Method and system for automatically counting microbial colonies

Abstract

The present application relates to a method and system for automatically counting microbial colonies. The present disclosure relates to an automated method for evaluating growth on a plate culture medium, the method comprising: providing a culture medium (202) inoculated with a biological sample; incubating the inoculated culture medium (204); after incubation, obtaining a first image (206) of the inoculated culture medium at a first time; after further incubation, obtaining a second image (206) of the inoculated culture medium at a second time; aligning the first image with the second image so that the pixel coordinates in the second image are approximately the same as the corresponding pixel coordinates in the first image; comparing image features of the second image with image features of the first image; classifying the image features of the second image as colony candidates based on changes in the image features from the first time to the second time; for colony candidates determined to be common microorganisms from the biological sample inoculated on the culture medium, counting the colony candidates; and determining whether the number of counted colonies reaches or exceeds a threshold count value stored in a memory and indicating significant growth.

Description

[0001] This application is a divisional application. The application date of the original application is April 22, 2016, the application number is 201680023381.2, and the name of the invention is “Method and system for automatically counting microbial colonies”.

[0002] CROSS-REFERENCE TO RELATED APPLICATIONS

[0003] This application claims the benefit of the filing dates of U.S. Provisional Application No. 62 / 151,688, filed on April 23, 2015, and U.S. Provisional Application No. 62 / 318,488, filed on April 5, 2016, the disclosures of which are hereby incorporated herein by reference. Background Art

[0004] There has been a greater focus on digital imaging of culture plates used to detect microbial growth. Techniques for imaging plates used to detect microbial growth are described in PCT Publication No. WO2015 / 114121, the entire contents of which are incorporated herein by reference. With this technology, the laboratory personnel no longer need to read the plates by direct visual inspection, but can use high-quality digital images to inspect the plates. Shifting laboratory workflow and decision making to inspecting digital images of culture plates can also improve efficiency. Images can be marked by the operator for further collation by the operator or another person with appropriate skills. Additional images can also be taken and used to guide secondary processes.

[0005] Colony detection, colony counting, colony population differentiation, and colony identification define the goals of modern microbiology imaging systems. Achieving these goals as early as possible achieves the goal of delivering results to patients quickly and providing these results and analyses economically. Automating laboratory workflow and decision making can improve the speed and cost at which these goals can be achieved.

[0006] Although significant advances have been made in imaging techniques for detecting signs of microbial growth, there is still a need to extend such imaging techniques to support automated workflows. Apparatus and methods for examining culture plates for signs of microbial growth are difficult to automate, in part due to the highly visual nature of plate inspection. In this regard, it is desirable to develop techniques that can automatically interpret culture plate images and determine the next steps to be performed (e.g., identification of bacterial colonies, susceptibility testing, etc.) based on the automated interpretation.

[0007] For example, counting bacterium colonies in plate cultures may be difficult, especially when bacterium colonies have different sizes and shapes and contact each other. When growth has reached convergence in some areas of the flat plate, these problems are aggravated. For these reasons, if possible, preferably CFU is counted in the early stage of the incubation process. However, the time for incubation is still needed to allow at least some growth of bacterium colonies. Therefore, on the one hand, the longer the bacterium colonies are allowed to grow, the more contrast they begin to form with the background and each other, and the easier they are to be counted. However, on the other hand, if bacterium colonies are allowed to grow too long, and they begin to fill the flat plate and / or contact each other, thereby forming a convergence area on the flat plate, it will become more difficult to contrast them with each other, making counting more difficult. If people can detect bacterium colonies at the incubation time when bacterium colonies are still small enough to separate from each other, although the contrast is relatively weak, or even when bacterium colonies are large enough to form a convergence area on the flat plate, if people can estimate the colony count, this problem can be solved. Summary of the invention

[0008] One aspect of the present disclosure relates to an automated method for evaluating growth on a plate culture medium, comprising: providing a culture medium inoculated with a biological sample; incubating the inoculated culture medium; after incubation, at a first time (t 1 ) to obtain a first image of the inoculated culture medium; after further incubation, at a second time (t 2 ) obtaining a second image of the inoculated culture medium; aligning the first image with the second image so that the coordinates of pixels in the second image are approximately the same as the coordinates of corresponding pixels in the first image; comparing image features of the second image with image features of the first image; and based on the image features from time t 1 To time t 2 classifying image features of the second image as colony candidates based on changes in the image features of the second image; counting the colony candidates determined to be common microorganisms from the biological sample inoculated on the culture medium; and determining whether the number of counted colonies reaches or exceeds a threshold count value stored in a memory and indicating significant growth.

[0009] In some examples, if the number of colonies counted reaches or exceeds the threshold count value, the method can further include: identifying at least one colony using matrix-assisted laser desorption ionization (MALDI); testing the antimicrobial sensitivity of the at least one colony; and outputting a report containing the MALDI and antimicrobial sensitivity test results. A plurality of threshold count values ​​can be stored in the memory, each threshold count value being associated with a different microorganism.

[0010] In some examples, classifying the image features of the second image as colony candidates may include: determining contrast information of the second image, the contrast information including at least one of spatial contrast information and temporal contrast information, the spatial contrast information indicating differences between pixels of the second image, the temporal contrast information indicating differences between pixels of the second image and corresponding pixels of a previous image; identifying an object in the second image based on the contrast information; and obtaining one or more object features of the identified object from pixel information associated in the first and second images, wherein the object is classified as a colony candidate based on the object features. The method may further include determining for each colony candidate whether the colony candidate is a colony or an artifact based on the pixel information associated with the colony candidate, wherein colony candidates determined to be artifacts are not counted. Determining whether a colony candidate is a colony or an artifact may further include determining whether the colony candidate is present in each of the first and second images and is larger in the second image than in the first image according to a threshold growth factor, wherein colony candidates that are present in both images and are larger in the second image according to at least the threshold growth factor are classified as colonies. Determining whether a colony candidate is a colony or an artifact may further include, for a colony candidate that is present in the second image and not present in the first image, obtaining one or more object features of an identified object from pixel information associated with the object in the second image; determining a probability that the colony candidate is a colony based on the one or more object features; and comparing the determined probability with a predetermined threshold probability value, wherein if the determined probability is greater than the predetermined threshold probability value, the colony candidate is classified as a colony. The method may further include: classifying colony candidates that (i) are present in both images and are not larger in the second image and (ii) are present in the first image and not present in the second image as definite artifacts; classifying colony candidates that are present in each of the first and second images and are larger in the second image according to the threshold growth factor as definite colonies; and calculating an artifact probability value based on a combination of the definite artifacts and the definite colonies, wherein the determined probability that the colony candidate is a colony is further based on the artifact probability.

[0011] In some instances, the object feature may include at least one of an object shape, an object size, an object edge, an object color, a color of an object pixel, a hue, a brightness, and a chromaticity. The method may further include obtaining background feature information, wherein the background feature information includes a culture medium type and a culture medium color, and wherein the object is further classified as a colony candidate based on the background feature information. In some instances, aligning the first image with the second image may include assigning polar coordinates to pixels of each of the first and second images so that the polar coordinates of a pixel in the second image are the same as the polar coordinates of a corresponding pixel in the first image.

[0012] Another aspect of the present disclosure relates to an automated method for estimating the number of colony forming units on a plate culture medium, which has been inoculated with a culture and incubated according to a predetermined pattern, comprising: obtaining a digital image of the plate culture medium after incubating the culture medium; identifying colony candidates in the image from the digital image; linearizing the digital image according to the predetermined pattern; drawing the colony candidates according to pixels of the linearized coordinates of the digital image; and estimating the number of colony forming units on the plate culture medium based on the pixels of the colony candidates in the linearized digital image.

[0013] In some examples, the plate medium from which the image is obtained may have been inoculated with magnetic control beads streaked along a continuous zigzag pattern, wherein the digital image may be linearized according to the zigzag streak pattern, the zigzag streak pattern being the major axis of the linearized image. An initial bead load of the magnetic control beads may be estimated from a plot of colony candidates. Estimating the initial bead load may include: selecting a distance from an origin along the major axis of the linearized image; determining a probability that a colony forming unit is released by the bead at the selected distance; and counting the number of colony forming units present in the digital image, the colony forming unit being further from the origin along the major axis than the selected distance, wherein the estimated initial bead load is equal to the ratio between the determined probability and the counted number of colony forming units. The distance may be selected such that there is no confluent area of ​​microbial growth in the image at a distance further from the origin of the linearized image than the selected distance. The method may further include: selecting a plurality of distances along the major axis of the linearized image; counting the number of colony forming units present in the digital image for each of the selected distances, the colony forming units being further away from the origin of the linearized image along the major axis than the selected distances; and calculating the probability of a colony forming unit being released onto the culture medium by the bead based on the counted number of colony forming units at each distance when a point of the bead containing the colony forming unit comes into contact with the culture medium. Determining the probability of a colony forming unit being released onto the culture medium by the bead at a given distance may be based on the calculated probability of a colony forming unit being released onto the culture medium by the bead when a point of the bead containing the colony forming unit comes into contact with the culture medium.

