System and method for monitoring bacterial growth of bacterial colonies and predicting colony biomass

Abstract

To provide an imaging method for early detection of microbial growth.SOLUTION: An automated method for evaluating microbial growth on a plate medium is provided, where the method uses images to determine colony biomass, and the colony biomass determines when the colony can be picked for analysis for identification or antibiotic susceptibility testing. If the sample source is not a pure sample source, additional incubation may be required to permit an increase in biomass of the colonies prior to picking.SELECTED DRAWING: Figure 19

Description

[Technical field]

[0001] [Related Applications] This application is a continuation of U.S. Provisional Patent Application No. 62 / 782,513, filed December 20, 2018. No. 6,313,563, filed on the same date herewith, the entire disclosure of which is incorporated herein by reference. It shall be so. [Background technology]

[0002] Digital imaging of culture plates to detect microbial growth is gaining increasing attention A method for imaging culture plates and detecting microbial growth thereon is described in "A System and Method for Image Acquisition Using Supervised High Quality Imaging Patent Document 1 published on August 6, 2016 under the name "Colony Contrast Gathering" Patent Document 2, published on October 27, 2016 under the title "A Method and System m for Automated Microbial Colony counting from Streaked Sample on Plated Media” This is described in Patent Document 3, published on October 27, 2017 under the title " The entirety of these three documents are incorporated herein by reference. are all assigned to the same assignee as this application.

[0003] In Patent Document 1, a colony object is identified by controlling the signal-to-noise ratio, and Techniques are described for distinguishing these objects from non-colony artifacts and background. Patent Document 2 describes a system that analyzes images of culture plates based on an automated imaging technique. A stem and method are described. Colony objects are identified early in the process of promoting microbial growth on a nutrient medium, and changes in these identified objects are observed over time and under conditions that support the growth of bacterial colonies. According to the system and method described in Patent Document 3, the identified colonies are counted, it is determined whether these colonies are from the same microorganism, and it is further determined whether the colony count is equal to or greater than a predetermined number. The detection, counting, population differentiation, and identification of colonies define the purpose of modern microbiological imaging systems. By realizing such a purpose as early as possible, the goal of quickly delivering results to patients and economically providing the results and analysis is achieved. By automating the research workflow and decision-making, the speed and cost at which the goal can be achieved can be improved. Although significant progress has been made in imaging techniques for detecting evidence of microbial growth, there is still a need to expand the imaging techniques to support an automated workflow. Devices and methods for examining culture plates to reveal microbial growth are difficult to automate, partly due to the very visual nature of plate confirmation. In this regard, it is desirable to develop a technique that can automatically analyze culture plate images and determine the next steps (e.g., colony identification, susceptibility testing, etc.) based on the automated analysis. According to the system and method described in Patent Document 3, the identified colonies are counted, it is determined whether these colonies are from the same microorganism, and it is further determined whether the colony count is equal to or greater than a predetermined number. The detection, counting, population differentiation, and identification of colonies define the purpose of modern microbiological imaging systems. By realizing such a purpose as early as possible, the goal of quickly delivering results to patients and economically providing the results and analysis is achieved. By automating the research workflow and decision-making, the speed and cost at which the goal can be achieved can be improved. Although significant progress has been made in imaging techniques for detecting evidence of microbial growth, there is still a need to expand the imaging techniques to support an automated workflow. Devices and methods for examining culture plates to reveal microbial growth are difficult to automate, partly due to the very visual nature of plate confirmation. In this regard, it is desirable to develop a technique that can automatically analyze culture plate images and determine the next steps (e.g., colony identification, susceptibility testing, etc.) based on the automated analysis.

[0004] The detection, counting, population differentiation, and identification of colonies define the purpose of modern microbiological imaging systems. By realizing such a purpose as early as possible, the goal of quickly delivering results to patients and economically providing the results and analysis is achieved. By automating the research workflow and decision-making, the speed and cost at which the goal can be achieved can be improved. Although significant progress has been made in imaging techniques for detecting evidence of microbial growth, there is still a need to expand the imaging techniques to support an automated workflow. Devices and methods for examining culture plates to reveal microbial growth are difficult to automate, partly due to the very visual nature of plate confirmation. In this regard, it is desirable to develop a technique that can automatically analyze culture plate images and determine the next steps (e.g., colony identification, susceptibility testing, etc.) based on the automated analysis. Although significant progress has been made in imaging techniques for detecting evidence of microbial growth, there is still a need to expand the imaging techniques to support an automated workflow. Devices and methods for examining culture plates to reveal microbial growth are difficult to automate, partly due to the very visual nature of plate confirmation. In this regard, it is desirable to develop a technique that can automatically analyze culture plate images and determine the next steps (e.g., colony identification, susceptibility testing, etc.) based on the automated analysis. The detection, counting, population differentiation, and identification of colonies define the purpose of modern microbiological imaging systems. By realizing such a purpose as early as possible, the goal of quickly delivering results to patients and economically providing the results and analysis is achieved. By automating the research workflow and decision-making, the speed and cost at which the goal can be achieved can be improved. Although significant progress has been made in imaging techniques for detecting evidence of microbial growth, there is still a need to expand the imaging techniques to support an automated workflow. Devices and methods for examining culture plates to reveal microbial growth are difficult to automate, partly due to the very visual nature of plate confirmation. In this regard, it is desirable to develop a technique that can automatically analyze culture plate images and determine the next steps (e.g., colony identification, susceptibility testing, etc.) based on the automated analysis.

[0005] Although significant progress has been made in imaging techniques for detecting evidence of microbial growth, there is still a need to expand the imaging techniques to support an automated workflow. Devices and methods for examining culture plates to reveal microbial growth are difficult to automate, partly due to the very visual nature of plate confirmation. In this regard, it is desirable to develop a technique that can automatically analyze culture plate images and determine the next steps (e.g., colony identification, susceptibility testing, etc.) based on the automated analysis. Although significant progress has been made in imaging techniques for detecting evidence of microbial growth, there is still a need to expand the imaging techniques to support an automated workflow. Devices and methods for examining culture plates to reveal microbial growth are difficult to automate, partly due to the very visual nature of plate confirmation. In this regard, it is desirable to develop a technique that can automatically analyze culture plate images and determine the next steps (e.g., colony identification, susceptibility testing, etc.) based on the automated analysis. Devices and methods for examining culture plates to reveal microbial growth are difficult to automate, partly due to the very visual nature of plate confirmation. In this regard, it is desirable to develop a technique that can automatically analyze culture plate images and determine the next steps (e.g., colony identification, susceptibility testing, etc.) based on the automated analysis. Devices and methods for examining culture plates to reveal microbial growth are difficult to automate, partly due to the very visual nature of plate confirmation. In this regard, it is desirable to develop a technique that can automatically analyze culture plate images and determine the next steps (e.g., colony identification, susceptibility testing, etc.) based on the automated analysis. Although significant progress has been made in imaging techniques for detecting evidence of microbial growth, there is still a need to expand the imaging techniques to support an automated workflow. Devices and methods for examining culture plates to reveal microbial growth are difficult to automate, partly due to the very visual nature of plate confirmation. In this regard, it is desirable to develop a technique that can automatically analyze culture plate images and determine the next steps (e.g., colony identification, susceptibility testing, etc.) based on the automated analysis. Although significant progress has been made in imaging techniques for detecting evidence of microbial growth, there is still a need to expand the imaging techniques to support an automated workflow. Devices and methods for examining culture plates to reveal microbial growth are difficult to automate, partly due to the very visual nature of plate confirmation. In this regard, it is desirable to develop a technique that can automatically analyze culture plate images and determine the next steps (e.g., colony identification, susceptibility testing, etc.) based on the automated analysis. Although significant progress has been made in imaging techniques for detecting evidence of microbial growth, there is still a need to expand the imaging techniques to support an automated workflow. Devices and methods for examining culture plates to reveal microbial growth are difficult to automate, partly due to the very visual nature of plate confirmation. In this regard, it is desirable to develop a technique that can automatically analyze culture plate images and determine the next steps (e.g., colony identification, susceptibility testing, etc.) based on the automated analysis.

[0006] In particular, when colonies are of different sizes and shapes and are in contact with each other, it can be difficult to identify and distinguish colonies in a plate culture. When colonies "grow into" each other, there is a risk of selecting microorganisms from adjacent colonies, making it more difficult to select the target colony when the adjacent colonies are different microorganisms. To select a sample of colonies for downstream processing, both an appropriate amount of colonies and the purity of the target colonies to avoid selecting multiple species of microorganisms are required. If multiple species of microorganisms are selected, the results of subsequent tests will be compromised. These problems worsen when growth has already overlapped in some areas of the plate. For these reasons, it is preferable to identify colonies and determine growth early in the process if possible. On the other hand, a culture time is still required to allow at least some growth of the colonies. Thus, on the one hand, the longer the time allowed for colonies to grow, the more colonies there are, creating contrast with the background and with each other, making it easier to identify them. However, on the other hand, if colonies are allowed to grow too long and begin to fill the plate and / or come into contact with each other, it becomes more difficult to select pure colonies. If it is possible to detect colonies at a culture time when the colonies are still small enough to be separated from each other (despite relatively poor contrast) but large enough to provide an appropriate amount of sample for testing, can this problem be minimized or even solved? This problem can be minimized or even solved if it is possible to detect colonies at a culture time when the colonies are still small enough to be separated from each other (despite relatively poor contrast) but large enough to provide an appropriate amount of sample for testing. This problem can be minimized or even solved if it is possible to detect colonies at a culture time when the colonies are still small enough to be separated from each other (despite relatively poor contrast) but large enough to provide an appropriate amount of sample for testing. This problem can be minimized or even solved if it is possible to detect colonies at a culture time when the colonies are still small enough to be separated from each other (despite relatively poor contrast) but large enough to provide an appropriate amount of sample for testing. This problem can be minimized or even solved if it is possible to detect colonies at a culture time when the colonies are still small enough to be separated from each other (despite relatively poor contrast) but large enough to provide an appropriate amount of sample for testing. This problem can be minimized or even solved if it is possible to detect colonies at a culture time when the colonies are still small enough to be separated from each other (despite relatively poor contrast) but large enough to provide an appropriate amount of sample for testing. This problem can be minimized or even solved if it is possible to detect colonies at a culture time when the colonies are still small enough to be separated from each other (despite relatively poor contrast) but large enough to provide an appropriate amount of sample for testing. This problem can be minimized or even solved if it is possible to detect colonies at a culture time when the colonies are still small enough to be separated from each other (despite relatively poor contrast) but large enough to provide an appropriate amount of sample for testing. This problem can be minimized or even solved if it is possible to detect colonies at a culture time when the colonies are still small enough to be separated from each other (despite relatively poor contrast) but large enough to provide an appropriate amount of sample for testing. This problem can be minimized or even solved if it is possible to detect colonies at a culture time when the colonies are still small enough to be separated from each other (despite relatively poor contrast) but large enough to provide an appropriate amount of sample for testing. It is possible.

