Method, system, device, medium and drug sensitivity detection method for single bacteria recognition based on lensless imaging

Abstract

The application provides a single bacterium identification method, system, device, medium and drug sensitivity detection method based on lens-free imaging, and belongs to the technical field of biological image processing. The single bacterium identification method based on lens-free imaging comprises steps of gray scale processing, edge detection and single bacterium identification, and further comprises steps of background removal, abnormal pixel correction, signal enhancement and outlier screening. The single bacterium identification method, system, device and medium have the advantages of high speed and high accuracy. Further used for bacterial drug sensitivity detection, the proliferation of bacteria can be obtained in a short time, and then the minimum inhibitory concentration of antibiotics is obtained, which has great application value.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of biological image processing, and in particular to a method and system for single-bacterium identification based on lens-free imaging, a device, a medium and a drug sensitivity detection method. BACKGROUND

[0002] Traditional antimicrobial susceptibility testing (AST), such as paper disc diffusion, agar dilution, broth microdilution, etc., requires culturing, isolating and enriching bacteria in clinical samples, and then observing the inhibitory effect of antibiotics on bacterial growth to obtain the drug sensitivity result of the bacteria. The process is complex and time-consuming (2-4 days).

[0003] In recent years, rapid phenotypic drug sensitivity detection technology has been applied to clinical diagnosis and treatment, which can significantly shorten the drug sensitivity report time and reduce the mortality rate of infected patients. Among them, the rapid drug sensitivity detection technology based on optical imaging has the advantage of high sensitivity of single-bacterium imaging, which can shorten the drug sensitivity detection time to 1 hour or even less, such as a microscope imaging system based on microfluidics, a Raman imaging system, etc.

[0004] However, such technology often requires a complex and precise optical imaging structure, which is high in cost and complex in operation. SUMMARY

[0005] To solve at least one of the above technical problems, the technical solution adopted by the present application is as follows.

[0006] The first aspect of the present application provides a method for single-bacterium identification based on lens-free imaging, which processes a microscopic image of a bacterial solution obtained based on lens-free imaging to identify single bacteria, comprising the following steps:

[0007] Gray processing: performing gray processing on the microscopic image to obtain a gray image;

[0008] Edge detection: using a Sobel operator to perform edge detection on the image:

[0009] Single-bacterium identification: first, using a second Gaussian filter to smooth the image, and then using a Laplacian filter to convolve the smoothed image to obtain the target bacteria in the image.

[0010] Lens-free imaging technology is a high-throughput microscopic imaging technology that does not require any lens, but directly places the object to be tested on or close to the surface of the image sensor for imaging. Due to its small size, low cost and other characteristics, it is suitable for application in POCT (Point of Care Testing) technology.

[0011] In some embodiments of the present application, the following formula is used for graying:

[0012]

[0013] wherein, G0(x, y) represents the original gray value of the pixel point with coordinates of (x, y), G1(x, y) and G2(x, y) respectively represent the R value and the B value of the pixel point, R G3(x, y) represents the G value of the pixel point. G B

[0014] Further, the present application uses an improved Sobel operator to perform edge detection on the image:

[0015]

[0016]

[0017]

[0018] wherein, and are defined by the convolution kernel function of the Sobel operator, G0(x, y) represents the edge gray value of the pixel point.

[0019] In some embodiments of the present application, the single bacteria identification step is specifically:

[0020]

[0021]

[0022]

[0023]

[0024] wherein, the kernel parameter of the first Gaussian filter is , the diameter of the target bacteria.

[0025] In some embodiments of the present application, after the graying step and before the edge detection step, the following steps are further included:

[0026] Background removal: a first Gaussian filter is used to convolve the image to obtain a background gray value, and a foreground gray value is obtained by difference operation with the original gray value, specifically:

[0027]

[0028]

[0029] ​​wherein, represents the estimated background gray value of the pixel point with coordinates , represents the first Gaussian kernel function with standard deviation , , , represents the calculated foreground gray value of the pixel point,

[0030] At this time:

[0031] .