[0014] In some examples, the method may further include: comparing the digital image to a plurality of distribution models stored in the memory, each distribution model showing an expected distribution of colony forming units on the imaging plate for a given original bead loading and a given probability of the colony forming units being released onto the culture medium when contacted with the culture medium; and determining the original bead loading based at least in part on the compared distribution models.

[0015] In some examples, the method can further include: selecting a distance from an origin along the major axis of the linearized image; determining a fraction of pixels associated with colony candidates at the selected distance; and estimating the raw bead load based on the determined fraction.

[0016] In some examples, the method may further include: obtaining multiple digital images of the plate culture medium after incubating the culture, each digital image containing one or more colony candidates; identifying a digital image in which at least some of the colony candidates form a confluent area; identifying an earlier digital image in which the colony candidates forming a confluent area in the digital image are not combined to form a confluent area; and evaluating the number of colony forming units in the confluent area based on the earlier digital image.

[0017] Another aspect of the present disclosure relates to a computer-readable memory storage medium having coded program instructions thereon, the coded program instructions being configured to cause a processor to implement a method. The method may be any of the above-described methods for evaluating microbial growth on a plate culture medium or for estimating the number of colony forming units on a plate culture medium.

[0018] Yet another aspect of the present disclosure relates to a system for evaluating growth in a culture medium inoculated with a biological sample. The system includes an image acquisition device for capturing a digital image of the culture medium; a memory; and one or more processors operable to execute instructions for implementing a method. In some embodiments, the memory may store information about the expected amount of microbial growth of one or more different organisms in one or more different culture media, and the method implemented by the executed instructions may be any of the above-described methods for evaluating microbial growth on a plate culture medium. In other embodiments, the memory may store information about a pattern for inoculating the culture medium with the biological sample, and the method implemented by the executed instructions may be any of the above-described methods for evaluating the number of colony forming units on a plate culture medium. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a schematic diagram of a system for imaging analysis and testing cultures according to one aspect of the present disclosure.

[0020] Figure 2 is a flow chart illustrating an automated laboratory work procedure for imaging analysis and testing cultures according to one aspect of the present disclosure.

[0021] Figure 3A , 3B 3C are images showing a visual representation of colony morphology over time according to one aspect of the present disclosure.

[0022] Figure 3D and 3E are images showing the visual representation of the colonies under different lighting conditions.

[0023] Figure 4 is a flow chart of an example procedure for counting colonies according to one aspect of the present disclosure.

[0024] Figure 5 is a flow chart of an example procedure for collecting a general list of colony candidates according to one aspect of the present disclosure.

[0025] Figure 6 is a flow chart of an example procedure for classifying colony candidates according to one aspect of the present disclosure.

[0026] Figure 7 is a flow chart of an example procedure for counting colonies based on statistical analysis according to one aspect of the present disclosure.

[0027] Figure 8 is a streak mode image according to one aspect of the present disclosure, for streaking a plate culture medium with a sample.

[0028] Fig.9A is an illustration of an image of identified colony candidates according to one aspect of the present disclosure.

[0029] Fig. 9B yes Figure 8 An illustration of the dash pattern shown in FIG.

[0030] Fig. 10A and 10B is a graphical representation of a distribution model for colony forming units (CFU) according to one aspect of the present disclosure.

[0031] Fig.11A According to one aspect of the present disclosure, Figure 8 A graphic representation of the confluence ratio of the slab spindles is shown in FIG.

[0032] Fig. 11B is a graphical representation of a colony growth simulation according to one aspect of the present disclosure.

[0033] Fig.12 is a side to side depiction of two images of a plate culture medium with colony growth.

[0034] Fig.13 is a series of images taken over time according to one aspect of the present disclosure.

[0035] Fig.14 According to one aspect of the present disclosure picture.

[0036] Fig.15A , 15B15C is a graphical depiction of isolation factor determination according to one aspect of the present disclosure.

[0037] Fig.16A and 16B A portion of the imaging plate is shown, with a magnified and reoriented image of the sample colonies imaged.

[0038] Fig. 16C According to one aspect of the present disclosure Fig. 16B The polar coordinate transformed images of the parts are enlarged respectively.

[0039] Fig.17 Yes Figure 2 A flowchart comparing the timeline of a procedure with the timeline of a comparable manually implemented process. DETAILED DESCRIPTION

[0040] The present disclosure provides apparatus and methods for identifying and analyzing microbial growth on plate culture medium, based at least in part on the number of identified colonies counted in one or more digital images of the plate culture medium. Many of the methods described herein can be fully or partially automated, such as being integrated as part of a fully or partially automated laboratory workflow.

[0041] The system described herein can be implemented in an optical system for imaging a microbial sample, for identifying microorganisms and detecting microbial growth of such microorganisms. There are many such commercially available systems, which are not described in detail herein. One example is the BD Kiestra TM ReadA compact intelligent culture and imaging system. Other exemplary systems include those described in PCT Publication No. WO2015 / 114121 and U.S. Patent Publication No. 2015 / 0299639, the entire contents of which are incorporated herein by reference. Such optical imaging platforms are well known to those skilled in the art and are not described in detail herein.

[0042] Figure 1 is a schematic diagram of a system 100 having a processing module 110 and an image acquisition device 120 (e.g., a camera) for providing high quality imaging of a plate culture. The processing module and image acquisition device may be further connected to, and thereby further interact with, other system components, such as an incubation module (not shown) for incubating the plate culture to allow the growth of cultures inoculated on the plate culture. Such connections may be fully or partially automated using a tracking system that receives samples for incubation and transports them to an incubator and then between the incubator and the image acquisition device.

[0043] The processing module 110 can instruct other components of the system 100 to perform tasks based on various types of information processing. The processor 110 can be hardware that performs one or more operations. The processor 110 can be any standard processor, such as a central processing unit (CPU), or can be a dedicated processor, such as an application-specific integrated circuit (ASIC) or a field programmable gate array (FPGA). Although a processor block is shown, the system 100 can also include multiple processors that can or can not operate in parallel, or other dedicated logic and memory for storing and tracking information related to sample containers in the incubator and / or image acquisition device 120. In this regard, the processing unit can track and / or store several types of information about the sample in the system 100, including but not limited to the location of the sample in the system (incubator or image acquisition device, location and / or orientation there, etc.), incubation time, pixel information of the captured image, type of sample, type of culture medium, preventive treatment information (e.g., dangerous samples), etc. In this regard, the processor can be capable of fully or partially automating the various procedures described herein. In one embodiment, instructions for executing the procedures described herein may be stored on a non-transitory computer-readable medium (eg, a software program).

[0044] Figure 2 FIG. 2 is a flow chart illustrating an exemplary automated laboratory process 200 for imaging, analyzing, and optionally testing cultures. The process 200 may be implemented by an automated microbiology laboratory system, such as a Kiestra TM Total LabAutomation or Kiestra TM Work Cell Automation, both manufactured by Becton, Dickenson & Co. The exemplary system includes interconnected modules, each module being configured to perform one or more steps of procedure 200 .

[0045] At 202, a culture medium is provided and inoculated with the biological sample. The culture medium can be an optically transparent container so that the biological sample can be observed in the container when illuminated from various angles. Inoculation can follow a predetermined pattern. Scribing patterns and automated methods for streaking samples onto a flat plate are well known to those skilled in the art. An automated method utilizes magnetically controlled beads to streak samples onto a flat plate.

[0046] In some examples of the present disclosure, the beads are arranged according to a zigzag pattern (e.g., see Figure 8) is scribed on the plate. The starting point and end point of the zigzag pattern can be located at opposite ends of the plate (e.g., separated by a distance approximately equal to the diameter of the plate). In this example, the "major axis" of the plate can be considered to be a straight line starting at the starting point of the zigzag pattern and ending at the end point.

[0047] At 204, the culture medium is incubated to allow growth of the biological sample.

[0048] At 206, one or more digital images of the culture medium and the biological sample are captured. As will be described in more detail below, during the incubation process (e.g., at the beginning of the incubation, at a time in the middle of the incubation, at the end of the incubation), multiple digital imaging of the culture medium can be performed so that changes in the culture medium can be observed and analyzed. Imaging of the culture medium can involve removing the culture medium from the incubator. In the case of taking multiple images of the culture medium at different times, the culture medium can be returned to the incubator for further incubation between imaging periods.