Prior Art Documents

Patent Documents

[0007]

Patent Document 1

Patent Document 2

Patent Document 3

Summary of the Invention

[0008] Described herein is an automated method for evaluating the growth of microorganisms in a plate medium. According to this method, a biological sample is inoculated and a culture medium is prepared in an optically substantially transparent container. The inoculated culture medium is cultured in an incubator. The inoculated culture medium is placed in an optically transparent container having the inoculated culture medium in a digital imaging device. The automated method also includes obtaining a first digital image of the cultured medium at a first time point (t )). The first digital image has a plurality of pixels. The automated method also includes determining the coordinates of the pixels in the first digital image with respect to the transparent container having the inoculated culture medium. The automated method also includes removing the transparent container having the inoculated culture medium from the digital imaging device and installing the inoculated culture medium in the incubator for further culture. The automated method also includes installing the transparent container having the inoculated culture medium in the digital imaging device after further culture. The automated method also includes, at a second time point (t ), the inoculated culture medium is cultured. The automated method also includes obtaining a second digital image of the inoculated culture medium at the second time point (t 0 ). The second digital image has a plurality of pixels. The automated method also includes determining the coordinates of the pixels in the second digital image with respect to the transparent container having the inoculated culture medium. The automated method also includes analyzing the first and second digital images to determine the growth of the microorganisms in the culture medium. The first digital image has a plurality of pixels. The automated method also includes determining the coordinates of the pixels in the first digital image with respect to the transparent container having the inoculated culture medium. The automated method also includes removing the transparent container having the inoculated culture medium from the digital imaging device and installing the inoculated culture medium in the incubator for further culture. The automated method also includes installing the transparent container having the inoculated culture medium in the digital imaging device after further culture. The automated method also includes, at a second time point (t ), the inoculated culture medium is cultured. x )). The second digital image has a plurality of pixels. It also includes obtaining a second digital image. The second digital image has a plurality of pixels . The automated method also includes aligning the first digital image with the second digital image . Thereby, the coordinates of the pixels in the second digital image will correspond to the coordinates of the corresponding pixels in the first digital image. The automated method also includes comparing the pixels in the second digital image with the corresponding pixels in the first digital image . The automated method also includes identifying the pixels that have changed between the first digital image and the second digital image . The pixels that have not changed between the first digital image and the second digital image indicate the background. The automated method also includes determining which of the identified pixels in the second digital image have a predetermined level of reference contrast in relation to the pixels indicating the background . The automated method also includes identifying one or more objects in the second digital image. Each object has the predetermined level of reference contrast in relation to the pixels indicating the background and has pixels that are separated from each other by the background pixels . The automated method also includes associating the identified objects with biomass . The automated method also determines whether the biological sample is identified as a pure sample, and if it is identified as a pure sample, further determines whether the biomass exceeds a first criterion . The first criterion is a predetermined area of the culture medium covered by the identified object. If the area of the identified object exceeds the first criterion, it also includes selecting at least a part of the growing material for further analysis . The automated method also includes inoculating in an optically transparent container when the biological sample is not a pure sample ​​​​​​ It also includes further culturing the resulting culture medium.

Brief Description of the Drawings

[0009]

Figure 1

Figure 2

Figure 3A.3B

Figure 3C

Figure 3D.3E

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13A.13B.13C

Figure 14

Figure 15A.15B.15C

Figure 16A

Figure 16B.16C

Figure 17

Figure 18

Figure 19

[0010] The present disclosure relates to a method for detecting the growth of microorganisms in a plate medium by detecting one or more digital signals of the plate medium. and analyzing the identified areas based at least in part on contrast detected in the digital image. Many of the methods described herein are fully can be fully or partially automated, for example in a fully or partially automated laboratory It becomes part of the workflow.

[0011] The systems described herein are useful for identifying microorganisms and detecting the microbial growth of such microorganisms. The present invention can be implemented in an optical system for imaging a microbiological sample for the purpose of There are many such commercially available systems not described in detail. One example is the BD Kiestra TM ReadA is a compact intelligent incubation and imaging system. Other exemplary systems are described in U.S. Patent Application Publication No. 2015 / 023634 and U.S. Patent Application Publication No. 2015 / 023634. No. 99639, the entireties of which are incorporated herein by reference. Such optical imaging platforms are known to those skilled in the art. and will not be described in detail here.

[0012] FIG. 1 illustrates a processing module 110 and an image acquisition module 112 for providing high quality imaging of plate media. FIG. 1 is a schematic diagram of a system 100 having an acquisition device 120 (e.g., a camera). The module and image acquisition device allow the growth of cultures grown on plate media. In order to achieve this, other system components such as a culture module (not shown) for culturing the plate medium may be included. Further connected to the components, thereby enabling further interaction with them Such a connection can be fully or partially automated using a tracking system that receives specimens for culturing, transfers them to an incubator, and then to an image acquisition device between the incubator and the image acquisition device ..

[0013] The processing module 110 can instruct other components of the system 100 to perform tasks based on the processing of various types of information. The processor 110 can be hardware that executes one or more operations. The processor 110 can also be any standard processor such as a central processing unit (CPU), or a dedicated processor such as an application-specific integrated circuit (ASIC) or a field-programmable gate array (FPGA) ). Although one processor block is shown, the system 100 can include multiple processors that may or may not operate in parallel, or other dedicated logic and memory for storing and tracking information related to the sample containers within the incubator and / or the image acquisition device 120. In this regard, the processing unit can track and / or store several types of information related to the samples within the system 100, including but not limited to the location of the samples within the system (incubator or image acquisition device, internal location and / or orientation, etc.), culture time, pixel information of the captured images, sample type, culture medium type, preventive treatment information (e.g., harmful samples), etc. In this regard, the processor may be capable of fully or partially automating the various routines described herein. In one embodiment, the routines described herein ​​​​​​​​​​​​​​ Instructions for performing a culturing can be stored on a non - transitory computer - readable medium (e.g., software program).

[0014] Figure 2 is a flowchart showing an exemplary automated laboratory routine 200 for imaging, analyzing, and optionally inspecting a culture. Routine 200 can be performed by an automated microbiology laboratory system such as BD Kiestra (trademark) Total Lab Automation or BD Kiestra (trademark) Work Cell Automation. An exemplary system includes a plurality of interconnected modules, and each module performs one or more steps of routine 200.

[0015] At 202, a culture medium is prepared and a biological sample is inoculated. The culture medium can be an optically transparent container. Thereby, the biological sample can be observed in its container as it is illuminated from various angles. The inoculation can follow a predetermined pattern. Streaking patterns and automated methods for streaking a sample onto a plate are known to those skilled in the art and will not be described in detail herein. One automated method streaks a sample onto a plate using magnetically controlled beads. At 204, the medium is cultured to enable the growth of the biological sample.

[0016] At 206, one or more digital images of the medium and the biological sample are captured. As will be described in more detail below, digital imaging of the medium can be performed multiple times (e.g., at the start of the culture, at a mid - point during the culture, at the end of the culture) during the culturing ​Changes in the ground can be observed and analyzed. Imaging of the culture medium can include the step of taking the culture medium out of the incubator. When images of the culture medium are acquired at multiple time points, the culture medium may be returned to the incubator for further culturing between two imaging sessions.

[0017] In 208, the biological sample is analyzed based on the information of the captured digital image. Analysis of the digital image can include analysis of the pixel information contained in the image. In some examples, the pixel information can be analyzed on a pixel-by-pixel basis. In other examples, the pixel information can be analyzed on a block-by-block basis. In yet further examples, the pixel can be analyzed based on the entire area of the pixel, whereby the pixel information of the individual pixels within the area can be derived by combining the information of the individual pixels, selecting sample pixels, or using other statistical methods such as the statistical histogram operations described in more detail below. In the present disclosure, the operations described as applicable to "pixels" are similarly applicable to blocks or other groups of pixels, and the term "pixel" in this specification is intended to include such applications.

[0018] Analysis can include determining whether growth is detected in the culture medium. From the perspective of image analysis, growth can be detected in the image by identifying the imaged object (based on the difference between the object and its adjacent surroundings) and then identifying the changes in the object over time. As will be described in more detail herein, these differences And both changes take the form of "contrast". In addition to detecting growth, 108 Image analysis at 108 can further include quantification of the amount of detected growth, identification of multiple colonies, identification of sister colonies, etc. - can be further included.

[0019] At 210, it is determined whether a biological sample (particularly, identified sister colonies) exhibits a quantitatively large growth. If no growth is found, or only a small amount of growth is found, routine 200 can proceed to 220, where a final report is output. When proceeding from 210 to 220, the final report is likely to indicate no large amount of growth or report the growth of normal flora.

[0020] If it is determined that the biological sample exhibits a quantitatively large growth, at 212, one or more colonies can be selected from the image based on the previous analysis. Selecting colonies can be a fully automated process, in which each of the selected colonies is sampled and examined. Alternatively, selecting colonies can be a partially automated process, in which multiple candidate colonies are automatically identified and visually presented to the operator in the digital image, whereby the operator can input the selection of one or more candidates for sampling and further examination. Sampling of the selected or picked colonies can be automated by the system itself.

[0021] At 214, by plate culturing the sample in an organism suspension, etc., the sample The printed colonies are prepared for further inspection. At 216, the sample is examined using matrix-assisted laser desorption ionization (MALDI) imaging, and the type of specimen sampled from the original medium is identified. At 218, the sample is further or alternatively subjected to antibiotic susceptibility testing (AST), and possible treatments for the identified specimen are determined.

[0022] At 220, the test results are output in a final report. The report can include the results of MALDI and AST. As described above, the report can also show the quantification of specimen growth. For this reason, the automated system can generate a final report regarding the specimens found in the culture, starting from the inoculated medium, with little or no additional input.

[0023] In routines such as the exemplary routine of FIG. 2, the detected and identified colonies are, in many cases, referred to as colony-forming units (CFUs). A CFU is a microscopic object that starts as one or several bacteria. Over time, the bacteria grow to form colonies. The earlier in time from when the bacteria were placed in the plate, the fewer bacteria are detected, and as a result, the colonies are smaller and the contrast against the background is lower. In other words, the smaller the colony size, the smaller the signal it produces, and the smaller the signal in a constant background, the smaller the resulting contrast. This is reflected by the

Number

[0024] Contrast can play an important role in identifying objects such as CFUs or other artifacts within an image. An object can be detected within an image when its brightness, color, and / or texture is significantly different from its surroundings. Once an object is detected, the analysis can also include identifying the type of the detected object. Such identification can also rely on contrast measurements such as the smoothness of the edges of the identified object, or the uniformity (or lack thereof) of the color and / or brightness of the object. This contrast must be large enough to overcome the noise (background signal) of the image in order to be detected by the image sensor. Human contrast perception (governed by Weber's law) is limited. Under optimal conditions, the human eye can detect a 1% difference in light levels. The quality and reliability of image measurements (e.g., brightness, color, contrast) can be characterized by the signal-to-noise ratio (SNR) of the measurement. An SNR value of 100 (or 40 dB using 20 log) corresponds to human detection ability regardless of pixel luminance. Digital imaging techniques that utilize high SNR imaging information and known SNR per pixel information can enable the detection of colonies even when the colonies are not yet visible to the human eye.

[0025] 10

[0026] ​​​​​​​​​​​​​​​In the present disclosure, contrast can be collected in at least two ways, namely spatially and temporally. Spatial contrast or local contrast quantifies the difference in color or brightness between a given region (e.g., a pixel or a group of adjacent pixels) in a single image and its surroundings. Temporal contrast or time contrast quantifies the difference in color or brightness between a given region in one image and the same region in another image obtained at a different point in time. The formula governing temporal contrast is similar to that of spatial contrast. However, t is a point in time later than t. Both the spatial contrast and the temporal contrast of a given image can be used to identify objects. The identified objects can be further inspected to determine their significance (e.g., whether they are CFUs, normal flora, dust, etc.). Figures 3A, 3B, and 3C provide a visual demonstration of the effects that temporal contrast can have on an imaged sample. The image shown in Figure 3A was captured at multiple points in time (from left to right, from the top row to the bottom row) and shows the overall growth in the sample. Although the growth is prominent in Figure 3A, it is even more prominent in the corresponding contrast-time image of Figure 3B, and can be noticed at an earlier stage of the sequence. For clarity, Figure 3C is Figure 3B.

Number

[0027] However, t 1 is a point in time later than t 0 Both the spatial contrast and the temporal contrast of a given image can be used to identify objects. The identified objects can be further inspected to determine their significance (e.g., whether they are CFUs, normal flora, dust, etc.). Figures 3A, 3B, and 3C provide a visual demonstration of the effects that temporal contrast can have on an imaged sample. The image shown in Figure 3A was captured at multiple points in time (from left to right, from the top row to the bottom row) and shows the overall growth in the sample. Although the growth is prominent in Figure 3A, it is even more prominent in the corresponding contrast-time image of Figure 3B, and can be noticed at an earlier stage of the sequence. For clarity, Figure 3C is Figure 3B.