[0032] In some embodiments of the present application, after the background removal step and before the edge detection step, the following step is further included:

[0033] Abnormal pixel correction: for each pixel point, the gray values of all pixels in the region around the pixel point not greater than the target bacteria size are obtained and the median and standard deviation are calculated, and the sizes of the following two values are judged:

[0034] ① the absolute value of the difference between the gray value of the pixel point and the median;

[0035] ② the product of the standard deviation and a first preset threshold,

[0036] If ①>②, the pixel point is identified as an outlier, and the median is replaced.

[0037] In some specific embodiments of the present application, a circular sampling window ( ) is used for each pixel to calculate the median and the standard deviation:

[0038]

[0039]

[0040] wherein, represents the foreground gray value median of the pixel point with coordinates , represents the foreground gray value standard deviation of the pixel point, and then the abnormal pixel is replaced or removed by the following method:

[0041]

[0042] wherein, represents the corrected gray value of the pixel point after the abnormal value replacement, represents a first preset threshold, in some preferred embodiments of the present application, the value is 4-5,

[0043] At this time,

[0044] In some embodiments of this application, after the edge detection step and before the single-bacterial identification step, the following steps are also included:

[0045] Signal enhancement: For any edge pixel, binarize it by comparing its gray value with a second preset threshold; then perform closing and opening operations to obtain the enhanced edge.

[0046] In some embodiments of this application, the purpose of binarization is to highlight the edges of the bacteria, specifically:

[0047]

[0048] in, This represents the edge grayscale value of the pixel. This indicates the second preset threshold. The value of is determined according to the following steps:

[0049] for Set an initial value and perform binarization using the formula above;

[0050] The bright spots in the binarized image are divided into two types: small bright spots (background noise points) and large bright spots with a ring-shaped outer circle (target bacteria points).

[0051] If regarding settings If the radius of the background noise point is no greater than 2 pixels, and the difference in radius between the outer and inner circles of the target bacteria point is no greater than 4 pixels, then this value is selected as the second preset threshold; otherwise, Add 1 to the current value and repeat the above steps until the radius of the background noise point is no more than 2 pixels and the difference in radius (ring width) between the outer and inner rings of the target bacteria point is no more than 4 pixels.

[0052] In some embodiments of this application, the radius of the background noise point is calculated using the perimeter or radius. As long as the perimeter or area is no greater than 4π, the radius is no greater than 2 pixels. Similarly, the annular width can also be calculated using the perimeter or area of ​​the inner and outer rings. In one specific embodiment of this application, the perimeter is used for calculation: Annular width = (Outer ring perimeter - Inner ring perimeter) / 2π.

[0053] In some embodiments of this application, the background noise consists of small particles, i.e., small areas, while the edge of the target bacteria is a ring (caused by light refraction). If the value is small, the area of ​​background noise points will be larger during binarization, and the ring width of the target bacteria points will also be larger. This noise points will be included in the subsequent counting, thus affecting the accuracy of the count. With... The value gradually increases, when the two requirements (i.e. the radius of the background noise point is not greater than 2 pixel points, and the ring width of the target bacteria point is not greater than 4 pixel points) can be met, the background noise point can be eliminated in the subsequent morphological calculation, and the target bacteria point only changes the edge image when the edge is enhanced without changing the original image of the bacteria center.

[0054] In order to eliminate the background noise point and enhance the edge, a 2-pixel disc is set to perform a closing operation and then an opening operation:

[0055]

[0056]

[0057] wherein, represents the closing operation gray value after filling the low brightness, represents the opening operation gray value for removing redundant white point noise, and represents a disc structure with a radius of 2 pixels;

[0058] Finally, the current image gray value is subtracted from the opening operation gray value to obtain an enhanced gray value .