[0049] At 208, the biological sample is analyzed based on the information from the captured digital image. Analyzing the digital image may involve analyzing the pixel information contained in the image. In some cases, the pixel information may be analyzed on a pixel basis. In other cases, the pixel information may be analyzed on a block basis. In further cases, the pixel may be analyzed based on the entire region of the pixel, thereby inferring the pixel information of each pixel in the region by combining the information of each pixel, selecting sample pixels or by utilizing other statistical methods, such as the statistical histogram operation described in more detail below. In this article, the operation described as being applied to "pixel" may be applied to blocks or other groups of pixels similarly, and the term "pixel" is intended to include such applications.

[0050] The analysis may involve determining whether growth is detected in the culture medium. From an image analysis perspective, growth can be detected in the image by identifying the imaged object (based on the differences between the object and its immediate surroundings) and then identifying changes in the object over time. As described in more detail herein, these differences and changes are two forms of "contrast." In addition to detecting growth, the image analysis at 208 may further involve quantifying the amount of growth detected, identifying different colonies, identifying sister colonies, etc.

[0051] At 210, it is determined whether the biological sample (particularly the identified sister colonies) exhibits quantitatively significant growth. If no growth or an insignificant amount of growth is found, then the process 200 may proceed to 220, where a final report is output. In the case of proceeding from 210 to 220, the final report will likely indicate a lack of significant growth, or report growth of normal flora.

[0052] If it is determined that the biological sample exhibits quantitatively significant growth, then one or more colonies can be picked from the image based on the previous analysis at 212. Picking colonies can be a fully automated process, where each colony picked is sampled and tested. Alternatively, picking colonies can be a partially automated process, where multiple colony candidates are automatically identified and visually presented to an operator in a digital image so that the operator can input the selection of one or more candidates for sampling and further testing. The sampling of the selected or picked colonies itself can be automated by the system.

[0053] At 214, the sampled colonies are prepared for further testing, such as by plating the samples in an organism suspension. At 216, the samples are tested using matrix-assisted laser desorption ionization (MALDI) imaging to identify the type of sample sampled from the original culture medium. At 218, the samples are also or alternatively subjected to antibiotic sensitivity testing (AST) to determine possible treatments for the identified samples.

[0054] At 220, the test results are output in a final report. The report may include MALDI and AST results. As mentioned above, the report may also indicate the quantification of sample growth. Thus, the automated system is able to start with the inoculated culture medium and generate a final report on the sample found in the culture with little or no additional input.

[0055] In the program, such as Figure 2 In the example program of, the colonies detected and identified are often referred to as colony forming units (CFU). CFU is a microscopic object that starts as one or several bacteria. Quantitative growth can be determined based on the number of CFUs that can be counted in a plate. However, as explained above, the number of CFUs cannot always be counted directly. For example, CFUs can contact or mix with each other to form a confluent area, and no independent unit is counted. In this case, the present disclosure provides a method for estimating colony counts, which is based on known information-such as a streak pattern applied to a plate culture medium, knowledge of how fast the streak tool is unloaded when it passes through the plate streak, a standard size and growth rate of a specific type of colony counted, etc.-and a combination of assay information collected from one or more digital images of the plate. This estimation can be automated by the system and program described above.

[0056] Determining whether the estimated growth is significant can be inferred by comparing the estimated colony count to a predetermined threshold. More than one threshold value can be set for a given plate and / or colony. For example, colony growth may be affected by the culture medium in which the colonies are grown. Therefore, what constitutes significant growth in one culture medium may not constitute significant growth in another culture medium, and different threshold values ​​can be set. In addition, while testing for one type of bacteria may not be guaranteed until a high threshold is met, testing for particularly harmful or dangerous bacteria (e.g., Group B Streptococcus in testing of pregnant women) can be guaranteed even if a low threshold is met, in some cases, even as low as one counted colony. Therefore, it should be understood that the system is capable of storing multiple thresholds and applying each of those different thresholds where appropriate.

[0057] Bacteria grow over time to form colonies. The earlier the bacteria are placed in the plate, the fewer bacteria will be detected, and accordingly, the smaller the colonies are and the lower the contrast with the background. In other words, smaller colony size produces smaller signal, and smaller signal on a constant background results in less contrast. This is reflected by the following equation:

[0058] (1)

[0059] Contrast can play an important role in identifying objects, such as CFU or other artifacts in an image. An object can be detected in an image if it is significantly different from its surroundings in terms of brightness, color and / or texture. Once an object has been detected, analysis can also involve identifying the type of object that has been detected. This identification can also rely on contrast measurements, such as the smoothness of the edges of the object being identified, or the uniformity (or lack of uniformity) of the color and / or brightness of the object. The contrast must be large enough to overcome the image noise (background signal) in order to be detected by the image sensor.

[0060] Human contrast perception (governed by Weber's law) is limited. Under optimal conditions, the human eye is able to detect 1% differences in brightness levels. The quality and confidence of image measurements (e.g., brightness, color, contrast) can be characterized by the signal-to-noise ratio (SNR) of the measurements, where an SNR value of 100 (or 40db) will match human detection capabilities, independent of pixel intensity. Digital imaging techniques that utilize high SNR imaging information and known SNR for each pixel of information can allow detection of colonies even though those colonies would still be invisible to the human eye.

[0061] In the present disclosure, contrast can be collected in at least two ways: spatially and temporally. Spatial contrast or local contrast quantifies the difference in color or brightness between a given area (e.g., a pixel, a group of adjacent pixels) in a single image and its surroundings. Temporal contrast or time contrast quantifies the difference in color or brightness of a given area of ​​one image taken at a different time relative to the same area in another image. The formula that specifies time contrast is similar to the formula used for spatial contrast:

[0062] (2)

[0063] Among them, t 2 Yes 1 At a later time. Both spatial and temporal contrast of a given image can be used to identify objects. Identified objects can then be further tested to determine their significance (e.g., whether they are CFU, normal flora, dust, etc.).

[0064] Figure 3A and 3B Visual demonstration of the effect that temporal contrast can exert on imaged samples is provided. Figure 3A The images shown in Figure 1 were captured at different time points (left to right, top row to bottom row) and show the total growth in the sample. Figure 3A is significant, but from Figure 3B The growth of the corresponding contrast time image is even more significant and can be noticed even earlier in the sequence. For clarity, Figure 3C Shows Figure 3B If you can Figure 3C As can be seen in Figure 1, the longer a portion of a colony has been imaged, the brighter the spot it forms in the contrast image. Therefore, the centroid of each colony can be represented by the center or peak of the colony's brightness. Image data acquired over time can therefore reveal important information about changes in colony morphology.

[0065] To maximize the spatial or temporal contrast of an object relative to its background, the system can capture images using different incident lighting on different backgrounds. For example, any of top lighting, bottom lighting, or side lighting can be used on a black or white background.

[0066] Figure 3D and 3E Provides visual demonstration of the effect that lighting conditions can have on imaged samples. Captured using top lighting Figure 3D , and captured at approximately the same time (e.g., close enough in time that no noticeable or significant growth occurs) using bottom illumination Figure 3E As can be seen, in Figure 3D and 3E In the sample, each image contains several colonies, but due to Figure 3E Back- or bottom-illumination in the image allows additional information about the colonies to be seen (in this case, hemolysis), whereas Figure 3D It is difficult to capture that same information in an image.

[0067] At a given point in time, multiple images can be captured under a variety of lighting conditions. Images can be captured using different light sources, which are spectrally different due to the illumination brightness level, illumination angle, and / or filters placed between the object and the sensor (e.g., red light, green light, and blue light filters). In this way, image acquisition conditions can be changed in terms of light source position (e.g., top, side, bottom), background (e.g., black, white, any color, any intensity), and spectrum (e.g., red channel, green channel, blue channel). For example, a first image can be captured using top illumination and a black background, a second image can be captured using side illumination and a black background, and a third image can be captured using bottom illumination and no background (i.e., white background). In addition, specific algorithms can be used to generate a set of variable image acquisition conditions in order to maximize spatial contrast usage. By changing the image acquisition conditions according to a given order and / or time range, these or other algorithms can also be useful for maximizing time contrast. Some such algorithms are described in PCT Publication No. WO2015 / 114121.

[0068] Figure 4 is a flow chart illustrating an example procedure for imaging a flat panel based at least in part on comparative analysis. Figure 4 The program can be considered as Figure 2 An exemplary subroutine of the program 200, thereby at least partially utilizing Figure 4 Program execution Figure 2 206 and 208.