[0028] Figures 3A, 3B, and 3C provide a visual demonstration of the effects that temporal contrast can have on an imaged sample. The image shown in Figure 3A was captured at multiple points in time (from left to right, from the top row to the bottom row) and shows the overall growth in the sample. Although the growth is prominent in Figure 3A, it is even more prominent in the corresponding contrast-time image of Figure 3B, and can be noticed at an earlier stage of the sequence. For clarity, Figure 3C is Figure 3B. Figures 3A, 3B, and 3C provide a visual demonstration of the effects that temporal contrast can have on an imaged sample. The image shown in Figure 3A was captured at multiple points in time (from left to right, from the top row to the bottom row) and shows the overall growth in the sample. Although the growth is prominent in Figure 3A, it is even more prominent in the corresponding contrast-time image of Figure 3B, and can be noticed at an earlier stage of the sequence. For clarity, Figure 3C is Figure 3B. Figures 3A, 3B, and 3C provide a visual demonstration of the effects that temporal contrast can have on an imaged sample. The image shown in Figure 3A was captured at multiple points in time (from left to right, from the top row to the bottom row) and shows the overall growth in the sample. Although the growth is prominent in Figure 3A, it is even more prominent in the corresponding contrast-time image of Figure 3B, and can be noticed at an earlier stage of the sequence. For clarity, Figure 3C is Figure 3B. Figures 3A, 3B, and 3C provide a visual demonstration of the effects that temporal contrast can have on an imaged sample. The image shown in Figure 3A was captured at multiple points in time (from left to right, from the top row to the bottom row) and shows the overall growth in the sample. Although the growth is prominent in Figure 3A, it is even more prominent in the corresponding contrast-time image of Figure 3B, and can be noticed at an earlier stage of the sequence. For clarity, Figure 3C is Figure 3B. shows the zoomed section. As can be seen from Figure 3C, the longer a portion of the colony is imaged, the brighter spots are created in the contrast image. In this way, the centroid of each colony can be indicated by the bright center or peak of the colony. Thus, the image data obtained over time can reveal significant information regarding changes in colony morphology. The more the colony is imaged, the brighter spots are created in the contrast image. In this way, the centroid of each colony can be indicated by the bright center or peak of the colony. Thus, the image data obtained over time can reveal significant information regarding changes in colony morphology.

[0029] To maximize the spatial or temporal contrast of the object against the background, the system can capture images using different incident light in different backgrounds. For example, any of top illumination, bottom illumination, or side illumination methods can be used with a black background or a white background. To maximize the spatial or temporal contrast of the object against the background, the system can capture images using different incident light in different backgrounds. For example, any of top illumination, bottom illumination, or side illumination methods can be used with a black background or a white background.

[0030] Figures 3D and 3E give a visual demonstration of the effects that the illumination conditions can have on the imaged sample. The image in Figure 3D was captured using top illumination, while the image in Figure 3E was captured using bottom illumination, at approximately the same time point (e.g., close enough to the time when neither significant nor massive growth had occurred). As can be seen, each of the images of the samples in Figures 3D and 3E contains several colonies, but additional information regarding the colonies (in this case, hemolysis) can be seen in the image of Figure 3D by backlighting or bottom illumination, while it is difficult to grasp the same information in the image of Figure 3E. The image in Figure 3D was captured using top illumination, while the image in Figure 3E was captured using bottom illumination, at approximately the same time point (e.g., close enough to the time when neither significant nor massive growth had occurred). The image in Figure 3E was captured using bottom illumination, at approximately the same time point (e.g., close enough to the time when neither significant nor massive growth had occurred). As can be seen, each of the images of the samples in Figures 3D and 3E contains several colonies, but additional information regarding the colonies (in this case, hemolysis) can be seen in the image of Figure 3D by backlighting or bottom illumination, while it is difficult to grasp the same information in the image of Figure 3E. As can be seen, each of the images of the samples in Figures 3D and 3E contains several colonies, but additional information regarding the colonies (in this case, hemolysis) can be seen in the image of Figure 3D by backlighting or bottom illumination, while it is difficult to grasp the same information in the image of Figure 3E. As can be seen, each of the images of the samples in Figures 3D and 3E contains several colonies, but additional information regarding the colonies (in this case, hemolysis) can be seen in the image of Figure 3D by backlighting or bottom illumination, while it is difficult to grasp the same information in the image of Figure 3E. As can be seen, each of the images of the samples in Figures 3D and 3E contains several colonies, but additional information regarding the colonies (in this case, hemolysis) can be seen in the image of Figure 3D by backlighting or bottom illumination, while it is difficult to grasp the same information in the image of Figure 3E.

[0031] At a given time point, multiple images can be captured under multiple illumination conditions. The images are At a given time point, multiple images can be captured under multiple illumination conditions. The images are then, using different light sources, different spectra can be captured due to red, green, and blue filters be captured. In this way, the image acquisition conditions can be varied in terms of the light source position (e.g., top, side surface, bottom surface), background (e.g., black, white, any color, any luminance), and light spectrum (e.g., red channel, green channel, blue channel). For example, the first image can be captured using top illumination and a black background, and the second image can be captured using side illumination and a black background, and the third image can be captured using bottom illumination and no background (i.e., a white background). Further, using a specific algorithm, a set of varying image acquisition conditions can be generated to maximize the spatial contrast used. These or other algorithms can be used to maximize the temporal contrast by varying the image acquisition conditions according to a given sequence and / or over a certain period of time. Some such algorithms are described in Patent Document 1.

[0032] FIG. 4 is a flowchart showing an exemplary routine for analyzing an imaging plate based at least in part on contrast. The routine of FIG. 4 can be regarded as an exemplary subroutine of routine 200 of FIG. 2, and 206 and 208 of FIG. 2 are adapted to be performed using at least in part the routine of FIG. 4.

[0033] At 402, a first digital image at time point t 0 is captured. Time point t 0 can be the time point immediately after the start of the culture process, and the bacteria in the imaging plate are still It has not started to form a visible colony.

[0034] In 404, coordinates are assigned to one or more pixels in the first digital image. . In some cases, the coordinates can be polar coordinates having radial coordinates spreading from the center point of the imaging plate and angular coordinates around the center point. The coordinates can be used, in a later step, to align the first digital image with other digital images of the plate acquired from another angle and / or at another time point. In some cases, the imaging plate can have a specific landmark or reference mark. It is advantageous for such a reference mark to be detectable by a sensor (i.e., optically detectable). Using a reference mark to orient an object in a coordinate space is known to those skilled in the art. Examples of suitable and optically detectable reference marks include off-center marks such as dots or lines on the bottom surface of an optically transparent culture dish. Such marks can be detected by a sensor "facing" the bottom surface of the culture dish. The sensor can be positioned below the culture dish and can detect the reference mark if the support of the culture dish is optically transparent. The sensor can detect such a reference mark when the sensor is mounted above the culture dish if the culture medium placed in the culture dish is optically transparent. To avoid the difficulty of detecting a reference mark on the bottom surface of the culture plate from the top or bottom surface of the plate, the reference mark can be a label such as a barcode label attached to the side surface of the culture dish. The barcode label can be detected by a sensor. 1 digital image with other digital images of the plate acquired from another angle and / or at another time point. In some cases, the imaging plate can have a specific landmark or reference mark. It is advantageous for such a reference mark to be detectable by a sensor (i.e., optically detectable). Using a reference mark to orient an object in a coordinate space is known to those skilled in the art. Examples of suitable and optically detectable reference marks include off-center marks such as dots or lines on the bottom surface of an optically transparent culture dish. Such marks can be detected by a sensor "facing" the bottom surface of the culture dish. The sensor can be positioned below the culture dish and can detect the reference mark if the support of the culture dish is optically transparent. The sensor can detect such a reference mark when the sensor is mounted above the culture dish if the culture medium placed in the culture dish is optically transparent. To avoid the difficulty of detecting a reference mark on the bottom surface of the culture plate from the top or bottom surface of the plate, the reference mark can be a label such as a barcode label attached to the side surface of the culture dish. The barcode label can be detected by a sensor. plate can have a specific landmark or reference mark. It is advantageous for such a reference mark to be detectable by a sensor (i.e., optically detectable). Using a reference mark to orient an object in a coordinate space is known to those skilled in the art. Examples of suitable and optically detectable reference marks include off-center marks such as dots or lines on the bottom surface of an optically transparent culture dish. Such marks can be detected by a sensor "facing" the bottom surface of the culture dish. The sensor can be positioned below the culture dish and can detect the reference mark if the support of the culture dish is optically transparent. The sensor can detect such a reference mark when the sensor is mounted above the culture dish if the culture medium placed in the culture dish is optically transparent. To avoid the difficulty of detecting a reference mark on the bottom surface of the culture plate from the top or bottom surface of the plate, the reference mark can be a label such as a barcode label attached to the side surface of the culture dish. The barcode label can be detected by a sensor. mark to be detectable by a sensor (i.e., optically detectable). Using a reference mark to orient an object in a coordinate space is known to those skilled in the art. Examples of suitable and optically detectable reference marks include off-center marks such as dots or lines on the bottom surface of an optically transparent culture dish. Such marks can be detected by a sensor "facing" the bottom surface of the culture dish. The sensor can be positioned below the culture dish and can detect the reference mark if the support of the culture dish is optically transparent. The sensor can detect such a reference mark when the sensor is mounted above the culture dish if the culture medium placed in the culture dish is optically transparent. To avoid the difficulty of detecting a reference mark on the bottom surface of the culture plate from the top or bottom surface of the plate, the reference mark can be a label such as a barcode label attached to the side surface of the culture dish. The barcode label can be detected by a sensor. ​​​​​​​​​​​​It can be detected. In these examples, the side or center of the barcode label can be used as a reference mark. Pixel coordinates can be assigned in relation to the reference mark. In this way, the orientation of the plate in the coordinate space of the imaging device can be reproduced for each imaging event. The coordinates of the pixel(s) covering the landmark in the first image can be assigned to the pixel(s) covering the same landmark in other images. Therefore, since the pixels in the relatively early image and the same pixels in the later image share the same coordinates, both pixels can be compared. In this way, the change of pixels for each image can be easily observed. In 406, a second digital image at time point t is captured. Time point t is a time point after t when the bacteria in the imaging plate have the opportunity to form visible colonies. In 408, the second digital image is aligned with the first digital image based on the coordinates assigned so far. Aligning the images can further include, for example, normalizing and standardizing the images using the methods and systems described in Patent Document 1. In 410, the contrast information of the second digital image is obtained. The contrast information can be collected on a pixel-by-pixel basis. For example, the pixels in the second digital image are compared with the corresponding pixels (at the same coordinates) in the first digital image, and the time course...

[0035] 406, at time point t x a second digital image is captured. At time point t x is a time point after t when the bacteria in the imaging plate have the opportunity to form visible colonies. 0 later than t.

[0036] 408, the second digital image is aligned with the first digital image based on the coordinates assigned so far. Aligning the images can further include, for example, normalizing and standardizing the images using the methods and systems described in Patent Document 1. Aligning the images can further include, for example, normalizing and standardizing the images using the methods and systems described in Patent Document 1. .