[0059] In some embodiments of the present application, after the single bacteria recognition step, the following steps are further included:

[0060] Outlier screening: by setting a threshold for at least one of the signal-to-noise ratio, contrast, minimum brightness and brightness variance, the false target bacteria are removed.

[0061] The second aspect of the present application provides a system for single bacteria recognition based on lensless imaging, comprising the following modules:

[0062] An image acquisition module is used to obtain a microscopic image of a bacterial solution based on lensless imaging;

[0063] A gray processing module is connected to the image acquisition module and is used to gray the microscopic image to obtain a gray image;

[0064] An edge detection module is connected to the gray processing module and is used to detect the edge of the image using a Sobel operator;

[0065] A single bacteria recognition module is connected to the edge detection module and is used to first convolve the image using a Gaussian filter, and then calculate the Laplace of the image to obtain the target bacteria in the image.

[0066] The third aspect of the present application provides a computer device, comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of the method according to any one of the first aspect of the present application.

[0067] The fourth aspect of the present application provides a computer-readable storage medium, having stored thereon a computer program, which when executed by a processor, implements the steps of the method according to any one of the first aspect of the present application.

[0068] The fifth aspect of the present application provides a method for detecting bacterial drug sensitivity, comprising the following steps:

[0069] exposing the bacteria to be tested to different concentrations of antibiotics in detection cavities, wherein a lens-free image sensor is disposed at the bottom of each detection cavity;

[0070] obtaining microscopic images of the bacterial liquid in each detection cavity by using the lens-free image sensor before and after the bacteria to be tested are exposed to different concentrations of antibiotics for a period of time, respectively;

[0071] identifying single bacteria in the microscopic images and counting the number of the single bacteria by using the method according to any one of the first aspect of the present application, the system according to any one of the second aspect of the present application, the computer device according to any one of the third aspect of the present application, or the computer-readable storage medium according to any one of the fourth aspect of the present application;

[0072] if the proliferation rate of the bacteria to be tested before and after being exposed to antibiotics at a certain concentration for a period of time is less than a third preset threshold, and the proliferation rate of the bacteria to be tested before and after being exposed to antibiotics at a concentration lower than the certain concentration for a period of time is not less than the third preset threshold, then the certain concentration is the minimum inhibitory concentration of the antibiotics to the bacteria.

[0073] In some embodiments of the present application, the antibiotics are selected from at least one of the following: β-lactams, aminoglycosides, macrolides, tetracyclines, quinolones, sulfonamides, glycopeptides, lincosamides, nitroimidazoles, and polypeptide antibiotics.

[0074] wherein the β-lactams further include penicillins, cephalosporins, and carbapenems, the penicillins include but are not limited to penicillin G, ampicillin, and piperacillin; the cephalosporins include but are not limited to cefazolin, ceftriaxone, ceftazidime, and cefotaxime; and the carbapenems include but are not limited to imipenem, meropenem, and ertapenem.

[0075] Aminoglycosides include, but are not limited to, Gentamicin, Tobramycin and Amikacin.

[0076] Macrolides include, but are not limited to, Erythromycin, Azithromycin and Clarithromycin.

[0077] Tetracyclines include, but are not limited to, Tetracycline, Doxycycline and Minocycline.

[0078] Quinolones include, but are not limited to, Ciprofloxacin, Levofloxacin and Moxifloxacin.

[0079] Sulfonamides include, but are not limited to, Sulfamethoxazole / Trimethoprim (SMZ-TMP).

[0080] Glycopeptides include, but are not limited to, Vancomycin and Teicoplanin.

[0081] Lincosamides include, but are not limited to, Clindamycin and Lincomycin.

[0082] Nitroimidazoles include, but are not limited to, Metronidazole.

[0083] Polypeptides include, but are not limited to, Polymyxin B and Colistin.

[0084] Of course, the present application can also be applied to the drug sensitivity detection of other antibiotics such as Linezolid, Daptomycin, Tigecycline and Fusidic acid.