[0069] At 402, at time t 1 A first digital image is captured. Time t 1 This may be a time after the incubation process has begun so that the bacteria in the imaging plate have at least begun to form some visible colonies, but those colonies have not yet begun to touch or overlap each other.

[0070] At 404, coordinates are assigned to one or more pixels of the first digital image. In some cases, the coordinates may be polar coordinates, with radial coordinates extending from a center point of the imaging flat panel and angular coordinates about the center point. The coordinates may be used in later steps to help align the first digital image of the flat panel with other digital images taken from different angles and / or at different times. In some cases, the imaging flat panel may have specific landmarks (e.g., off-center points or lines), so that coordinates of pixels covering the landmarks in the first image may be assigned to pixels covering the same landmarks in other images. In other cases, the image itself may be considered as a feature for future alignment.

[0071] At 406, at time t 2 A second digital image is captured. Time t 2 Yes 1 At a later time, the colonies in the imaging plate have had a chance to grow further. 1 Additional colonies that are too small to be seen at t 2 Moreover, the presence of colonies at time t 2 There is a possibility that they have begun to touch or overlap with each other.

[0072] At 408, the second digital image is aligned with the first digital image based on the previously assigned coordinates.Aligning the images may further involve normalization and standardization of the images, for example using the methods and systems described in PCT Publication No. WO2015 / 114121.

[0073] At 410, a general list of colony candidates is collected based on the first and second digital images. The general list of colony candidates may identify any objects in the first and second digital images that may be colonies that may be desirable for further testing (e.g., Figure 1 in the program).

[0074] At 412, each colony candidate included in the overall list is classified. Classifying colony candidates involves identifying, for each candidate, whether the candidate is actually an artifact or a colony. As described in more detail below, in some cases, it may not be possible to determine exactly whether a given candidate is an artifact or a colony. However, a probabilistic or ambiguous determination may be made, and the candidate may be classified based on the determination.

[0075] At 414, the classified colony candidates identified as colonies are counted. As explained in more detail below, counting colonies is not always unambiguous due to confluence between individual colonies. Therefore, the present disclosure provides methods and techniques for counting based on statistical analysis of the second digital image.

[0076] At 416, a final report containing the estimated colony count is output. The final report may optionally include additional information that affects the accuracy of the estimated colony count, such as clustering probabilities between the counted colonies. Clustering refers to the confluence of colonies, resulting in a cluster such that individual colonies cannot be identified independently. In some cases, clustering probabilities may be reported as long as the clustering probability exceeds a pre-set threshold (e.g., 50%).

[0077] Figure 5 is a flow chart illustrating an example process 500 for collecting a global list of colony candidates. Figure 5 The program can be considered as Figure 4 An exemplary subroutine of the program 400, thereby at least partially utilizing Figure 5 Program execution Figure 4 of 410.

[0078] At 502, contrast information for the second digital image is determined. The contrast information may be collected on a pixel-by-pixel basis. For example, a pixel of the second digital image may be compared to a corresponding pixel of the first digital image (at the same coordinates) to determine that there is a temporal contrast. Additionally, adjacent pixels of the second digital image may be compared to each other or to other pixels known to be background pixels to determine that there is a spatial contrast. Changes in pixel color and / or brightness indicate contrast, and the magnitude of such changes from one image to another or from one pixel (or pixel region) to another pixel (or pixel region) may be measured, calculated, estimated, or otherwise determined. Where temporal and spatial contrasts are determined for a given image, an overall contrast for a given pixel of the image may be determined based on a combination of the spatial and temporal contrasts (e.g., averages, weighted averages) of the given pixel.

[0079] At 504, objects in the second digital image are identified based on the contrast information calculated at 502. Adjacent pixels of the second digital image having similar contrast information may be considered to belong to the same object. For example, if the brightness difference between adjacent pixels and their background or the difference between the pixel brightness and their brightness in the first digital image is approximately the same (e.g., within a predetermined threshold amount), the pixels may be considered to belong to the same object. As an example, the system may assign "1" to any pixel with significant contrast (e.g., greater than a threshold amount), and then identify a group of adjacent pixels all assigned to "1" as an object. Objects may be given a specific tag or mask so that pixels with the same tag share certain features. During later processing, the tag may help distinguish the object from other objects and / or the background.

[0080] Identifying objects in a digital image may involve segmenting or separating the digital image into multiple regions (e.g., foreground and background). The goal of segmentation is to turn an image into a representation of multiple parts so that the parts can be more easily analyzed. Image segmentation is used to locate objects of interest in an image. [Add cross reference here]

[0081] At 506, features of a given object (identified at 504) may be characterized. Characterization of object features may involve inferring descriptive statistics of the object (e.g., area, reflectivity, size, optical density, color, plate position, etc.). Descriptive statistics may ultimately quantitatively describe certain features of a collection of information collected about the object (e.g., from SHQI images, from contrast images). Such information may be evaluated as a function of species, concentration, mixture, time, and culture medium. However, in at least some cases, characterizing an object may start with a collection of qualitative information about the features of the object, whereby the qualitative information is then quantitatively represented. Table 1 below provides a list of example features that may be qualitatively evaluated and subsequently converted to a quantitative representation:

[0082] Table 1: Qualitative properties of objects and the criteria used to convert properties quantitatively

[0083]

[0084]

[0085] Some features of an object, such as shape or the time until it is visually observed, can be measured once for the object as a whole. Other features can be measured several times (e.g., for each pixel, for each row of pixels with the same y-coordinate, for each column of pixels with the same x-coordinate, for each ray of pixels with the same angular coordinate, for a circle of pixels with the same radial coordinate), and then combined into a single measurement, for example using a histogram. For example, color can be measured for each pixel, growth rate or size can be measured for each row, column, ray or circle of pixels, etc.

[0086] At 508, it is determined whether the object is a colony candidate based on the characterized features. Colony candidate determination may involve inputting quantitative features (e.g., scores shown in Table 1 above) or a subset thereof into a classifier. The classifier may include a confusion matrix for implementing a supervised machine learning algorithm or a matching matrix for implementing an unsupervised machine learning algorithm to evaluate the object. In the case where the object will be distinguished from a limited set (e.g., two or three) of possible organisms (in which case the algorithm can be trained on a relatively limited set of training data), supervised learning may be preferred. In contrast, in the case where the object will be distinguished from the entire database of possible organisms, unsupervised learning may be preferred, in which case it will be difficult to provide comprehensive or even sufficient training data. In the case of a confusion matrix or a matching matrix, differences within a certain range can be measured numerically. For example, for a given pair of objects, "0" may mean that the two objects should be distinguished from each other, while "1" may mean that the objects are difficult to distinguish from each other.

[0087] If the object is determined to be a colony candidate at 508, it is added to the overall list of colony candidates at 510. Otherwise, process 500 ends (and may continue with 412 of process 400) without adding the object to the overall list.

[0088] Additional procedures and subroutines for identifying colony candidates based on comparative information of digital images are discussed in commonly owned and co-pending application entitled "COLONY CONTRASTGATHERING," the entire contents of which are incorporated herein by reference.

[0089] Figure 6 is a flow chart illustrating an example process 600 for classifying colony candidates. Figure 6 The program can be considered as Figure 4 An exemplary subroutine of the program 400, thereby at least partially utilizing Figure 6 Program execution Figure 4 of 412. As Figure 4 The subroutine of process 600 can be iteratively applied to each colony candidate appearing on the overall list.

[0090] At 602, it is determined whether the colony candidate is growing. Growth may be indicated by (a) the colony candidate being present in the first and second digital images and (b) the size of the colony candidate being significantly larger in the second digital image than in the first digital image. Whether a size change is considered significant may be determined by comparing the size change to a predetermined growth threshold, whereby changes that meet or exceed the growth threshold are considered significant.

[0091] If the colony candidate is determined to be growing, the colony candidate is verified and identified as a colony. Otherwise, the process 600 continues at 604 where it is determined whether the colony candidate is present in both the first and second digital images.

[0092] If the colony candidate is determined to be present in both images (meaning there is no significant growth between the two images), then the colony candidate is identified as an artifact. Otherwise, process 600 continues at 606 where it is determined whether the colony candidate is present in the second digital image.

[0093] If a colony candidate does not exist in the second image (meaning it only exists in the first image and then disappears), the colony candidate is identified as an artifact (e.g., 1 and t 2 Otherwise, further analysis is performed to determine whether the colony candidate is only present at time t 1 Is it an artifact such as the colonies that did not grow to be visible at time t 1 and t 2 A piece of dust was blown onto the flat panel.