[0037] 410, the contrast information of the second digital image is obtained. The contrast information can be collected on a pixel-by-pixel basis. For example, the pixels in the second digital image are compared with the corresponding pixels (at the same coordinates) in the first digital image, and the time course... els in the second digital image are compared with the corresponding pixels (at the same coordinates) in the first digital image, and the time course... ​​​​The presence of contrast can be determined. Further, adjacent pixels of the second digital image can be compared with each other or with other pixels known to the background pixels to determine the presence of spatial contrast. Changes in the color and / or brightness of the pixels indicate contrast, and the magnitude of such changes from one image to the next, or from one pixel (or region of pixels) to the next pixel (or region of pixels), can be measured, calculated, estimated, or otherwise determined. When both temporal contrast and spatial contrast are determined for a given image, the overall contrast of a given pixel of the image can be determined based on a combination (e.g., average, weighted average) of the spatial contrast and temporal contrast of that given pixel. Adjacent pixels of the second digital image can be compared with each other or with other pixels known to the background pixels to determine the presence of spatial contrast. Changes in the color and / or brightness of the pixels indicate contrast, and the magnitude of such changes from one image to the next, or from one pixel (or region of pixels) to the next pixel (or region of pixels), can be measured, calculated, estimated, or otherwise determined. Changes in the color and / or brightness of the pixels indicate contrast, and the magnitude of such changes from one image to the next, or from one pixel (or region of pixels) to the next pixel (or region of pixels), can be measured, calculated, estimated, or otherwise determined. Changes in the color and / or brightness of the pixels indicate contrast, and the magnitude of such changes from one image to the next, or from one pixel (or region of pixels) to the next pixel (or region of pixels), can be measured, calculated, estimated, or otherwise determined. When both temporal contrast and spatial contrast are determined for a given image, the overall contrast of a given pixel of the image can be determined based on a combination (e.g., average, weighted average) of the spatial contrast and temporal contrast of that given pixel. When both temporal contrast and spatial contrast are determined for a given image, the overall contrast of a given pixel of the image can be determined based on a combination (e.g., average, weighted average) of the spatial contrast and temporal contrast of that given pixel. When both temporal contrast and spatial contrast are determined for a given image, the overall contrast of a given pixel of the image can be determined based on a combination (e.g., average, weighted average) of the spatial contrast and temporal contrast of that given pixel. When both temporal contrast and spatial contrast are determined for a given image, the overall contrast of a given pixel of the image can be determined based on a combination (e.g., average, weighted average) of the spatial contrast and temporal contrast of that given pixel.

[0038] At 412, an object in the second digital image is identified based on the contrast information calculated at 410. Adjacent pixels of the second digital image having similar contrast information can be considered to belong to the same object. For example, if the difference in brightness between a plurality of adjacent pixels and their background, or the difference in brightness between the plurality of pixels and those pixels in the first digital image, is generally the same (e.g., within a predetermined criterion), those pixels can be considered to belong to the same object. As an example, a system can assign a "1" to any pixel having a large contrast (e.g., exceeding a reference amount), and then identify a group of adjacent pixels all assigned a "1" as an object. Objects can have pixels with the same label. At 412, an object in the second digital image is identified based on the contrast information calculated at 410. Adjacent pixels of the second digital image having similar contrast information can be considered to belong to the same object. Adjacent pixels of the second digital image having similar contrast information can be considered to belong to the same object. For example, if the difference in brightness between a plurality of adjacent pixels and their background, or the difference in brightness between the plurality of pixels and those pixels in the first digital image, is generally the same (e.g., within a predetermined criterion), those pixels can be considered to belong to the same object. For example, if the difference in brightness between a plurality of adjacent pixels and their background, or the difference in brightness between the plurality of pixels and those pixels in the first digital image, is generally the same (e.g., within a predetermined criterion), those pixels can be considered to belong to the same object. For example, if the difference in brightness between a plurality of adjacent pixels and their background, or the difference in brightness between the plurality of pixels and those pixels in the first digital image, is generally the same (e.g., within a predetermined criterion), those pixels can be considered to belong to the same object. As an example, a system can assign a "1" to any pixel having a large contrast (e.g., exceeding a reference amount), and then identify a group of adjacent pixels all assigned a "1" as an object. As an example, a system can assign a "1" to any pixel having a large contrast (e.g., exceeding a reference amount), and then identify a group of adjacent pixels all assigned a "1" as an object. As an example, a system can assign a "1" to any pixel having a large contrast (e.g., exceeding a reference amount), and then identify a group of adjacent pixels all assigned a "1" as an object. Specific labels or masks can be given so that the loops share certain characteristics. This label can help distinguish the object from other objects and / or the background during the process after subroutine 400. Identifying an object in a digital image can include dividing or partitioning the digital image into multiple regions (e.g., foreground and background). The goal of the partitioning is to change the image into a representation of multiple elements so that the elements are easier to analyze. Image partitioning is used to identify the position of the object within the image.

[0039] At 414, the nature of a given object (identified at 412) can be characterized. Characterizing the nature of an object can include deriving descriptive statistics of the object (e.g., area, reflectivity, size, optical density, color, plate location, etc.). Descriptive statistics can ultimately quantitatively describe certain characteristics of a set of information collected regarding the object (e.g., from a SHQI image, from a contrast image). Such information can be evaluated depending on the species, concentration, mixture, time, and medium. On the other hand, in at least some cases, characterizing an object can start from a set of qualitative information regarding the characteristics of the object. The qualitative information can then be expressed quantitatively. Table 1 below gives an exemplary list of characteristics that can be qualitatively evaluated and then changed to a quantitative representation.

[0040]

Table 1

[0041] ​​​​​​Some features of the object, such as shape or the time until visually observable, 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 having a common y - coordinate, for each column of pixels having a common x - coordinate, for each ray of pixels having a common angular coordinate, for each circle of pixels having a common radial coordinate), and then combined into a single measurement value using, for example, a histogram. For example, color can be measured for each pixel, and growth rate or size can be measured for each row, column, ray, or circle of pixels, and so on.

[0042] In 416, it is determined whether the object is a candidate for a colony based on the characterized properties. The determination of colony candidates can include inputting quantitative features (e.g., the scores shown in Table 1 above), or a part thereof, into a classifier. The classifier can 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. Supervised learning may be preferred when the object is to be distinguished from a limited set (e.g., two or three) of possible organisms (in this case, the algorithm can be trained on a relatively limited set of training data). In contrast, unsupervised learning may be preferred when the object is to be distinguished from the entire database of possible organisms. In this case, it is difficult to provide comprehensive or even sufficient training data. In the case of a confusion matrix or a matching matrix, the discrimination can be numerically measured in terms of range. It can be done. For example, for a given pair of objects, "0" means that the two objects are distinguishable from each other while "1" can mean that it is difficult to differentiate the objects from each other from each other.

[0043] Colony candidates can be stored in the memory of the automation system for further use (e.g., inspection, split routines described below, etc.). for the purpose of

[0044] [Use of multiple media] In the above example, the evaluation of the culture for a single medium was described. On the other hand, these examples are equally applicable when the culture is evaluated in multiple media.

[0045] The characteristics of the bacteria (e.g., color, growth rate, etc.) may vary depending on the type of culture medium ("medium") used, so different confusion matrices can be applied for each medium during classification (e.g., 416 of subroutine 400). For this reason, while a classifier for one medium outputs "0" for two objects, it is perfectly reasonable for a classifier for multiple media to output "1" for the same two objects. Next, the collective results of the classifiers can be evaluated together (manually or based on further machine-driven relationships) to arrive at an overall or final differentiation or classification for the objects. The evaluation of multiple media can be carried out using a single container. The single container can be configured to hold multiple media (e.g., double plates, triple plates, quadruple plates, etc.) so that multiple media can be imaged simultaneously. Alternatively, several

[0046] media can be evaluated using a single container. The single container can be configured to hold multiple media (e.g., double plates, triple plates, quadruple plates, etc.) so that multiple media can be imaged simultaneously. Alternatively, several media (e.g., double plates, triple plates, quadruple plates, etc.) can be held so that multiple media can be imaged simultaneously. Or, several It is possible to evaluate a plurality of culture media by streaking a culture sample in the container. Each container holds one or more culture media. Next, each of the plurality of containers can be subjected to the imaging routine described above. In order to identify the growth found in various culture media with more sufficient information, information derived from each of the culture media (e.g., characterized features) can be input into the classifier all together. can. Each container holds one or more culture media. Next, each of the plurality of containers can be subjected to the imaging routine described above. For more sufficient identification of the growth found in various culture media, information derived from each of the culture media (e.g., characterized features) can be input into the classifier all together.

[0047] [Contrast information] FIG. 5 is a flowchart showing an exemplary subroutine 500 for obtaining spatial contrast as part of 410 of FIG. 4. The subroutine 500 receives, as input, a combination of one or more background conditions and illumination conditions 551 and a filter 554. At 502, a digital image is obtained from the input combination 551 under the specified illumination and background conditions. Next, at 5 04, the image is replicated. At 506, one of the replicated images is filtered using the filter 554. In the example of FIG. 5, a low-pass kernel is used as the filter, but those skilled in the art will recognize other filters that can be used. At 5 08, the difference between the unfiltered image and the filtered image is calculated, and the ratio of the sum of the unfiltered image and the filtered image is calculated. At 510, a spatial contrast image is obtained based on the calculated ratio of 508. This routine 500 can be repeated for each of the background conditions and illumination conditions 551. As a result of each iteration of the routine 500, another spatial contrast image is obtained, and this is used to update the spatial contrast image stored so far at 510. image is obtained, and this is used to update the spatial contrast image stored so far at 510. image is obtained, and this is used to update the spatial contrast image stored so far at 510. image is obtained, and this is used to update the spatial contrast image stored so far at 510. image is obtained, and this is used to update the spatial contrast image stored so far at 510. image is obtained, and this is used to update the spatial contrast image stored so far at 510. image is obtained, and this is used to update the spatial contrast image stored so far at 510. ​The image can be updated iteratively. Thus, an overall contrast image (including the contrast from each of the illumination conditions can be constructed iteratively. In one embodiment , in each iteration, a cleared contrast image in which the contrast setting is still set to zero (compared to the iteratively constructed contrast image) can be provided as an input for each illumination setting . At 512, when it is determined that the last image has been processed , routine 500 ends.

[0048] FIG. 6 is a flowchart showing an exemplary subroutine 600 for obtaining a temporal contrast as part of 410 of FIG. 4 as well. Subroutine 600 receives, as inputs, a set 651 of one or more background and illumination conditions and a filter 655. At 602, a first digital image and a second digital image obtained under a particular illumination condition and background condition are each obtained. At 604, the t image is filtered. In the example of FIG. 6, a low-pass kernel is used as the filter, but those skilled in the art will recognize other filters that can be used . At 606, the ratio of the unfiltered t image minus the filtered t image to the unfiltered t 0 image plus the filtered t image is calculated. At 608 , a temporal contrast image is obtained based on the ratio calculated at 606. This routine 6 x 00 can be repeated under different illumination conditions and / or different background conditions. Routine 0 x 0 , 608, a temporal contrast image is obtained based on the ratio calculated at 606. This routine 6 00 can be repeated under different illumination conditions and / or different background conditions. Routine As a result of each iteration of 600, another temporal contrast image is obtained and used to iteratively update the temporal contrast images previously stored at 608. Similar to the construction of the spatial contrast image, the temporal contrast image can be iteratively constructed, and the cleared contrast image is provided as input for each lighting condition. Routine 600 ends when it is determined at 610 that the last image has been

[0049] processed. To make an overall or global determination regarding contrast, the results of spatial contrast and temporal contrast can be further combined. The combination of spatial contrast and temporal contrast is referred to herein as "mixed contrast" (MC). In one embodiment, the mixed contrast can be derived from the spatial contrast (SC) image at 0 time point t, the spatial contrast image at time point t x and the temporal contrast (TC) image derived from the comparison of the t image and the t 0 image and the t x image, according to the following equation.

Number

[0050] [Filtering] To improve image analysis, additional processes can be included in subroutine 400 of FIG. 4. For example, the first digital image can be analyzed for objects appearing within the 0 image at time point t. For the object at t 0 ​where it is known that the bacteria have not yet started to grow significantly so that any object found at time point t 0 is simply assumed to be dust, bubbles, artifacts, condensation, etc. that do not constitute a candidate for a colony and can be assumed to be.