[0085] For different antibiotics / bacteria, the corresponding third preset threshold needs to be adjusted accordingly, and the third preset threshold can be obtained by using a population sample.

[0086] Compared with the prior art, the present application has the following beneficial effects:

[0087] The method, system, device and medium of the application have the advantages of fast speed and high precision in single bacteria identification. The method, system, device and medium of the application are further used for bacteria drug sensitivity detection, and the proliferation of bacteria can be obtained in a short time, and then the minimum inhibitory concentration of antibiotics is obtained, which has great application value.

[0088] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the application, nor to limit the scope of the application. Other features of the application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0089] The above and other objects, features and advantages of the exemplary embodiments of the present application will be more apparent from the following detailed description read in conjunction with the accompanying drawings, in which:

[0090] Figure 1 A flowchart of image processing and single bacteria identification for a lensless image in Embodiment 2 of the application is shown;

[0091] Figure 2 Effects of abnormal pixel removal when different values are selected in Embodiment 2 of the application are shown, A: before removal, B: value is 2, C: value is 4, D: value is 20;

[0092] Figure 3 Effects of pseudo-target identification and screening using different features in Embodiment 2 of the application are shown;

[0093] Figure 4 Bacterial microscopic images obtained based on lensless imaging of the highest antibiotic concentration group and the control group in Embodiment 3 of the application are shown;

[0094] Figure 5 Microscopic images after image processing and single bacteria identification of the highest antibiotic concentration group and the control group in Embodiment 3 of the application are shown;

[0095] Figure 6 Escherichia coli growth curves at different antibiotic concentrations in Embodiment 3 of the application are shown. DETAILED DESCRIPTION

[0096] ​Unless otherwise indicated, all parts and percentages expressed herein are based upon weight, and all tests and measurements are made in accordance with the methods synchronized with the filing date of this application. To the extent that any patent, patent application, or publication is cited in this application, the contents of all such citations are incorporated herein by reference, and the equivalents of all such cited documents are incorporated herein by reference, particularly with respect to the definitions of terms used therein. If there is a discrepancy between the definitions of terms as provided in any cited document and the definitions provided in this application, the definitions provided in this application shall control.

[0097] In order to make the technical problems, technical solutions and beneficial effects solved in the present application more apparent, the present application will be further described in detail in connection with the embodiments.

[0098] The following examples are presented herein to demonstrate preferred embodiments of the present application. Those skilled in the art will appreciate that the technology disclosed in the following examples represents the best attempts of the inventors to use the technology discovered to practice the present application, and therefore can be considered as preferred modes of practicing the present application. However, those skilled in the art will further appreciate, in light of the present specification, that many modifications of the particular examples disclosed herein can be made, and will obtain the same or similar results without departing from the spirit or scope of the present application.

[0099] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs, and the meaning ascribed to in the materials incorporated herein by reference.

[0100] Those of ordinary skill in the art will realize and appreciate that many modifications of the specific embodiments of the application described herein can be made without departing from the spirit and scope of the application. Accordingly, it is not intended that the application be limited, except as by the appended claims.

[0101] The experimental methods in the following examples are routine methods unless otherwise specified. The instruments and equipment used in the following examples are routine laboratory instruments and equipment unless otherwise specified. The test materials used in the following examples are purchased from routine biochemical reagent stores unless otherwise specified.

[0102] Example 1 Obtaining images of E. coli liquid lens-free imaging

[0103] Preparation of detection cavity using polydimethylsiloxane (PDMS): First, mix PDMS monomer and crosslinking agent (DOW CORNING) in a mass ratio of 10:1, stir evenly, then remove bubbles, then pour the PDMS glue without bubbles into a plastic culture dish with a flat surface and a diameter of 9 cm, place it on a heating plate at 75°C and bake for 1.5 hours, then cut the PDMS film into a square block of 10 mm x 10 mm, the thickness of the PDMS film is about 2 mm, use a punch to punch holes in the cut PDMS film to make detection cavities, the size of the detection cavity is 3.76 mm x 2.74 mm.