[0094] At 608, given that it is known that the colony candidate does not appear in the first digital image, the colony candidate is characterized based solely on information from the second digital image. Characterization can rely on static features such as color, size, shape, and surface (described above in conjunction with Figure 5 506 of FIG. 10 ).

[0095] At 610, an overall probability that a colony candidate is in fact a colony is determined based at least in part on the characterization. For example, the characterized features of the object may be compared to an expected colony type (i.e., a colony type that is included in the overall list and is Figure 6 In one embodiment, the characteristics of the colony type counted in the sample are compared with Figure 5 The comparison is performed in the same manner as step 508, whereby "0" will mean that the object is of the expected colony type and "1" will mean that the object is not of the expected colony type, and the number between "0" and "1" will indicate the probability that the object is of the expected colony type, also referred to as the colony probability.

[0096] In some cases, the colony probability may be the overall probability at 610. Optionally, the decision at 610 may be further based on information collected about artifacts in the image. Such information may include artifact probability, which measures the likelihood that an object in the image is an artifact. Figure 6 In the example of 1 and t 2No growth between, only appears in t 1 and does not appear in t 2 ) or colonies (e.g., at t 1 and t 2 The objects (i.e., objects that have grown significantly between the colonies) are provided as input at 612 to determine the artifact probability. The artifact probability at 612 is then combined with the colony probability to produce an overall probability. In one embodiment, the colony probability and the artifact probability are combined according to the following equation:

[0097] (3) P(total) = P(colony) × (1-P(illusion))

[0098] At 614, the overall probability is compared to a determined threshold (eg, 50%). If the overall probability meets or exceeds the threshold, the colony candidate is then identified as a colony. Otherwise, the colony candidate is identified as an artifact.

[0099] although Figure 5 and 6 Programs are used to classify the identified colonies, but those programs do not ensure that each colony is an independent colony and is not a confluence of multiple colonies. Therefore, the colony candidates identified as colonies do not have to be counted as discrete elements, but are counted using estimation techniques.

[0100] Figure 7 is a flow chart illustrating an example procedure 700 for counting colonies based on statistical analysis. Figure 7 The program can be considered as Figure 4 An example subroutine of the program 400, thereby at least partially utilizing Figure 7 Program execution Figure 4 of 414. Figure 7 The procedure assumes that the colonies are streaked onto the imaging plate using magnetically controlled beads according to a predetermined streaking pattern. Those skilled in the art will appreciate that in addition to those described below, Figure 7 The underlying concept of the procedure can be adapted to a variety of streaking media, techniques, and formats.

[0101] At 702, the second digital image is linearized according to a streaking pattern along which the imaging plate is streaked by the magnetron beads. To illustrate the streaking pattern, Figure 8 An image of a sample grown in a flat plate culture medium is shown. The image is digitally superimposed with a zigzag pattern that starts toward the lower right of the image and ends toward the upper left. The zigzag pattern indicates the streaking pattern of the magnetically controlled beads used to streak the culture medium.

[0102] For clarity, linearizing a digital image can be viewed as plotting the pixels of a zigzag pattern along the x-axis of the linearized image, so that the zigzag pattern is expanded into a straight line along the x-axis. For each pixel of the digital image, the zigzag pattern is not directly superimposed, and the pixel can be associated with the closest portion of the zigzag pattern to the pixel (e.g., along the y-axis of the linearized image). The linearized image is used to indicate the density gradient of colonies deposited by the beads onto the culture medium as the beads move along the streak pattern over time.

[0103] At 704, each colony candidate is plotted along the linearized coordinates of the second digital image. In other words, the density gradient of the colonies deposited onto the culture medium is evaluated using the linearized image. Fig.9A is a graphic representation of a previously identified colony candidate (e.g., from Figure 5 Procedure 500). Fig. 9B is a diagram of a zigzag pattern. Fig.9A and 9B As shown in the figure, the distance of each colony candidate from the pattern origin (the far left end of the image) along the zigzag pattern can be calculated.

[0104] At 706, the original bead load (concentration measured in CFU per milliliter) is estimated based on the mapped colonies present in the linearized second digital image. 2 The surface area (SA) of the path-marked beads is measured as “SA” and the calculations used to estimate the initial bead loading are presented herein.

[0105] As an initial point, note that for a given point on the bead surface, on average, the point will contact the plate once for every "B" mm that the bead travels along the zigzag pattern, where:

[0106] (4)

[0107] If it is assumed that a given spot is loaded with colony forming units (CFU), when the given spot comes into contact with the culture medium, the CFU is released into the culture medium (P R ) can be represented as a number between 0 and 1.

[0108] By the time the bead has travelled a distance x (measured in mm) along the streaked path, the probability that a CFU at a given point has been released (P NR (x)):

[0109] (5)P NR (x) = (1-P R ) x / B (18)

[0110] As the beads advance and release CFUs, the CFU load present on the beads decreases. At a given time in which a bead has advanced such a great distance x along the streaking pattern, the total CFU load present on the bead can be characterized as K(x), at which time K(x) can be further expressed as the original bead load K 0 A function of (i.e., the CFU load of the beads before streaking mode begins and thus before any CFUs are released):

[0111] (6) K(x) = K 0 ×P NR (x) = K 0 ×(1-P R ) x / B (19)

[0112] K(x) can also be estimated based on the linearized digital image. In particular, it can be assumed that all CFU initially loaded onto the bead will be released into the medium by completing the streaking pattern, and therefore, the remaining load on the bead at any given distance x can be characterized as The number of colonies traversing distance x is shown in the digital image (actually, the upper limit of the summation should be the length of the streak pattern, not ∞, but assuming that all CFUs are released at the end of the streak pattern, an upper limit of infinity is equally acceptable).

[0113] With an estimate of K(x), this estimate can be substituted into the above equation to resolve the original bead loading K 0 In fact, for any given distance x (e.g., x1, x2, x3, etc.) at which K(x) can be estimated, K 0 can also be solved independently, as shown in the following equation:

[0114] (7)

[0115] Although in the above example it is assumed that the release probability P R , but further note that the above series of equations can also be used to utilize two or more along the dash pattern The estimated analytical P R The following is the analysis of P using the estimated values ​​K(x) of the distances xl and x2 R An example of a formula.

[0116] (8)

[0117] P has been resolved R , using P according to the following equation R The determined values ​​of K(x) can be used to estimate K 0 :

[0118] (9)

[0119] It should be noted that the more K(x) values ​​are estimated, the greater the R and K 0 The more accurate the estimate of K can become. Therefore, although the above example only uses xl and x2 to estimate K 0 , but other examples may use other distances (e.g., x3).

[0120] Alternatively, P is determined based on known characteristics of the colony, culture medium, beads, or any combination thereof. R It can be a predetermined value, in which case K can be estimated based on the estimated value K(x) at a single given distance x. 0 .

[0121] Distribution Model

[0122] In addition to the above calculations, the number of colonies on the imaged plates can be estimated based on the comparison between the digital image and the distribution model. Fig. 10A and 10B The release probability of a given bead (P R )’s CFU distribution model. Fig. 10A and 10B Each of the images shows the distribution of CFU for varying original bead loadings (ranging from a small original loading of e.g. 10 2 To the rightmost image, the large original bead load is, for example, 10 5 ). The distribution model can also be modified to account for variables such as release probability and colony size. Fig. 10A and 10B In the example, the release probability is set to 0.14. Fig. 10A In the , the colony size was set to 1.66 mm in diameter. Fig. 10B Therefore, when the release probability and colony size of a given sample are known, the distribution model can be used to estimate the original bead loading for images with similar distribution appearance.

[0123] Confluence ratio

[0124] The confluence ratio can also be used to improve colony count estimates. The confluence ratio is the fraction of pixels at a specific distance along the major axis of the plate that are associated with colony candidates. The confluence ratio can be characterized according to the following formula: (10)

[0126] Confluence a％ = x, so

[0127] Fig.11AThe confluence ratio of the plate measured along the major axis of the streaking pattern from the origin to the end point is illustrated. Fig.11A In the example of , the expected CFU on the plate is 4800. Fig. 11B The results of the simulation are described, with an initial load of 4800 CFU having a release probability of 0.185 and a colony size of approximately 2 mm in diameter. As can be seen in this example, Fig.11A The plate shows confluence and Fig. 11B The simulations are quite similar to each other.

[0128] The confluence ratio can also be used to identify tangent zero crossings, which are points along the major axis where the confluence region substantially or typically ends (e.g., there are more independent colonies than confluent colonies, the confluence region makes up less than 50% of the pixels along a line passing through the point and perpendicular to the major axis, etc.).