[0051] One filtering process can be used on the captured image to subtract dust and other artifacts on the imaging plate or lens . When considering a transparent medium (e.g., MacConkey agar medium, CLED agar medium, CHROM agar medium, etc.), it is expected that a certain level of dust will be present in the captured image. The influence of dust in a given image can be shown to be at least partially based on the specific lighting conditions and background conditions under which the image is obtained . For example, when using a white medium, reflective artifacts and dust are most observable when the medium is illuminated from above with a black background on the bottom . As another further example, when using a colored or dark medium, artifacts and dust are most observable when the medium is illuminated from above with a white background on the bottom. As a further example, in almost any medium, light-absorbing artifacts and dust are observable when the medium is illuminated from the bottom, regardless of the background . In any case, the management of dust and artifacts is a complex image processing problem that can greatly affect the detection of microbial growth . For example, when using a white medium, reflective artifacts and dust are most observable when the medium is illuminated from above with a black background on the bottom . . As another example, when using a colored or dark medium, artifacts and dust are most observable when the medium is illuminated from above with a white background on the bottom. As a further example, in almost any medium, light-absorbing artifacts and dust are observable when the medium is illuminated from the bottom, regardless of the background . In any case, the management of dust and artifacts is a complex image processing problem that can greatly affect the detection of microbial growth . For example, in almost any medium, light-absorbing artifacts and dust are observable when the medium is illuminated from the bottom, regardless of the background . In any case, the management of dust and artifacts is a complex image processing problem that can greatly affect the detection of microbial growth .

[0052] Dust and artifacts can be divided into two types, namely, (A) those whose position can change and , (B) those whose position cannot change ​​can accumulate over time, which means that the numbers of both types A and B may vary over time. Nevertheless, the observation results indicate that type A is more likely to cause a change in the amount over time than type B. Naturally, type A is also more likely to cause changes, for example, due to the movement of the plate in and out of the imaging chamber. Even so, the observation results indicate that type A is more likely to cause a change in the amount over time than type B. Naturally, type A is also more likely to cause changes, for example, due to the movement of the plate in and out of the imaging chamber. Even so, the observation results indicate that type A is more likely to cause a change in the amount over time than type B. Naturally, type A is also more likely to cause changes, for example, due to the movement of the plate in and out of the imaging chamber.

[0053] Typically, type B is caused by artifacts related to the plate itself, such as ink dots (brand, lot number, and information printed on the bottom of the plate), defects related to plastic injection molding points, or frosted areas. Type B can also be caused by clogging on top of the medium, being trapped within the medium, or electrostatically clogged dust or air bubbles on the underside of the plate. Typically, type B is caused by artifacts related to the plate itself, such as ink dots (brand, lot number, and information printed on the bottom of the plate), defects related to plastic injection molding points, or frosted areas. Type B can also be caused by clogging on top of the medium, being trapped within the medium, or electrostatically clogged dust or air bubbles on the underside of the plate. Typically, type B is caused by artifacts related to the plate itself, such as ink dots (brand, lot number, and information printed on the bottom of the plate), defects related to plastic injection molding points, or frosted areas. Type B can also be caused by clogging on top of the medium, being trapped within the medium, or electrostatically clogged dust or air bubbles on the underside of the plate.

[0054] From the perspective of imaging, even type A dust and artifacts themselves hardly change their positions. On the other hand, due to the plasticity of the plate and the medium that act as filters and lenses, the observed characteristics and positions of type A artifacts may change slightly depending on the color of the medium, the medium level, and the plastic. Type B dust and artifacts also do not change their positions. On the other hand, as long as type B dust and artifacts are connected to the medium and the medium undergoes slight movement and shift over time (mostly due to slight drying over time in the incubator), type B dust and artifacts may move relative to the medium. Therefore, the positions of type B dust and artifacts are also likely to undergo at least some minute changes. From the perspective of imaging, even type A dust and artifacts themselves hardly change their positions. On the other hand, due to the plasticity of the plate and the medium that act as filters and lenses, the observed characteristics and positions of type A artifacts may change slightly depending on the color of the medium, the medium level, and the plastic. Type B dust and artifacts also do not change their positions. On the other hand, as long as type B dust and artifacts are connected to the medium and the medium undergoes slight movement and shift over time (mostly due to slight drying over time in the incubator), type B dust and artifacts may move relative to the medium. From the perspective of imaging, even type A dust and artifacts themselves hardly change their positions. On the other hand, as long as type B dust and artifacts are connected to the medium and the medium undergoes slight movement and shift over time (mostly due to slight drying over time in the incubator), type B dust and artifacts may move relative to the medium. Therefore, the positions of type B dust and artifacts are also likely to undergo at least some minute changes. From the perspective of imaging, even type A dust and artifacts themselves hardly change their positions.

[0055] From the perspective of contrast, a small piece of dust of type A can be present in the last image at position "p 0 " at time t 0 spatial contrast and in the spatial contrast image at position "p x " at time t x can exist. Assuming that p and p 0 and p x are different locations, the dust or artifact fact also exists in the temporal contrast images at both locations (for example, p x shows positive contrast at the location, and p 0 shows negative con trast at the location). By comparison, a small piece of dust of type B exists at the common location of the spatial contrast images at times t 0 and t x but not in the temporal contrast image . As described above, the spatial contrast image and the temporal contrast image can be combined to derive a mixed contrast result. The effects of dust and artifacts of both type A and type B can be further removed from the mixed contrast result

[0056] . In one embodiment, if an object (e.g., a CFU candidate) is identified in the mixed contrast result, this object can be compared with the dust and artifacts detected by the neighborhood N(x,y) of the object in the spatial contrast result at time t . Thus, if a similar object is found in the spatial contrast result at time t , the object identified in the mixed contrast result is of type A or type B . For this reason, if a similar object is found in the spatial contrast result at time t , the object identified in the mixed contrast result can be compared with the dust and artifacts detected by the neighborhood N(x,y) of the object in the spatial contrast result at time t 0 . If a similar object is found in the spatial contrast result at time t in the spatial contrast result at time t , the object identified in the mixed contrast result is of type A or type B 0 . If a similar object is found in the spatial contrast result at time t , the object identified in the mixed contrast result is of type A or type B It is flagged as a false positive. Even if the object is not initially flagged as a false positive of type A or type B, if it is found that the size of the object has not changed significantly over time, it can still later be determined that the object is a false positive of type B. The false positives are remembered and can be applied to subsequent images through a filtering mask (e.g., a binary mask) etc. which will be further explained below. Another filtering process can be used to subtract condensation formed on the plate (e.g., while transferring from the refrigerator to the incubator at the start of the culture session). In one exemplary condensation filter, the plate is illuminated using bottom illumination, and thus the light passing through the location of the condensation is less than that passing through locations without condensation. Next, the optical density of the image can be evaluated, and areas of low optical density can be subtracted from the image. Furthermore or alternatively, an image mask can be constructed to exclude an object from consideration in any analysis of the t image and / or subsequent digital images. Figure 7 is a flowchart showing an exemplary routine 700 for generating an image mask using the spatial contrast of the t

[0057] In the example of routine 700, the only input provided is the S HQI image 751 obtained at time t . At 702, the spatial contrast of the t image is determined. At 704, using the spatial contrast information, the mean and standard deviation (e.g., of brightness) etc. of the t image are calculated.

[0058] image are calculated. 0 image and / or subsequent digital images. In the example of routine 700, the only input provided is the S 0 HQI image 751 obtained at time t . At 702, the spatial contrast of the t image is determined. 0 image is determined. At 704, using the spatial contrast information, the mean and standard deviation (e.g., of brightness) etc. of the t 0 image are calculated. At 704, using the spatial contrast information, the mean and standard deviation (e.g., of brightness) etc. of the t image are calculated. 0Collect statistical information about the pixels within the object region of the image. At 706 Adjust the contrast criterion so that an appropriate number of pixels exceed the same criterion. For example, if more pixels than a given percentage are not considered as the background of the image, the criterion can be increased. At 708, the same criterion is further adjusted based on the statistical information about these pixels below the same criterion. Finally, at 710, a binary mask is generated. The binary mask distinguishes various artifacts that are not valid pixels from other pixels that are considered valid. Next, the binary mask is used at subsequent time points where potential colonies exist to detect objects occupying invalid pixels and prevent these objects from becoming candidate colonies.

[0059] The above filtering process avoids erroneously including dust, dew, or other artifacts as objects, and at 414, the characterization of properties needs to be performed only for valid pixels, so the subroutine 400 can be improved by accelerating the characterization.

[0060] [Definition of Object and Label] Another process that can be added to the subroutine 400 in FIG. 4 is to assign labels to the objects identified at 412. A specific label can be given to an object, so that pixels having the same label share a certain characteristic. The label can be useful for differentiating that object from other objects and / or the background in a process after the subroutine 400. FIG. 8 shows an image acquired at time t x obtained at That is, an exemplary routine 800 for labeling the pixels of the "t" x image") is shown as a flowchart. In the example of FIG. 8, a binary mask 851 (e.g., the output of routine 700), and an uninitialized candidate mask 852 for the t image, and a temporal contrast x image 853 (e.g., the output of subroutine 600) are received as inputs. At 802, the candidate mask 852 is initialized. The initialization can include specifying the object region on the imaging plate and identifying the " valid pixels" in the image acquired at time t x using the binary mask 851. A valid pixel is a pixel of the image that has not been excluded from consideration as a candidate and is being considered for labeling. At 804, the temporal contrast image 853 is used to collect statistical information regarding the valid pixels within the object region of the t image, such as the mean and standard deviation (e.g., of brightness). Next, at 806, the statistical information of each of the temporal contrast image 853 and the binary mask x 851 (preferably generated under similar illumination and background conditions) is combined to form a reference contrast image. Using the reference defined by the reference contrast image, at 808, the "connex component" of the t image is labeled. A connex component is a label that substantially indicates the connection (or grouping) between adjacent pixels, and thus indicates that multiple pixels are part of the same object. The connex component at t x is... omponent) is labeled. A connex component is a label that substantially indicates the connection (or grouping) between adjacent pixels, and thus indicates that multiple pixels are part of the same object. And this indicates that multiple pixels are part of the same object.

[0061] The connex component at tx Once the image is defined, each connection component can be individually analyzed and its status verified as a single object. In the example of Figure 8, at 812, statistical calculations of the pixels associated with the label are performed. This calculation can utilize a histogram to determine the average and / or standard deviation of the brightness or color of the pixels. At 814, it is determined whether the pixel matches the reference area. If it does not match the reference area, the process proceeds to 830, where the label is updated. Updating the label can involve holding the analyzed component as one label or dividing the component into two labels. If the reference area does not match, the component is held as a single label. If the reference area matches, at 816, the histogram is smoothed, and at 818, the peaks of the pixels labeled separately are identified. The peak can further be defined by having the smallest area. This is because peaks smaller than the smallest area can be ignored. At 820, the number of identified peaks can be counted. If only one peak exists, the process proceeds to 830, where the label is updated, thereby holding the component as one object. If two or more peaks exist, at 824, using the reference contrast image, it is further evaluated whether the contrast between the peaks is large. Next, the process proceeds to 830, where the label is updated based on the multiple It is.

[0062] [Split] Another process that can be included as part of subroutine 400 is, at time t x at a splitting process for separating a colony in a confluent state (dense state) into separate objects is. At time t x when the colonies have grown to where they overlap or contact each other it may be necessary to draw boundaries through the confluent region in order to evaluate the separate colonies within the region.

[0063] In some examples, when two adjacent colonies have different characteristics (e.g., different colors, different textures), the splitting can simply involve feature analysis of the confluent region and can be done. On the other hand, spatial and temporal contrast alone is not always sufficient to identify the boundaries between colonies. FIG. 9 is a flowchart showing an exemplary routine 900 for separating such colonies into separate objects (e.g., having separate labels), or in other words, for splitting the colonies. The exemplary routine 900 of FIG. 9 uses as inputs a first digital image 951 obtained at time t and a second digital image obtained at time t and an image binary mask 953 (e.g., a mask generated by routine 700). At 902, a temporal contrast image is generated based on the t 0 image 951 and the t x image 952. At 904, the temporal contrast image is split using the binary mask 953. At 906, the segmentation of the image 0 is using the mask) as inputs. At 902, a temporal contrast image is generated based on the t 0 image 951 and the t x image 952. At 904, the temporal contrast image is split using the binary mask 953. At 906, the segmentation of the image Labels are attached to the mentos. At 908, the peak or maximum value of each label is identified . The maximum value of a given segment is usually the center point or center of mass of the segment. 91 0, for each label, using the maximum value of (e.g., the neighboring labels of the label being analyzed) to make a further determination as to whether a given label is unique with respect to its neighbors or should be combined with one or more neighboring labels . When the label is trimmed down to a unique component , the characterization of the properties for each label (e.g., steps 414 and 416 of routine 400) can be performed at 912, and a global list of candidate colonies can be generated at 914 .