[0104] Place the PDMS film containing the detection cavity on a sterile glass slide, and place a lensless image sensor at the bottom of each detection cavity. In this application, the lensless image sensor is obtained by removing the optical lens of the OV5647 (Omnivision) CMOS camera.

[0105] Use a densitometer to configure 0.5 McFarland of E. coli (ATCC No. 25922) initial bacterial solution (concentration about 1.5 x 10 8 CFU / mL), dilute the initial bacterial solution to 1 x 10 5 CFU / mL using CaMHB medium, and take 20 μL and add it to the detection cavity and incubate at 35°C.

[0106] After 60 minutes of incubation, obtain the bacterial image of the detection cavity through the lensless image sensor, i.e. the lensless imaging image.

[0107] Example 2: Single bacterial recognition based on lensless imaging image of bacterial solution

[0108] Reference Figure 1 , the lensless imaging image of the bacterial solution obtained in Example 1 is processed automatically and counted for single bacterial recognition by the following steps.

[0109] 1. Color space dimension reduction (grayscale)

[0110] Based on the CIE 1931 color space conversion model, the data fusion of the RGB three channels of the bacterial lensless microscopic image is realized by weighted average method:

[0111]

[0112] wherein, represents the original gray value of the pixel point with coordinates , R , G and B represent the R value, G value and B value of the pixel point, respectively.

[0113] This process converts 24-bit true color images into 8-bit grayscale images (dynamic range 0-255) while reducing the data dimensionality to preserve texture features for subsequent steps to quantify spatial phenotypic features of bacterial images.

[0114] 2. Spatio-temporal filtering denoising

[0115] A convolution algorithm with large-scale anisotropic Gaussian kernel (first Gaussian filter) is used to construct the background field estimator, and the foreground feature is extracted by difference operation to realize background field compensation and noise suppression:

[0116]

[0117]

[0118] wherein, represents the estimated background grayscale value of the pixel point with coordinates , and represents the first Gaussian kernel function with a standard deviation of ; and represents the calculated foreground grayscale value of the pixel point.

[0119] In this embodiment, the first standard deviation value is confirmed according to the typical size of the bacterial image (the bacterial radius or diameter is used to represent it), and the value is set to , for example, the bacterial radius is 5 pixels, and the value is 10. Through this kernel function, the spatial non-uniformity features in the bacterial culture solution can be effectively captured.

[0120] To avoid signal aliasing caused by spatial high-frequency sampling and accurately model the spatial heterogeneity of the background, the filter kernel size selected in this application is more than 3 times the diameter of the bacteria, i.e. , to ensure that the covered background area is much larger than the area of the bacteria.

[0121] 3. Abnormal pixel correction

[0122] In view of the problem of pulse noise interference commonly existing in microscopic imaging, the abnormal pixel correction algorithm based on spatial range outlier removal is proposed in this embodiment to break through the limitations of traditional nonlinear filtering (such as the traditional median filtering method which causes image blur and reduces data accuracy), and to realize accurate discrimination and repair of noise by constructing a local statistical feature space.

[0123] First, a ring-shaped sampling window with a radius much smaller than the typical size of the bacterial image is established for each pixel (i.e. ​For example ) to calculate the median and standard deviation of its gray scale statistics:

[0124]

[0125]

[0126] wherein, represents the median of the foreground gray scale value of the pixel point with coordinates represents the standard deviation of the foreground gray scale value of the pixel point.

[0127] In order to maximize the ratio of the foreground intensity of the target bacteria to the background intensity, a first threshold value is set to remove outliers, specifically: for any one pixel point , judge the size of the absolute value of the difference between the foreground gray scale value and the median and the product of and the standard deviation, if the former is larger, the pixel point is identified as an outlier, and the median is replaced. That is:

[0128]

[0129] wherein, represents the corrected gray scale value of the pixel point after outlier replacement.