[0129] It will be appreciated from the above examples that the expected confluence ratio along the major axis of the plate depends largely on the initial load (CFU / ml), the size of the isolated colonies at a given time when the plate is imaged, and the given incubation time. To emphasize these factors, Fig.12 is a side-by-side depiction of two plates with similar confluence ratios but distinctly different CFU loads. The top plate contains a total of approximately 39,500 CFU of S. aureus, while the bottom plate contains approximately 305 CFU of P. aeruginosa. Significantly, the confluence ratios of these plates are similar to Fig.11A The plates shown in are approximately the same (containing approximately 4,800 CFU of Serratia marcescens after 18 hours of incubation on blood agar medium).

[0130] Time Series Analysis

[0131] Another way to assess the CFU content within a confluent region is to perform a time series analysis by splitting the confluent region into individual colonies using previous images. As noted above, colonies that are confluent at a given time may still be independent and individually countable at an earlier time. Thus, for at least some colonies, when the confluent condition has not yet been met, images of the confluent region from an earlier incubation time may be used for analysis (e.g., running a segmentation program to establish a time series). This analysis can then be used to track changes over time to help maintain the identification of individual colonies at subsequent times.

[0132] Fig.13 illustrates a series of images taken over time. Fig.13In the figure, each row contains a digital image (left), a segmentation result (middle), and a comparison image (right) at a specific time point during the incubation: at 8 hours, 12 hours, 16 hours, and 20 hours (from bottom to top). The segmentation and comparison images are described in more detail in the commonly owned, co-pending patent application entitled "Colony Versus Aggregation", the disclosure of which is incorporated herein in its entirety.

[0133] Those skilled in the art will appreciate that the results of the above colony estimation techniques, distribution models, confluence ratios, and time series analysis can be used in conjunction with each other to provide a more accurate estimate, or to confirm the accuracy of a previous estimate.

[0134] Object characteristics

[0135] As above combined Figure 5 As discussed, features of an object on an imaging plate may be characterized as part of an image analysis performed on the imaging plate. The features characterized may include static features (associated with a single image) and dynamic features (associated with multiple images).

[0136] Static features are intended to reflect the properties of an object and / or the surrounding context at a given time. Static features include the following:

[0137] (i) Center of gravity: This is a static feature that provides the center of gravity of the imaged object in a coordinate space (e.g., xy coordinates, polar coordinates). The center of gravity of an object, like the polar coordinates of an object, provides invariance in the feature set under given lighting and background conditions. The center of gravity can be obtained by first determining the weighted center of mass of all colonies in the image (M is a binary mask of all detected colonies). The weighted center of mass can be determined based on the assumption that each pixel of the image has an equal value. The center of gravity of a given colony can then be described in xy coordinates by the following equation (where E = {p|p∈M} (E is the binary mask of the current colony), the range of the x-coordinate is, the range of the y-coordinate is, and each pixel is a unit):

[0138] (11)

[0139] (ii) Polar coordinates: This is also a static feature and can be used to further characterize the location on the imaging plate, such as the center of gravity. Typically, polar coordinates are measured along the radial axis (d) and the angular axis (θ), where the coordinates of the center of the plate are igv (x，y) The coordinates d and θ (d in millimeters and θ in degrees) are given by the following equation (where k is the pixel density, corresponding to pixels in millimeters, and the "barcode" is a landmark feature of the imaging plate to ensure alignment of the plate with previous and / or future images):

[0140] (12)d=k×dist(igν (x，y) , 0 (x，y) )

[0141] (13)θ=angle(barcode, 0 (x，y) ,igv (x，y) )

[0142] (iii) Image vectors: Two-dimensional polar coordinates can in turn be converted to one-dimensional image vectors. Image vectors can characterize the pixel intensity of an image as a function of a radial axis (usually, the center of a colony has the highest intensity) and / or an angular axis. In many cases, image vectors can be more accurate in classifying similarities / differences between imaged objects.

[0143] (iv) Morphometric characteristics, which describe the shape and size of a given object.

[0144] (a) Area: This is a morphometric characteristic and can be determined based on the number of pixels in the imaged object (also called a “blob”), without counting holes in the object. When pixel density is available, area can be measured in actual size (e.g., mm 2 ). Otherwise, when pixel density is not available, the total number of pixels may indicate the size, and the pixel density (k) is set equal to 1. In one embodiment, the area is calculated using the following equation:

[0145] (14)A=k 2 ×∑ p∈E 1

[0146] (b) Perimeter: The perimeter of an object is also a morphometric feature and can be determined by measuring the edges of the object and adding together the total length of the edges (e.g., a single pixel with an area of ​​1 square unit has a perimeter of 4 units). As with area, length can be measured in pixel units (e.g., when k is not available) or in physical length (e.g., when k is available). In some cases, perimeter can also include the perimeter of any holes in the object. In addition, by counting interior corners as Instead of 2, the staircase effect (which occurs when diagonal edges are digitized as ladder-shaped boxes) can be compensated. In one embodiment, the perimeter can be determined using the following equation:

[0147] (15)P=k×∑ p∈E q(n p )

[0148] (16)

[0149] (17) If:

[0150] {∑(t∈M, l∈M, r∈M, b∈M)=2, (l∈M≠r∈M), (t∈M≠b∈M)}

[0151] (p is interior and p is corner)

[0152] Then:

[0153] Otherwise: q(n p )=4-∑(t∈M,l∈M,r∈M,b∈M)

[0154] (c) Roundness: The roundness of an object is also a morphometric feature and can be determined based on a combination of area and perimeter. In one embodiment, the roundness is calculated using the following equation:

[0155] (18)

[0156] (d) Radius coefficient of variation (RCV): This is also a morphometric feature and is used to calculate the radius of the object by taking the average radius of the object in all N directions or angles θ extending from the center of gravity. The standard deviation of the radius σ R The ratio between indicates the change in the radius of the object. In one embodiment, the value can be calculated using the following equation:

[0157] (19)

[0158] (20)

[0159] (twenty one)

[0160] (v) Contextual features that describe the topographical relationship of the object under surveillance to other objects under observation and to the edges of the plate walls. For example, in the case of imaging bacterial colonies, a contextual feature of the colony could be whether the colony is free, has limited free space, or competes with other surrounding colonies for resources. Such features tend to help classify colonies growing in the same perceived environment and / or distinguish colonies growing in different environments.

[0161] (a) Influence area: This is a characteristic of the environment that takes into account the space between an object and its neighbors and predicts the area that the object under analysis is likely to occupy without other distinct objects first occupying the same area. The influence area can be expressed as In the form of a graph, such as Fig.14, the influence region (shaded) is shown based on the distance d between colony 1401 and its neighboring colonies, such as 1405. In one embodiment, the distance (D) from the edge of the object to the edge of the influence region is NC ) can be characterized by the following equation:

[0162] (twenty two)

[0163] (b) Distance to the flat wall: This is a characteristic of the environment. The distance from the edge of the object to the nearest flat wall (D pw ). In one embodiment, the distance can be characterized by the following equation:

[0164] (twenty three)

[0165] (c) Isolation Factor: This is a characteristic of the environment that characterizes the relative isolation of a given object based on its size and distance to the nearest edge (e.g., another object, a flat wall). Fig.15A -C clarifies the isolation factor aspect. Fig.15A The case where the closest edge is the distance d from the colony to the plate wall is illustrated. Fig. 15B and 15C The case where the closest edge belongs to another colony is illustrated. In this case, a circle is drawn with the colony under analysis as the center and then expanded (smaller at first, as in Fig. 15B As in, then larger, as in Fig. 15C until the circle touches the adjacent colony. Fig.15A In the embodiment of -C, the isolation factor (IF) can be characterized by the following equation:

[0166] (twenty four)

[0167] (d) Neighborhood Occupancy Ratio: This is a characteristic of the environment that characterizes the bounded area of ​​a given object plate. The area fraction of the impact region (V) within a given distance d. In one embodiment, the neighbor occupancy ratio (OR) can be characterized by the following equation (wherein for this equation,

[0168] E={p|p∈V,dist(p,igv (x，y) )<d}):

[0169] (25)

[0170] (e) Relative neighborhood occupancy ratio: In some cases, the average radius of the objects is multiplied by a predetermined factor. can be derived for a given distance d. The result is the relative neighbor occupancy ratio (RNOR) and for a given factor x it can be derived using the following equation:

[0171] (26)RNOR(x)=NOR(d)

[0172] (vi) Spectral features, which describe the optical properties of a given object. Color (red, green and blue light channels; hue, brightness and chroma, or any other color space transformation), texture and contrast (over time and / or in space) are examples of such features. Spectral features can be derived from images captured at various time points during the incubation and / or under various lighting conditions using colony masks, and can be further combined with the spectral features for a given colony. The impact area is associated.