[0064] Using various coefficients such as the inclusion factor, it can be determined whether the local maximum value of a given label belongs to one colony or to separate colonies . The inclusion factor is a coefficient indicating whether a plurality of neighboring pixels are associated with an object that is adjacent . Such coefficients are used in a splitting strategy to determine whether the two local maximum values at a given label should be separated into two separate objects or integrated into a single object .

[0065] Figure 10 is a flowchart showing such an exemplary way of splitting. Routine 1000 can be used as a subroutine for step 910 in Figure 9 . As shown in Figure 10 , two local maximum values 1051 and 1052 are identified. At 1002, the surrounding region is identified for each maximum value . In the example of Figure 10, region "A" encloses the maximum value 1051, and region "B" encloses the maximum value 1052. For the following exemplary formula, that of region A ​​​ The size of each region is equal to or greater than the size of region B. In some examples, each region is The horizontal distance (xA, xB) along the horizontal axis and the vertical distance (yA, yB) along the vertical axis of the region FIG. 11 is a simplified version of the routine in FIG. 10. To clarify, an exemplary diagram of regions A and B and their respective maximum values ​​is provided.

[0066] In 1004, for each local maximum value 1051 and 1052, the edge of the object is calculated from the maximum value. In some examples, the distance to the edge is determined. The distance map is the average or median of the distances of the assigned regions. The distance map of region A is hereafter referred to as rA, and the distance map of region B is hereafter referred to as rB.

[0067] In 1006, the distance “d” between the two local maxima and the distance “d” between the two local maxima A coverage factor is calculated based on the distance. In one embodiment, the coverage factor is calculated as follows: It is calculated using the formula:

number

[0068] In 1008, the coverage factor is less than a predetermined range (for example, 0.5 to 1) or It is determined whether the coverage factor is greater than or within a specified range. If the difference is less than the predetermined range, then both maxima are determined to be associated with the same object. If both maxima are greater than 1, then it is determined that they are associated with different objects.

[0069] If the coverage factor falls within the above range, both maxima belong to the same object or to different objects. Whether it belongs to [the relevant category] is not immediately obvious, and further processing is required. Subsequently, routine 1 000 advances to 1010, and at 1010, the convexity of the regions around each of the two maximum values is calculated using the coordinates of the third region "C" at the position between the two maximum values. In several examples, this region can be the weighted center of the two regions, and the center point of region C is closer to the smaller region B than to the larger region A. For region C, the horizontal distance xC, the vertical distance yC, and the distance map H can also be calculated For example, convexity can be calculated using the above values and d(A,C) according to the following formula. d(A,C) is the distance between the center point of region C and the maximum value A. For region C, the horizontal distance xC, the vertical distance yC, and the distance map H can also be calculated. For example, convexity can be calculated using the above values and d(A,C) according to the following formula. d(A,C) is the distance between the center point of region C and the maximum value A. d(A,C) is the distance between the center point of region C and the maximum value A.

Equation

[0070] At 1012, it is determined whether the convexity value is greater than a given criterion (more convex). For example, ΔΗ can be compared with the criterion 0. If the convex value is greater than the criterion, both maximum values are determined to be associated with separate objects. Otherwise, at 1014, one or more parameters of region C are updated so that the size of region C increases For example, distOffset can be updated based on ΔH. For example, ΔH has an upper limit of a value from 0 to 1 (if ΔH is greater than 1, ΔH is rounded to 1 ), and then added to distOffset. For example, distOffset can be updated based on ΔH. For example, ΔH has an upper limit of a value from 0 to 1 (if ΔH is greater than 1, ΔH is rounded to 1 ), and then added to distOffset. ) and then added to distOffset.

[0071] At 1016, it is determined whether the size of region C is above the criterion. If it is above the criterion, both maximum values are determined to be associated with the same object. In other words, the region ​Since the difference between region A and region B is not distinct, region C begins to cast a shadow on regions A and B When increasing to such an extent that the maximum values 1051 and 1052 belong to the same object, this is preferably indicated. In the above example, this can be indicated by the fact that distOffset is greater than or equal to the distance d between the maximum values. Otherwise, the operation returns to 1010, and the convexity of regions A and B is recalculated based on the updated parameter(s) of region C.

[0072] When an association is required for all maximum values, the obtained association can be stored in, for example, a matrix (also called an association matrix). Using the stored information, the entire list of maximum values can be converted into a final list of candidate objects. For example, in the case of an association matrix, a master list can be created from the entire list of maximum values, and then each maximum value is reviewed iteratively. If the associated maximum value still remains on the list, this maximum value can be removed from the master list.

[0073] In the examples of FIGS. 9 and 10, the time point t at which the second image is acquired x (for this reason, the earliest time point at which routine 900 can be executed) can be a few hours after the start of the culture process. Such a time point is usually considered too early to identify fully formed colonies, but may be sufficient to create a segmented image. The segmented image may be applied to future images obtained at subsequent time points as needed. For example, to predict the expected growth of colonies, the boundaries between colonies can be drawn. Next, when the colonies become confluent, the boundaries can be used to handle the confluent colonies. Ronnie can be separated.

[0074] [Analysis using two or more images after time point t 0 is performed] The processes and routines described above require only one image obtained after time point t 0 (for example the first digital image at time point t 0 and the second digital image at time point t x ), but another process requires at least a second image obtained after time point t . For example, if it is discovered that the image at time point t 0 contains colonies in a confluent state, another image obtained at time point t (where 0 < n < x) can be used to identify and separate individual colonies. x For example, if t = 0 hours after the start of cultivation (at this time, no growth has occurred), and n t = 24 hours after the start of cultivation (at this time, a lot of growth has occurred, so the colonies are in a confluent state), then for t

[0075] = 12 hours (at this time 0 the colonies have started to grow but are not yet in a confluent state), the presence of individual colonies becomes apparent from the image at this time point. Next, based on the image at time point t the growth of the colonies can be predicted, and the boundaries between the confluent colonies at time point t x can be estimated. In this regard, the image at time point t n can help distinguish fast-growing colonies from slow-growing colonies. A person skilled in the art would understand that time point t and time point From the image at, the presence of individual colonies becomes apparent. Next, based on the image at time point t n the growth of the colonies can be predicted, and the boundaries between the confluent colonies at time point t can be estimated. x In this regard, the image at time point t can help distinguish fast-growing colonies from slow-growing colonies. A person skilled in the art would understand that time point t n The image at can help distinguish fast-growing colonies from slow-growing colonies. A person skilled in the art would understand that time point t and time point t 0 and time point t x As the number of images obtained between t increases, it should be recognized that the growth rate of the colony can be predicted more accurately. It should be recognized that the growth rate of the colony can be predicted more accurately as the number of images obtained between t increases.

[0076] In one application of the above concept, at time point t n the image obtained at t (or, more comprehensively and including the images obtained between time point t 0 and t x can be used to identify the seeds of the colony, which is an object that appears to grow over time, and these seeds can be associated with the corresponding masks and labels. Each seed is assigned a unique label, and the label is stored together with a plurality of features (e.g., position, morphology, and histogram, red channel, green channel, blue channel, luminance, chrominance hue, or composite image) generated from the SHQI image, and characteristics (e.g., separated state / non-separated state, other information for predicting time-series reproduction ). Some of the stored features (e.g., histogram ) can be calculated at the plate level rather than being attributed to a specific seed for extracting global indicators of the plate. Next, using the features stored by the seeds, the colonies at time point t can be extracted, and these features are also provided as inputs to the classifier for training and / or inspection. x

[0077] During the workflow loop, dust, artifacts, or other foreign objects that appear on the plate or within the imaging lens can also be detected from the tracking of the growth rate using a plurality of images obtained after t t 0 For example, a small piece of dust appears after t but before t but after t 0 but before tn Before the imaging lens If it falls on top, the spot generated by this small piece is at time t 0 was not visible, so it may initially be interpreted as a growing colony. On the other hand, if subsequent imaging reveals that there is no change in size for the spot, it can be determined that the spot is not a colony because it is not growing. Apart from tracking growth rate and division, other aspects of the colony can be tracked with the help of additional images between t

[0078] and t 0 and t x and. Minor morphological changes that progress slowly over time within the colony can be identified more quickly by capturing more images. In some cases, growth can be measured along the z-axis in addition to or instead of the normal x and y axes. For example, Streptococcus pneumoniae is known to slowly form a sunken center when growing in blood agar, but the sunken center is usually not visible until the second day of analysis. By examining the time course of bacterial growth it is possible to detect the initial sunken center and identify the bacteria much earlier than if one had to wait for the center to fully sink. In other cases, the colony may be known to change color over time. Thus, imaging of a colony that has a first color (e.g., red) at a time point later than t and then has a second color (e.g., green) at a subsequent time point can be used to identify the bacteria growing within the colony. Color changes can be in a color space (e.g., RGB, CMYK and can be identified more quickly by capturing more images. In some cases, growth can be measured along the z-axis in addition to or instead of the normal x and y axes. For example, Streptococcus pneumoniae is known to slowly form a sunken center when growing in blood agar, but the sunken center is usually not visible until the second day of analysis. By examining the time course of bacterial growth it is possible to detect the initial sunken center and identify the bacteria much earlier than if one had to wait for the center to fully sink. In other cases, the colony may be known to change color over time. Thus, imaging of a colony that has a first color (e.g., red) at a time point later than t

[0079] and then has a second color (e.g., green) at a subsequent time point can be used to identify the bacteria growing within the colony. Color changes can be in a color space (e.g., RGB, CMYK and can be identified more quickly by capturing more images. In some cases, growth can be measured along the z-axis in addition to or instead of the normal x and y axes. For example, Streptococcus pneumoniae is known to slowly form a sunken center when growing in blood agar, but the sunken center is usually not visible until the second day of analysis. By examining the time course of bacterial growth 0 and then has a second color (e.g., green) at a subsequent time point can be used to identify the bacteria growing within the colony. Color changes can be in a color space (e.g., RGB, CMYK Using imaging of colonies that have a first color (e.g., red) at a time point later than t and then have a second color (e.g., green) at a subsequent time point, the bacteria growing within the colony can be identified. Color changes can be in a color space (e.g., RGB, CMYK etc.) It can be measured as a vector or path (etc.). Changes to other color characteristics of the colony can be measured in the same way.

[0080] [Properties of the object] As previously described in connection with FIG. 4, the nature of the object on the imaging plate can be characterized as part of the image analysis performed on the imaging plate. The characterized properties can include both static features (relating to a single image) and dynamic images (relating to multiple images).

[0081] Static features are intended to reflect the attributes of the object and / or the surrounding background at a given point in time. Static features include the following. (i) Centroid: This is a static feature that gives the center of mass of the object imaged in a coordinate space (e.g., x - y, polar). The centroid of the object gives invariance to a set of features under given illumination and background conditions, such as the polar coordinates of the object. The centroid can first be obtained by finding the weighted center of mass for all colonies in the image (where M is the binary mask for all detected colonies). The weighted center of mass can be determined based on the assumption that each pixel in the image has an equal value. Next, the centroid for a given colony can be represented in x - y coordinates by the following formula (where E = {p|p ∈ M} (E is the binary mask of the current colony), the range of the x coordinate is [0, image width], the range of the y coordinate is [0, image height], and each pixel is 1 unit). [Equation] (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. Usually, polar coordinates can be measured along the radial axis (d) and the angular axis (Θ), and the coordinates of the plate center are [0,0]. igv (x,y) of The coordinates d and Θ are given by the following formula (where d is in millimeters and Θ is in degrees): (However, k is the pixel density in pixels corresponding to millimeters, and the "barcode" is a landmark feature of the imaging plate to ensure the alignment of the plate with past and / or future images.)