[0130] Finally, the window traverses the entire image to remove all abnormal pixels caused by pulse noise and to retain the original target data.

[0131] wherein, the threshold value According to a large number of experiments, the original data intensity is compared with the background noise intensity, and finally the optimal reasonable value is obtained. For the escherichia coli in the embodiment, when takes different values, the abnormal pixel removal results are significantly different, as shown in Figure 2 When takes a value of 2, the bacteria will also be removed; when takes a value of 20, the denoising is very insufficient; when takes a value of 4~5, for example, takes a value of 4, the abnormal pixel removal is very ideal.

[0132] 4. Signal-to-noise ratio enhancement modeling

[0133] The low-illumination annular feature generated by the refraction effect of the bacterial cell wall in bright field microscopic imaging can be used to construct a signal-to-noise ratio enhancement model based on gradient field reconstruction. First, an improved Sobel operator is used for anisotropic edge detection to obtain an edge detection image. For any one pixel point, the following formula is used to obtain the edge gray scale value:

[0134] ​

[0135]

[0136]

[0137] wherein, and is defined as the convolution kernel function of Sobel operator, represents the edge gray value of the pixel point.

[0138] In order to obtain the shape of the low-illumination annular image of the bacteria, the inventor first sets a second threshold value , and the edge of the detected image is binarized:

[0139]

[0140] That is, if the gray value (0~255 range) of the pixel point is not less than , it is directly assigned as 0, and otherwise, it is directly assigned as 255.

[0141] wherein, represents the edge gray value of the pixel point. The value of is determined according to the following steps:

[0142] is set as an initial value (default 1 starts), and binarization is performed by using the above formula; The bright spots in the binarized image are divided into two types, the small-area bright spots (background noise points), and the large-area bright spots with annular outer ring (target bacteria points);

[0143] If the initial value of

[0144] is set, if the radius of the background noise point is not greater than 2 pixel points, and the radius difference between the outer ring and the inner ring of the target bacteria point is not greater than 4 pixel points, the value is selected as the second preset threshold value, otherwise, 1 is added to the initial value, and the above steps are repeated until the radius of the background noise point is not greater than 2 pixel points, and the radius difference (annular width) between the outer ring and the inner ring of the target bacteria point is not greater than 4 pixel points. In this embodiment,

[0145] the final value is 60. The low-brightness annular image is completed by morphological closing operation, and the redundant white point noise is removed by morphological opening operation, that is, the complete bacteria annular image after noise reduction is obtained.

[0146]

[0147]

[0148] ​

[0149] wherein, represents the closed operation gray value after complementing low brightness, represents the open operation gray value after removing redundant white point noise, and represents a disc structure with a radius of 2 pixels.

[0150] Finally, the enhanced gray value is obtained by subtracting the open operation gray value from the corrected gray value, so as to further enhance the signal-to-noise ratio.

[0151]

[0152] 5. Automatic identification of single bacterial particles based on image segmentation

[0153] The contrast between the bacterial body and the edge is enhanced by the fourth step, so that the peak value of the extreme point of the bacteria is higher and the response is greater.

[0154] Based on the scale space theory, an isotropic filter is constructed, The goal of the filter is to find the spot-like structures (i.e. target bacteria) in the image, which are usually brighter than the surrounding area and approximately circular. The filter combines Gaussian smoothing and Laplacian second derivative:

[0155]

[0156]

[0157]

[0158]

[0159] First, the image is convolved with a second Gaussian filter, the purpose of which is to suppress noise. The Gaussian filter is a weighted average filter, and the weight is determined by the Gaussian function. The farther away from the center, the smaller the weight. The degree of smoothing is controlled by the second standard deviation of the Gaussian function. The larger the second standard deviation, the more blurred the image, and the larger the size of the detectable spots.