[0173] (a) Channel image: This is a spectral feature in which a specific color channel (e.g., red (R), green (G), blue (B)) is used to spectrally resolve the image.

[0174] (b) Brightness: This is also a spectral feature that is used to characterize the brightness of an image using RGB channels as input.

[0175] (c) Hue: This is a spectral feature where areas of an image are characterized as appearing similar to a perceived color (e.g., red, yellow, green, blue) or a combination thereof. 2 ) is usually characterized by the following equation:

[0176] (27)H 2 = atan 2(β, α)

[0177] (28)

[0178] (29)

[0179] (d) Chroma: This is a spectral feature that characterizes the color of an image region relative to its brightness if the region were illuminated similarly to white light. 2 ) is usually characterized by the following equation:

[0180] (30)

[0181] Maximum contrast:

[0182] (31)[Keep maximum contrast]

[0183] (vii) Background features, which describe changes in the culture medium near the object being analyzed. For example, in the case of imaging bacterial colonies, microbial growth around the colonies can cause changes (e.g., signs of hemolysis, changes in pH, or specific enzyme reactions).

[0184] Dynamic features are intended to reflect changes in object properties and / or surrounding context over time. Time series processing allows static features to be correlated over time. Independent first-order and second-order derivatives of these features provide the instantaneous "speed" and "acceleration" (or reaching equilibrium or deceleration) of these feature changes characterized over time. Examples of dynamic features include the following:

[0185] (i) Time series processing, for tracking the above static features over time. Each feature measured at a given incubation time can be referenced according to its relative incubation time to allow the feature to be a related feature measured at a subsequent incubation time. Time series images can be used to detect objects, such as the appearance and growth of CFU over time, as described above. The time points for imaging can be pre-set or defined by an automated process based on the uninterrupted analysis of previously captured object images. At each time point, the image can be of a given acquisition configuration, either for the entire sequence of a single acquisition configuration, or as a complete sequence of images captured from multiple acquisition configurations.

[0186] (ii) Discrete first and second derivatives of the above characteristics are used to provide the instantaneous velocity and acceleration (either reaching equilibrium, or decelerating) of these characteristics over time (e.g., tracking growth rate as described above):

[0187] (a) Velocity: The first derivative of a feature with respect to time. The velocity (V) of a feature x can be expressed in terms of x units per hour based on the following equation, where Δt is the time span expressed in hours:

[0188] (32)

[0189] (33)

[0190] (34)

[0191] (b) Acceleration: The second derivative of the characteristic with time, which is also the first derivative of velocity. Acceleration (A) can be characterized based on the following equation:

[0192] (35)

[0193] Dynamic features can include changes in the color acquisition signature of an object during incubation. Dynamic color changes allow for further differentiation of objects that may exhibit the same color at a given time point but different colors at different time points. Thus, different temporal signatures for the color acquisition of two objects will help to infer that the two objects are different (e.g., different organisms). Conversely, if the color changes of two objects over time are the same (e.g., follow the same path in color space), the two objects can be considered the same (e.g., the same type or species of organism).

[0194] The above image features are measured from the object or the environment of the object and are intended to capture the specificity of the organism grown in various media and incubation conditions. The listed features are not meant to be exhaustive, and anyone knowledgeable in the art can modify, amplify or limit the feature set based on the diversity of various known image processing based features known in the art.

[0195] Image features may be collected for each pixel, group of pixels, object, or group of objects in the image. The distribution of the collected features may be constructed in a histogram to more generally characterize a region of the image or even the entire image. The histogram itself may rely on several statistical features to analyze or otherwise process the incoming image feature data, such as those described in the commonly owned, co-pending application entitled "Colony Versus Clustering."

[0196] Image alignment

[0197] When multiple images are taken over time, very precise alignment of the images is required in order to obtain valid time estimates from them. Such alignment can be achieved by mechanical alignment devices and / or algorithms (e.g., image tracking, image matching). Those knowledgeable in the art are aware of these schemes and techniques to achieve this goal.

[0198] For example, where multiple images of an object on a flatbed are acquired, the coordinates of the object's location may be determined. Image data of the object acquired at a subsequent time may then be correlated with previous image data based on the coordinates and then used to determine changes in the object over time.

[0199] In order to use images quickly and meaningfully (e.g., when used as input to a classifier), it is important to store the images in a spatial reference system to maximize their invariance. When the basic shape descriptor of a colony is a common circle, a polar coordinate system can be used to store colony images. When a colony is first detected, the colony centroid can be identified as the center of the colony's location. This center point can later be used as the origin of the polar coordinate transformation of each subsequent image of the colony. Fig.16AAn enlarged portion of an imaging plate is shown with a center point "O". Two rays "A" and "B" extending from point "O" are shown superimposed on the image (for clarity). Each ray intersects a corresponding colony (circled). Fig.16A The circled colony is shown in even more detail. Fig. 16B In images 1611 and 1612. Fig. 16B , image 1611 (the colony intersected by ray "A") is reoriented into image 1613 ("A'") so that the radial axis of image 1613 is aligned with the radial axis of image 1612, so that the leftmost portion of the reoriented image is separated Fig.16A The rightmost portion of the reoriented image is closest to point "O", and the rightmost portion of the reoriented image is farthest from point "O". This polar coordinate reorientation allows easier analysis of colonies in different orientations (with respect to factors such as illumination) of the imaged plate.

[0200] exist Fig. 16C In Fig. 16B Each of the images 1611, 1612 and 1613 completes polar coordinate conversion. In the polar coordinate conversion images 1621, 1622 and 1623, Fig. 16C The radial axes (extending from the center of each corresponding imaged colony) of the corresponding re-oriented images 1611, 1612 and 1613 are drawn from left to right in the image of FIG. 16A, and the angular axes (of the corresponding colonies) are drawn from top to bottom.

[0201] For each polar coordinate image, for example, a summary one-dimensional vector set can be generated using shape features and / or histogram features (e.g., the mean and / or standard deviation of the color or intensity of the object) along the radial axis and / or the angular axis. Even though shape and histogram features are generally invariant when rotation is considered, it is possible that some texture features will show significant changes when rotated; therefore, invariance is not guaranteed. Therefore, since the texture differences of objects can then be used to distinguish one from another, there is a significant benefit in presenting each colony image from the same viewpoint or angle lighting. Since lighting conditions generally show changes associated with angular position around the center of the plate being imaged, a ray passing through the center of the colony and the plate (shown as a line in each of images 1611, 1612, and 1613 of FIG. 16) can serve as the origin (θ) for the polar coordinate transformation of each image.

[0202] Improve SNR

[0203] Under typical lighting conditions, photon scatter noise (the statistical variation in the arrival speed of incident photons on the sensor) limits the SNR of the detection system. Modern sensors have a full well capacity of about 1,700 to about 1,900 electrons per effective square micrometer. Therefore, when imaging an object on a flat panel, the primary concern is not the number of pixels used to image the object, but the area of ​​the sensor space covered by the object. Increasing the area of ​​the sensor improves the SNR for imaging the object.

[0204] By specifying the photon noise Capturing images under lighting conditions can improve image quality without saturating the sensor (the maximum number of photons that can be recorded per pixel per frame). To maximize the SNR, image averaging techniques are often used. These techniques are used to process images with significant brightness (or color) differences, since the SNR in dark areas is much lower than that in bright areas, as shown by the following formula:

[0205] (36)

[0206] Where I is the average current produced by the electron flow at the sensor. Because color is perceived due to differences in the absorption / reflection of matter and light across the electromagnetic spectrum, the confidence in the captured color will depend on the system's ability to record intensity with a high SNR. Image sensors (e.g., CCD sensors, CMOS sensors, etc.) are known to those skilled in the art and are not described in detail herein.

[0207] To overcome the traditional SNR imaging limitations, the imaging system can perform an analysis of the imaged panel during image acquisition and adjust the lighting conditions and exposure time in real time based on the analysis. This process is described in PCT Publication No. WO2015 / 114121, which is incorporated by reference, and is generally referred to as supervised high quality imaging (SHQI). The system can also customize imaging conditions for various brightness regions of the panel in different color channels.