Number

Number

Number

Number

Number

Equation

Equation

[0082] Dynamic features aim to reflect changes in the attributes of the object and / or the surrounding background over time. Through time-series processing, it becomes possible to relate static features over time. These Dynamic features aim to reflect changes in the attributes of the object and / or the surrounding background over time. Through time-series processing, it becomes possible to relate static features over time. These The discrete first and second derivatives of the features give the instantaneous "speed" and "acceleration" (or constancy or deceleration) of the changes in such properties characterized over time in . Examples of dynamic features include the following (i) Time series processing to track the above static features over time. Each feature measured at a given culture time point is referenced according to its relative culture time so that the feature can be related to the feature measured at a later culture time point . Using a time series of images, as described above, objects such as CFUs that appear and grow over time can be detected . Based on the ongoing analysis of the images captured so far for the object, the time points for imaging can be preset or determined by an automated process. At each time point , the image can be a given acquisition configuration for the entire series with a single acquisition configuration or for the entire series of images captured by multiple acquisition configurations . (ii) Discrete first and second derivatives of such features to give the instantaneous speed and acceleration (or constancy or deceleration) of the changes to the above features over time (e.g., tracking the growth rate as described above). (a) Speed: The first derivative of the feature over time. The speed (V) of feature x can be characterized in terms of (x units) / time using a period Δt that is a period expressed in time units based on the following equation . . (b) Acceleration: The second derivative of the feature over time, which is also the first derivative of the speed. The acceleration (A ) can be characterized based on the following equation .

Number

Number

[0083] The above image features are measured from the object or the context of the object and are intended to capture the specificity of organisms growing in various media and culture conditions. The listed features are not intended to be comprehensive, and those skilled in the art can modify, expand, or limit this set of features according to a wide variety of known image - processing - based features known in the art.

[0084] Image features can be collected for each pixel, each pixel group, each object, or each group of objects within the image. To more comprehensively characterize a region of the image or even the entire image, the distribution of the collected features can be configured into a histogram. The histogram itself can rely on several statistical features to analyze or otherwise process the incoming image feature data.

[0085] Statistical histogram features can include the following. (i) Minimum value: The smallest value of the distribution captured within the histogram. This can be characterized by the following relationship

Number

Number

Number

Number

Number

Number

Number

Mathematics

Mathematics

Mathematics

Number

Number

Number

Number

[0086] The above - mentioned statistical methods are useful for analyzing the spatial distribution of grayscale values by calculating local features at each point in the image and deriving a set of statistics from the distribution of the local features. Using these statistical methods, the texture for the analyzed region can be described and statistically defined.

[0087] Texture can be characterized using texture descriptors. Texture descriptors can be calculated over a given region of the image (described in more detail below). 1 One commonly applied texture method is the co-occurrence method introduced in Haralick, R. et al., "Texture features for image classification", IEEE Transactions of System, Man and Cybernetics, Vol. 3, pp. 610-621 (1973), which is incorporated herein by reference. In this method, the relative frequencies (relative occurrences) of pairs of gray levels of multiple pixels separated by a distance d in a direction θ are combined to form a relative displacement vector (d, θ). The relative displacement vector is calculated and stored in a matrix called the gray level co-occurrence matrix (GLCM). This matrix is used to extract second-order statistical texture features. Haralick proposed 14 different features to describe the two-dimensional probability density function p Four of these features are more commonly used than the others.

[0088] ij Textures can be characterized using texture descriptors. Texture descriptors can be calculated over a given region of an image (described in more detail below). One commonly applied texture method is the co-occurrence method introduced in Haralick, R. et al., "Texture features for image classification", IEEE Transactions of System, Man and Cybernetics, Vol. 3, pp. 610-621 (1973), which is incorporated herein by reference. In this method, the relative frequencies (relative occurrences) of pairs of gray levels of multiple pixels separated by a distance d in a direction θ are combined to form a relative displacement vector (d, θ). The relative displacement vector is calculated and stored in a matrix called the gray level co-occurrence matrix (GLCM). is the co-occurrence method. In this method, the relative frequency (relative occurrence) of intensity value pairs of a plurality of pixels separated by a distance d in the direction θ is combined to form a relative displacement vector (d, θ). The relative displacement vector is calculated and stored in a matrix called a grey level co-occurrence matrix (GLCM). This matrix is used to extract second-order statistical texture features. Haralick proposed 14 different features for describing the two-dimensional probability density function p ij and four of these features are more commonly used than the others. (i) Angular Second Moment (ASM) in the angular direction is calculated as follows.

Equation

Equation

Equation

Equation

[0089] These four features are hereby incorporated by reference as part of this specification. ​​​​​Strand, J. in his "Local frequency features for the texture classification" Pattern R ecognition, Vol. 27, No. 10, pp 1397-1406 (1994) [Strand94] is also mentioned.

[0090] For a given image, the region of the image where the above features are evaluated can be defined by a mask (e.g., a colony mask) or by a Voronoi influence region that extends beyond the mask. This can be done.

[0091] Figure 14 shows some of the regions that can be considered. Region 1410 is a colony mask that has a spread similar to that of the colony itself. Region 1420 is the Voronoi influence region of the colony (bounded by the edges of the image or plate). For further illustration, pixel 1430 is within region 1420 but outside region 1410. In other words, the colony represented by region 1410 could potentially extend to pixel 1630 but has not yet done so. Pixel 1440 is outside both regions 1410 and 1420. In other words, not only does the colony not occupy pixel 1410 at this point in the image, but it is also not predicted that pixel 1410 will be occupied by the colony at any future point (in this case, this pixel is already occupied by another colony).

[0092] Using the colony mask at multiple time points along the culture process and the associated Voronoi influence regions as described above, different aspects of the colony and the surrounding local ​​​​​​​​A plurality of histograms can be generated that show their effects on the growing medium. The colony mask and the Voronoi influence regions themselves can be adjusted over time, for example as the colonies grow. For example, FIGS. 15A - 15C show how the colony mask can be adjusted over time. FIG. 15A is a portion of a blood culture after 24 hours of growth on an agar medium. FIG. 15B is a contrast image compared to an image captured at a past t FIG. 15C is a grayscale image showing growth at 9 hours (brightest), 12 hours (intermediate), and 24 hours (darkest). Each shade in FIG. 15C can be used to create a different mask for each colony. 0 Alternatively or additionally, as growth occurs, the masks can be separated according to their respective Voronoi influence regions. (brightest), 12 hours (intermediate), and 24 hours (darkest). Each shade in FIG. 15C can be used to create a different mask for each colony. Alternatively or additionally, as growth occurs, the masks can be separated according to their respective Voronoi influence regions. Alternatively or additionally, as growth occurs, the masks can be separated according to their respective Voronoi influence regions. Alternatively or additionally, as growth occurs, the masks can be separated according to their respective Voronoi influence regions.

[0093] Any one or combination of the features in the above - listed feature set can be used as a set of features for capturing the specificity of organisms growing in various media on an imaging plate under various culture conditions. This list is not intended to be comprehensive, and one of ordinary skill in the art can modify, expand, or limit this set of features according to the object intended to be imaged and a wide variety of image - processing - based features known in the art. For this reason, the above - exemplified features are given by way of example and are not limiting. One of ordinary skill in the art is aware of other measurements and techniques for determining the shape and features of an object, and the above examples are given by way of example and are not limiting. One of ordinary skill in the art is aware of other measurements and techniques for determining the shape and features of an object, and the above examples are given by way of example and are not limiting. For this reason, the above - exemplified features are given by way of example and are not limiting. For this reason, the above - exemplified features are given by way of example and are not limiting.

[0094] One of ordinary skill in the art recognizes other measurements and techniques for determining the shape and features of an object, and the above examples are given by way of example and are not limiting. One of ordinary skill in the art recognizes other measurements and techniques for determining the shape and features of an object, and the above examples are given by way of example and are not limiting. ​

[0095] [Construction of Contrast] Which image in a series of images yields a value for growth detection, counting, or identification It is often difficult to initially predict which. This is due in part to the fact that image contrast varies for different colony forming units (CFUs) and across different media In a given image of some colonies, for growth detection, some colonies may have a very desirable contrast to the background, while another colony may not have an appropriate contrast to the background. This makes it difficult to identify colonies in the medium using a single approach.

[0096] Therefore, it is desirable to construct contrast from all available materials through space (spatial difference) and time (time difference under common imaging conditions), and using various imaging conditions (e.g., red channel, green channel, and blue channel, bright background and dark background, spectral images, or any other color space transformation). It is also desirable to collect contrast from multiple available sources and provide a standardized image as input to an algorithm for colony detection.

[0097] Image data can be segmented based on any number of factors. For example, image data can be limited to a specific point in time and / or specific information required (e.g., spatial image information may not require as many points in time as temporal image information may require). Lighting configuration and color space can also be selected to achieve a specific contrast goal. The desired size ​​​​To detect an object (e.g., a colony) having a size within a range (or a target range), the spatial frequency can also be varied.

[0098] To detect separate objects, the contrast can be set to an absolute value in the range [0,1] or a signed value in the range [-1, -1]. The scale and offset of the contrast output can also be specified (e.g., for an 8-bit image with signed contrast, the offset can be 127.5 and the scale can be 127.5). In the case of). In an example where the contrast is set to an extreme value, the absolute offset can be set to zero and the scale can be set to 256. Separate objects on a uniform background can be detected using spatial contrast.

[0099] Using a certain formula, the spatial contrast on the image I at the location (x,y) within the distance r can be automatically evaluated. The distance r is limited to

Number

Number

Number

[0100] Using temporal contrast, moving objects or objects that change over time (such as CFUs that appear and / or disappear on the imaging plate) can be detected. Using a certain formula, at time point t using a certain formula, at time point t0 and t x the temporal contrast on the image I at the location (x, y) between

Equation

Equation

[0101] The collection of spatial contrast can be performed in an automated fashion by generating a plurality of SHQI images of the plate according to a pre-programmed sequence. Multiple images can be generated at a given culture time to advance the colony detection investigation. In one embodiment, the image data (in particular, the vector "vect" used to provide the contrast input to the contrast collection operator) is detected at a given point in time according to the following equation from several different radii (R ~R ~R ~R ) from the colonies detected and collected over several different configurations (CFG min ~CFG max ) over several different configurations (CFG ~CFG 1 ~CFG N ).

Equation

[0102] For a given image I (x,y) (e.g., when SHQI imaging is the source), if the SNR is known, the configuration that maximizes the SNR-weighted contrast is

Equation

Equation

[0103] The contrast operator K further benefits from this known SNR information, and the above equation becomes as follows as follows.

Number

[0104] Temporal contrast collection can also be performed in an automated manner by generating a plurality of SHQI images of the plate according to a pre-programmed sequence. A plurality of images are generated over multiple culture time points to advance the colony detection investigation. At least one of the multiple culture time points is t is. In one embodiment, the image data is as follows Thus, over multiple culture time points, a plurality of images are generated to conduct a colony detection investigation. For some configurations at one or more subsequent culture time points up to time point t and time point t 0 is. In one embodiment, the image data is as follows Thus, at time point t 0 and at one or more subsequent culture time points up to time point t x is collected over some configurations. is collected over.

Number

[0105] In the above example, the vector can be the vector between these two time points based on the difference in images at two time points (e.g., t 0 and t x ). On the other hand, in other applications where another time point between t and t 0 and t x is included, the vector can be mapped over as many points as the time points at which the images are acquired. Mathematically speaking, there is no limit to the number of points that can be included in the vector. is included, the vector can be mapped over as many points as the time points at which the images are acquired. Mathematically speaking, there is no limit to the number of points that can be included in the vector. is included, the vector can be mapped over as many points as the time points at which the images are acquired. Mathematically speaking, there is no limit to the number of points that can be included in the vector. is included, the vector can be mapped over as many points as the time points at which the images are acquired. Mathematically speaking, there is no limit to the number of points that can be included in the vector.