[0160] Next, the Laplacian (second spatial derivative) of the image after Gaussian smoothing is calculated. The Laplacian operator is very sensitive to rapid changes in gray scale (edges) in the image, and it will give a strong negative response (for bright spots) or positive response (for dark spots) at the center of the spot (i.e. local intensity maximum or minimum point), and will give a response of opposite sign at the edge of the spot. These strong response points can be used to establish the coordinates of the extreme points, i.e. the coordinates of the bacteria.

[0161] Experiments have shown that when the LoG operator... The value makes Able to interact with bacterial radius During matching, the response at the extreme point coordinates is the strongest, which is beneficial for bacterial detection. Therefore, This yields the Gaussian kernel parameters.

[0162] 6. Outlier removal

[0163] To remove false targets and ensure accurate bacterial identification, the detected targets need further screening. Utilizing the spatial phenotypic features of each bacterial particle obtained in the previous step, false targets can be screened by setting multiple reasonable thresholds for signal-to-noise ratio, contrast, minimum brightness, and brightness variance. This allows the image detection results to approximate the true values, resulting in accurate bacterial counts within the field of view.

[0164]

[0165] in SNR Indicates the signal-to-noise ratio. C Indicates contrast. I min This indicates the lowest brightness (minimum value). STD Indicates variance.

[0166] For Escherichia coli, the effectiveness of using the above characteristics for identification and screening is as follows: Figure 3 As shown.

[0167] Example 3: Sensitivity test of Escherichia coli to amikacin

[0168] This example tests the sensitivity of the Escherichia coli quality control strain ATCC No. 25922 to the antimicrobial drug amikacin.

[0169] Similarly, an initial bacterial culture of 0.5 McFarland (concentration approximately 1.5 × 10⁻⁶) was prepared using a densitometer. 8 (CFU / mL), the initial bacterial culture was diluted to 1×10⁻⁶ using CaMHB medium. 5 CFU / mL was used as the initial bacterial culture for drug susceptibility testing. Amikacin solution was prepared in CaMHB medium to concentrations of 0.125 μg / mL, 0.25 μg / mL, 0.5 μg / mL, 1 μg / mL, 2 μg / mL, 4 μg / mL and 8 μg / mL.

[0170] Add 1 μL of bacterial culture to each detection chamber. Add different concentrations of amikacin solution to different detection chambers (20 μL added, one concentration per detection chamber). Use the detection chamber with an equal amount of blank CaMHB medium as a control.

[0171] The detection cavities were placed in 35℃ for culture. At the beginning of the culture (0 min) and every 30 min, the bacterial images of each detection cavity were acquired by the lensless image sensor (the microscopic images of the highest antibiotic concentration group and the control group are shown in FIG. 2) until the culture reached 120 min. The image processing and single-bacterium recognition were performed by the method of Example 2 (the single-bacterium recognition images of the highest antibiotic concentration group and the control group are shown in FIG. 3), so as to complete the bacterial counting in each detection cavity. Figure 4 Figure 5 The growth curves of each antibiotic group and the control group within 120 min were drawn, and the minimum value was selected as the minimum inhibitory concentration of amikacin to Escherichia coli from the amikacin concentration corresponding to the antibiotic group in which the bacteria were in the inhibited state.

[0172] In this embodiment, the growth curves of each group are shown in FIG. 4, and it can be seen that when the amikacin concentration is greater than 2 μg / mL, the growth of Escherichia coli is inhibited, so that the minimum inhibitory concentration (MIC) of amikacin to Escherichia coli is 2 μg / mL.

[0173] In this embodiment, the growth curves of each group are shown in FIG. 4, and it can be seen that when the amikacin concentration is greater than 2 μg / mL, the growth of Escherichia coli is inhibited, so that the minimum inhibitory concentration (MIC) of amikacin to Escherichia coli is 2 μg / mL. Figure 6

[0174] It can be seen that the drug sensitivity detection by the method of the present application can be completed within 2 hours, which is very fast and convenient.

