis (LDA), and naive Bayes method.

4. The method for detecting foodborne pathogens using a water-soluble CsPbBr3 perovskite quantum dot fluorescence sensor array based on a machine learning algorithm according to claim 2, characterized in that: The bacterial species include at least one of the following bacterial species: Escherichia coli, Staphylococcus aureus, Salmonella typhimurium, Listeria monocytogenes, and Pseudomonas aeruginosa.

5. The method for detecting foodborne pathogens using a water-soluble CsPbBr3 perovskite quantum dot fluorescence sensor array based on a machine learning algorithm according to claim 2, characterized in that: The relative signal intensity change value was calculated according to the following formula: ΔRGB = (R0− R) / R0 + (G0 − G) / G0 + (B0 − B) / B0, where R0, G0, B0 are the initial red, green, and blue values ​​of the image without the addition of bacteria, and R, G, and B are the red, green, and blue values ​​of the image extracted after the addition of bacteria.

6. Application of a water-soluble CsPbBr3 perovskite quantum dot fluorescence sensor array in bacterial inactivation based on a machine learning algorithm, characterized in that: The following steps are involved: a. Observe the sterilization effect by plate counting: a bacterial suspension with an OD value of 0.2 was placed in a sterile centrifuge tube, and the bacterial suspension was randomly divided into 4 groups: control group, PQD1 group, PQD2 group and PQD3 group; a PQDs solution with a concentration of 12.5 mg / mL was added to each centrifuge tube, and incubated at room temperature for 30 minutes; after the incubation, the mixed solution was centrifuged at 3000-5000 rpm for 1-5 minutes, and the precipitate was washed several times to finally obtain a bacterial resuspension; then it was diluted 10,000 times, 100 μL was evenly spread on the plate, and cultured at 37°C until a single colony grew, and the colonies were counted; b. Observation of sterilization effect by live-dead cell staining: A calcein AM / PI staining kit was used to observe the bactericidal properties of PQDs; in brief, after incubation with the bacterial suspension for 30 min, the samples were washed several times with ultrapure water and the bacterial resuspension was prepared; subsequently, calcein-AM (1.5 μL / mL) and PI (1.5 μL / mL) were added to the samples, and then incubated at 37 °C in the dark for 45 min; the prepared samples were observed under an inverted fluorescence microscope.

7. The use of a water-soluble CsPbBr3 perovskite quantum dot fluorescence sensor array based on a machine learning algorithm in bacterial inactivation according to claim 6, characterized in that: After incubation with the bacterial suspension for 30 min, the samples were washed several times with ultrapure water and the bacterial resuspension was prepared; subsequently, calcein-AM 1.5 μL / mL and PI 1.5 μL / mL were added to the samples, followed by incubation at 37°C in the dark for 45 min.

8. The use of a water-soluble CsPbBr3 perovskite quantum dot fluorescence sensor array based on a machine learning algorithm in bacterial inactivation according to claim 6, characterized in that: The bacterial species of the bacterial suspension include at least one of the following bacterial species: Escherichia coli, Staphylococcus aureus, Salmonella typhimurium, Listeria monocytogenes, and Pseudomonas aeruginosa.

Images