[0208] For a given pixel x, y of an image, the SNR information of the pixel acquired during the current frame N can be combined with the SNR information of the same pixel acquired during the previous or subsequent acquisition frames (e.g., N-1, N+1). For example, the combined SNR is represented by the following formula:

[0209] (37)

[0210] After updating the image data with the new acquisition, the acquisition system is able to predict the best next acquisition time that will maximize the SNR subject to environmental constraints (e.g., the minimum required SNR per pixel in the region of interest). For example, averaging 5 images captured in unsaturated conditions will result in a dark area SNR improvement. (10% of maximum intensity), when merging information from two images captured in bright and dark conditions, optimal illumination will result in a dark region SNR improvement in both acquisitions only.

[0211] application

[0212] This disclosure is primarily based on testing performed in saline at various dilutions to simulate typical urine reporting volumes (CFU / ml, Bucket Group). Suspensions of each isolate were adjusted to 0.5 McFarland Standard and used to collect urine in BD Urine Vacutainer tubes (Cat. No. 364951) at an estimated 1×10 6 , 1×10 5 , 5×10 4 , 1×10 4 , 1×10 3 and 1×10 2 Dilutions were prepared using a CFU / ml suspension. Sample tubes were processed using Kiestra InoqulA (WCA1) using a standard urine streak pattern - #4 zigzag pattern (0.01 ml was dispensed per plate).

[0213] All acquired images are corrected for lens geometry and chromatic aberrations, spectrally balanced with known object pixel size, standardized illumination conditions, and high signal-to-noise ratio per pixel per band. Suitable cameras for use in the methods and systems described herein are well known to those skilled in the art and are not described in detail herein. As an example, capturing images of a 90 mm plate with a 4-megapixel camera should allow counting of up to 30 colonies / mm when the colonies are in the 100 μm diameter range with appropriate contrast. 2 Local density (>10 5 CFU / plate).

[0214] The magnetic ball used to mark the sample plate has a diameter of 5 mm, a circumference of 15.7 mm, and a surface area of ​​78 mm 2 The average surface in contact with the culture medium is approximately 4 mm 2 , which represents a contact disk with a diameter of approximately 2.2 mm.

[0215] The following media were used to evaluate the contrast of colonies grown on them:

[0216] TSAII 5% Blood of Sheep (BAP): A non-selective medium with broad application for urine cultures.

[0217] BAV: For colony counts and presumptive IDs based on colony morphology and hemolysis.

[0218] MacConkey II Agar (MAC): A selective medium for most common Gram-negative UTI pathogens. MAC is used to differentiate lactose-producing colonies. MAC also inhibits Proteus colonization. BAP and MAC are commonly used extensively for urine culture. Some media are not recommended for colony counts due to partial inhibition of some Gram-negative bacteria.

[0219] Polymyxin Nalidixic Acid Agar (CNA): A selective medium for most common Gram-positive UTI pathogens. If there is an overgrowth of Gram-negative colonies, CAN is not routinely used for urine culture as MAC, but is helpful in identifying colonies.

[0220] CHROMAgar Orientation (CHROM): A non-selective medium widely used for urine culture. CHROM is used for colony counts and ID based on colony color and morphology. E. coli and Enterococcus are recognized by the medium and no confirmatory testing is required. CHROMA is used less frequently than BAP due to cost reasons. CLED medium is also used for mixed samples.

[0221] Cystine lactose electrolyte deficient agar (CLED): For colony counts and presumptive ID of lactose-fermenting urinary pathogens.

[0222] Sample processing BD Kiestra TM InoqulA TM Used to automate the processing of bacteriological specimens in order to standardize and ensure consistent, high-quality streaking. BD Kiestra TM InoqulA TM The sample processor utilizes magnetic bead technology to streak culture plates using customizable patterns.

[0223] Fig.17 A timeline of automating the testing process 1700 is shown (eg, Figure 2200) with a timeline of a comparable manually performed test process 1705. Each process begins with a sample 1710, 1715 received at the laboratory for testing. Each process then proceeds to incubation 1720, 1725, during which the sample can be imaged several times. In the automated process, an automated evaluation 1730 is performed after nearly 12 hours of incubation, after which it can be clearly determined whether there is significant growth or no growth (or normal growth) 1740 in the sample. As shown from the above disclosure, the use of statistical methods to classify and count colonies in the automated process greatly improves the ability to determine whether there is significant growth, even after only 12 hours. In contrast, in the manual process, the manual evaluation 1735 cannot be performed until nearly 24 hours into the incubation process. Only after 24 hours can it be clearly determined whether there is significant growth or no growth (or normal growth) 1745 in the sample.

[0224] The use of automated processes also allows for more rapid AST and MALDI testing. Such testing 1750 in an automated process can begin shortly after the original assessment 1730, and by the 24 hour mark, results can be obtained 1760 and reported 1775. In contrast, such testing 1755 in a manual process is often not performed until close to the 36 hour mark, and takes an additional 8 to 12 hours to complete before the data can be reviewed 1765 and reported 1775.

[0225] In summary, the manual testing process 1705 was shown to take up to 48 hours, requiring an 18-24 hour incubation period, only thereafter evaluating the growth of the plates, and furthermore having no way to track how long the sample has been in incubation. In contrast, because the automated testing process 1700 is able to detect even fairly weak contrasts between colonies (compared to the background and to each other), and is able to perform imaging and incubations without the microbiologist having to keep track of timing, only 12-18 hours of incubation are necessary before the sample can be identified and prepared for further testing (e.g., AST, MALDI), and the entire process can be completed in about 24 hours or less. Thus, the automated process of the present disclosure, supplemented by the contrast processing described herein, provides for more rapid sample testing without adversely affecting the quality or accuracy of the test results.

[0226] Although the present invention has been described herein with reference to specific embodiments, it is to be understood that these embodiments are merely illustrative of the principles and applications of the present invention. It is therefore to be understood that many modifications may be made to the illustrative embodiments, and other arrangements may be designed without departing from the spirit and scope of the present invention as defined in the appended claims.

Claims

1. An automated method for estimating the number of colony forming units on a plate medium which has been inoculated with a culture according to a predetermined pattern and incubated, said method include: obtaining a digital image of the plate culture after incubating the culture, wherein the plate culture from which the image is obtained is inoculated with magnetically controlled beads streaked along a continuous zigzag pattern, wherein the digital image is linearized according to the zigzag streak pattern, wherein the zigzag streak pattern is a major axis of the linearized image; from the digital image, identifying colony candidates in the image; linearizing the digital image according to the predetermined pattern; drawing the colony candidates according to pixels of linearized coordinates of the digital image; wherein the original bead loading of the magnetic control beads is estimated from the mapping of colony candidates and estimating the number of colony forming units on the plate medium based on the pixels of the colony candidates in the linearized digital image; Wherein estimating the initial bead loading comprises: Selecting a distance from the origin along the principal axis of the linearized image; determining the probability of a colony forming unit being released by the bead at a selected distance; counting the number of colony forming units present in the digital image that are further away from the origin along the major axis than the selected distance, and wherein the estimated original bead load is equal to the ratio between the determined probability and the counted number of colony forming units.

2. The method of claim 1, wherein the distance is selected such that no confluent areas of microbial growth exist in the image at distances further from the origin of the linearized image than the selected distance.

3. The method according to claim 2, further comprising: include: selecting a plurality of distances along a major axis of the linearized image; counting, for each selected distance, a number of colony forming units present in the digital image that are further away from an origin of the linearized image along a major axis than the selected distance; and when one point of the bead containing the colony forming unit comes into contact with the plate medium, calculating the probability that the colony forming unit is released by the bead onto the plate medium based on the number of colony forming units counted for each distance, wherein when a point of the bead containing the colony forming unit comes into contact with the plate medium, the probability of the colony forming unit being released by the bead at a given distance is determined based on the calculated probability of the colony forming unit being released by the bead onto the plate medium.

4. The method according to any one of claims 2 to 3, further comprising: include: comparing the digital image to a plurality of distribution models stored in a memory, each distribution model illustrating an expected distribution of colony forming units on an imaged plate for a given initial bead loading and a given probability of the colony forming units being released onto the culture medium upon contact with the culture medium; and The initial bead loading is determined based at least in part on the compared distribution models.

5. The method according to any one of claims 2 to 3, further comprising: include: selecting a distance from an origin along the major axis of the linearized image; determining a fraction of pixels associated with colony candidates at the selected distance; and The original bead loading is estimated based on the determined fraction.

6. The method according to any one of claims 2 to 3, further comprising: include: After incubating the culture, obtaining a plurality of digital images of the plate culture medium, each digital image comprising one or more colony candidates; identifying a digital image in which at least some of the colony candidates form a confluent region; identifying an earlier digital image in which the colony candidates forming a confluent region in the digital image were not combined to form a confluent region; and The number of colony forming units in the confluent area is estimated based on the earlier digital image.

Images