[0175] In addition, it should be understood that, after reading the above teachings of the present application, those skilled in the art can make various modifications or changes to the present application, and these equivalent forms also fall within the scope defined by the claims attached to the present application.​​

Claims

1. A method for single bacterium identification based on lensless imaging, characterized in that, The method processes a microscopic image of a bacterial solution obtained based on lensless imaging to identify single bacteria, comprising the following steps: gray processing: the microscopic image is grayed to obtain a gray image; edge detection: edge detection is performed on the image using a Sobel operator; single bacterium identification: first, a second Gaussian filter is used to smooth the image, and then a Laplace filter is used to convolve the smoothed image to obtain the target bacteria in the image, wherein the method for obtaining the microscopic image is as follows: a PDMS film containing detection cavities is placed on a sterile glass slide, and a lensless image sensor is arranged at the bottom of each detection cavity, the lensless image sensor is obtained by removing the optical lens of a CMOS camera, bacteria are cultured in the detection cavities, and the bacterial images of the detection cavities are obtained by the lensless image sensor, after the gray processing step and before the edge detection step, the following steps are further included: background removal: a first Gaussian filter is used to convolve the image to obtain a background gray value, and a foreground gray value is obtained by difference operation with the original gray value, after the background removal step and before the edge detection step, the following steps are further included: abnormal pixel correction: for each pixel point, the gray values of all pixels in a region around the pixel point that is not greater than the size of the target bacteria are obtained, and the median and standard deviation are calculated, and the following two values are judged: ① the absolute value of the difference between the gray value of the pixel point and the median; ② the product of the standard deviation and a first preset threshold value, if ①>②, the pixel point is identified as an outlier, and the median is used instead, wherein the bacteria are Escherichia coli, and the first preset threshold value is 4-5.

2. The method of claim 1, wherein, after the edge detection step and before the single bacterium identification step, the following steps are further included: signal enhancement: for any edge pixel point, if the gray value is not less than a second preset threshold value, the gray value of the pixel point is set to 0, otherwise it is set to 255; then closed operation and open operation are performed to obtain the enhanced edge.

3. The method of claim 1, wherein, after the single bacterium identification step, the following steps are further included: outlier exclusion: by setting a threshold value for at least one of the signal-to-noise ratio, contrast, minimum brightness and brightness variance, false target bacteria are removed.

4. A computer device, comprising: including: a memory for storing a computer program; a processor for executing the computer program to implement the steps of the method of any one of claims 1-3.

5. A computer readable storage medium, characterized in that, the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the steps of the method of any one of claims 1-3.

6. A method for detecting bacterial antimicrobial susceptibility, characterized in that, including the following steps: exposing the bacteria to be tested to different concentrations of antibiotics in the detection cavities, the bottom of the detection cavities being provided with a lensless image sensor; before and after the bacteria to be tested are exposed to different concentrations of antibiotics for a period of time, the microscopic images of the bacterial solutions in the detection cavities are obtained using the lensless image sensor; using the method of any one of claims 1-3, the computer device of claim 4, or the computer readable storage medium of claim 5 to identify single bacteria in the microscopic images and count them; If the proliferation rate of the bacteria to be tested before exposure to an antibiotic at a certain concentration and after exposure to the antibiotic at the certain concentration for a period of time is less than a third preset threshold value, and the proliferation rate of the bacteria to be tested before exposure to the antibiotic at a concentration lower than the certain concentration and after exposure to the antibiotic at the concentration lower than the certain concentration for a period of time is not less than the third preset threshold value, the certain concentration is the minimum inhibitory concentration of the antibiotic to the bacteria.

7. The method of antibiotic susceptibility testing according to claim 6, wherein, The antibiotic is selected from at least one of β-lactams, aminoglycosides, macrolides, tetracyclines, quinolones, sulfonamides, glycopeptides, lincosamides, nitroimidazoles and polypeptide antibiotics.

Images