[0106] Similar to spatial contrast, for a given image I(x、y) When the SNR is known , the configuration that maximizes the SNR-weighted contrast is

Number

Number

[0107] The contrast operator K further benefits from this known SNR information, and the above equation becomes as follows .

Number

[0108] In the above example, the Max operator can be replaced with any other statistical operator such as a percentile (e.g., Q1, median, Q3, or any other percentile), or a weighted sum. The weighting values result from preliminary work extracted from the training database, thereby expanding the field of supervised contrast extraction to neural networks . Furthermore, multiple algorithms can be used, and the results of multiple algorithms are further combined using another operator such as the Max operator.

[0109] [Image Alignment] When multiple images are obtained over time, very accurate alignment of those images is required to obtain a valid temporal estimate from them. Such alignment can be achieved by mechanical alignment devices and / or algorithms (e.g., image tracking, image matching ing). Those skilled in the art will understand these for achieving this goal ​​​ recognizes the solutions and techniques.

[0110] For example, when multiple images of an object on a plate are collected, the coordinates of the location of the object can be determined. Next, the image data of the object collected at a subsequent time point can be associated with the previous image data based on the coordinates, and then these can be used to determine the temporal changes of the object.

[0111] For the fast and useful use of the images (when used as input to a classifier, for example), it is important to store these images in a spatial reference such that the invariance of the images is maximized. Since the basic shape descriptor of the colony is usually circular, the colony images can be stored using a polar coordinate system. The center of mass of the colony can be specified as the center of the location of the colony when the colony is first detected. The center point can then serve as the origin center for the polar transformation of each subsequent image of the colony. FIG. 16A shows a zoomed-in portion of an imaging plate having a center point "O". Two rays "A" and "B" extending from the point "O" are overlaid on the image (for clarity). Each ray intersects the respective (circled) colony. The colony circled in FIG. 16A is shown in more detail in images 1611 and 1612 of FIG. 16B. In FIG. 16B, image 1611 (the colony intersecting ray "A") is rotated to become image 1613 ("A'"), and the radial axis of image 1613 is aligned with the radial axis of image 1612, whereby the orientation is The leftmost part of the transformed image is closest to the point "O" in FIG. 16A, and the rightmost part of the image with the changed orientation is farthest from the point "O". By this change in the direction of the pole, easier analysis of colonies oriented in multiple directions of the imaging plate (with respect to elements such as illumination, etc.) becomes possible. The rightmost part of the image with the changed orientation is farthest from the point "O". By this change in the direction of the pole, easier analysis of colonies oriented in multiple directions of the imaging plate (with respect to elements such as illumination, etc.) becomes possible. The rightmost part of the image with the changed orientation is farthest from the point "O". By this change in the direction of the pole, easier analysis of colonies oriented in multiple directions of the imaging plate (with respect to elements such as illumination, etc.) becomes possible. The rightmost part of the image with the changed orientation is farthest from the point "O". By this change in the direction of the pole, easier analysis of colonies oriented in multiple directions of the imaging plate (with respect to elements such as illumination, etc.) becomes possible.

[0112] In FIG. 16C, the pole transformation is completed for each of the images 1611, 1612, and 1613 in FIG. 16B. In the pole-transformed images 1621, 1622, and 1623, the radial axes (extending from the center of each imaged colony) of the images 1611, 1612, and 1613 with their respective changed orientations are plotted from left to right in the multiple images of FIG. 16C, and the angular axes (of each colony) are plotted from top to bottom. In FIG. 16C, the pole transformation is completed for each of the images 1611, 1612, and 1613 in FIG. 16B. In the pole-transformed images 1621, 1622, and 1623, the radial axes (extending from the center of each imaged colony) of the images 1611, 1612, and 1613 with their respective changed orientations are plotted from left to right in the multiple images of FIG. 16C, and the angular axes (of each colony) are plotted from top to bottom. In FIG. 16C, the pole transformation is completed for each of the images 1611, 1612, and 1613 in FIG. 16B. In the pole-transformed images 1621, 1622, and 1623, the radial axes (extending from the center of each imaged colony) of the images 1611, 1612, and 1613 with their respective changed orientations are plotted from left to right in the multiple images of FIG. 16C, and the angular axes (of each colony) are plotted from top to bottom. In FIG. 16C, the pole transformation is completed for each of the images 1611, 1612, and 1613 in FIG. 16B. In the pole-transformed images 1621, 1622, and 1623, the radial axes (extending from the center of each imaged colony) of the images 1611, 1612, and 1613 with their respective changed orientations are plotted from left to right in the multiple images of FIG. 16C, and the angular axes (of each colony) are plotted from top to bottom. In FIG. 16C, the pole transformation is completed for each of the images 1611, 1612, and 1613 in FIG. 16B. In the pole-transformed images 1621, 1622, and 1623, the radial axes (extending from the center of each imaged colony) of the images 1611, 1612, and 1613 with their respective changed orientations are plotted from left to right in the multiple images of FIG. 16C, and the angular axes (of each colony) are plotted from top to bottom.

[0113] For each pole image, a set of 1-dimensional vectors can be generated using, for example, shape features and / or histogram features (such as the average and / or standard deviation of the color or luminance of the object) along the radial axis and / or angular axis. Considering rotation, even when the shape and histogram features are almost invariant, some texture features may show large variations during rotation, so invariance is not guaranteed. Therefore, there is a great advantage in presenting each colony image from the same viewpoint or angle for each illumination. This is because then the differences in the texture of the objects can be used to distinguish them from each other. Since the illumination conditions show variations linked to the angular position around the center of the plate imaging in most cases, a ray passing through the centers of the colony and the plate (in the images 1611, 1 in FIG. 16B) For each pole image, a set of 1-dimensional vectors can be generated using, for example, shape features and / or histogram features (such as the average and / or standard deviation of the color or luminance of the object) along the radial axis and / or angular axis. Considering rotation, even when the shape and histogram features are almost invariant, some texture features may show large variations during rotation, so invariance is not guaranteed. Therefore, there is a great advantage in presenting each colony image from the same viewpoint or angle for each illumination. This is because then the differences in the texture of the objects can be used to distinguish them from each other. Since the illumination conditions show variations linked to the angular position around the center of the plate imaging in most cases, a ray passing through the centers of the colony and the plate (in the images 1611, 1 in FIG. 16B) For each pole image, a set of 1-dimensional vectors can be generated using, for example, shape features and / or histogram features (such as the average and / or standard deviation of the color or luminance of the object) along the radial axis and / or angular axis. Considering rotation, even when the shape and histogram features are almost invariant, some texture features may show large variations during rotation, so invariance is not guaranteed. Therefore, there is a great advantage in presenting each colony image from the same viewpoint or angle for each illumination. This is because then the differences in the texture of the objects can be used to distinguish them from each other. Since the illumination conditions show variations linked to the angular position around the center of the plate imaging in most cases, a ray passing through the centers of the colony and the plate (in the images 1611, 1 in FIG. 16B) For each pole image, a set of 1-dimensional vectors can be generated using, for example, shape features and / or histogram features (such as the average and / or standard deviation of the color or luminance of the object) along the radial axis and / or angular axis. Considering rotation, even when the shape and histogram features are almost invariant, some texture features may show large variations during rotation, so invariance is not guaranteed. Therefore, there is a great advantage in presenting each colony image from the same viewpoint or angle for each illumination. This is because then the differences in the texture of the objects can be used to distinguish them from each other. Since the illumination conditions show variations linked to the angular position around the center of the plate imaging in most cases, a ray passing through the centers of the colony and the plate (in the images 1611, 1 in FIG. 16B) For each pole image, a set of 1-dimensional vectors can be generated using, for example, shape features and / or histogram features (such as the average and / or standard deviation of the color or luminance of the object) along the radial axis and / or angular axis. Considering rotation, even when the shape and histogram features are almost invariant, some texture features may show large variations during rotation, so invariance is not guaranteed. Therefore, there is a great advantage in presenting each colony image from the same viewpoint or angle for each illumination. This is because then the differences in the texture of the objects can be used to distinguish them from each other. Since the illumination conditions show variations linked to the angular position around the center of the plate imaging in most cases, a ray passing through the centers of the colony and the plate (in the images 1611, 1 in FIG. 16B) For each pole image, a set of 1-dimensional vectors can be generated using, for example, shape features and / or histogram features (such as the average and / or standard deviation of the color or luminance of the object) along the radial axis and / or angular axis. Considering rotation, even when the shape and histogram features are almost invariant, some texture features may show large variations during rotation, so invariance is not guaranteed. Therefore, there is a great advantage in presenting each colony image from the same viewpoint or angle for each illumination. This is because then the differences in the texture of the objects can be used to distinguish them from each other. Since the illumination conditions show variations linked to the angular position around the center of the plate imaging in most cases, a ray passing through the centers of the colony and the plate (in the images 1611, 1 in FIG. 16B) For each pole image, a set of 1-dimensional vectors can be generated using, for example, shape features and / or histogram features (such as the average and / or standard deviation of the color or luminance of the object) along the radial axis and / or angular axis. Considering rotation, even when the shape and histogram features are almost invariant, some texture features may show large variations during rotation, so invariance is not guaranteed. Therefore, there is a great advantage in presenting each colony image from the same viewpoint or angle for each illumination. This is because then the differences in the texture of the objects can be used to distinguish them from each other. Since the illumination conditions show variations linked to the angular position around the center of the plate imaging in most cases, a ray passing through the centers of the colony and the plate (in the images 1611, 1 in FIG. 16B) For each pole image, a set of 1-dimensional vectors can be generated using, for example, shape features and / or histogram features (such as the average and / or standard deviation of the color or luminance of the object) along the radial axis and / or angular axis. Considering rotation, even when the shape and histogram features are almost invariant, some texture features may show large variations during rotation, so invariance is not guaranteed. Therefore, there is a great advantage in presenting each colony image from the same viewpoint or angle for each illumination. This is because then the differences in the texture of the objects can be used to distinguish them from each other. Since the illumination conditions show variations linked to the angular position around the center of the plate imaging in most cases, a ray passing through the centers of the colony and the plate (in the images 1611, 1 in FIG. 16B) For each pole image, a set of 1-dimensional vectors can be generated using, for example, shape features and / or histogram features (such as the average and / or standard deviation of the color or luminance of the object) along the radial axis and / or angular axis. Considering rotation, even when the shape and histogram features are almost invariant, some texture features may show large variations during rotation, so invariance is not guaranteed. Therefore, there is a great advantage in presenting each colony image from the same viewpoint or angle for each illumination. This is because then the differences in the texture of the objects can be used to distinguish them from each other. Since the illumination conditions show variations linked to the angular position around the center of the plate imaging in most cases, a ray passing through the centers of the colony and the plate (in the images 1611, 1 in FIG. 16B) (shown as lines in 612 and 1613 respectively) can serve as the origin (Θ) for the polar transformation of each image. can serve as the origin (Θ).

[0114] Another alignment issue arises because the plate medium is not completely frozen and solid, which may cause a slight shift from one obtained one to the next. Therefore, the area of the plate at a specific coordinate of an image acquired at a certain point in time may not necessarily exactly match the area of the plate at the same coordinate acquired at a later point in time. In other words, a slight distortion of the medium may introduce a slight uncertainty regarding the exact matching between a given pixel and the corresponding pixel captured at different points in time during the cultivation process.

[0115] To account for this uncertainty, the given pixel luminance value a at time point t

Number

Number

[0116] A person skilled in the art would b from the source image of t b two images of t, that is, b the gray level dilation (referred to as [Mathematics]) [Mathematics] and the first image corresponding to (referred to as [Mathematics]) b the gray level erosion [Mathematics] and generating the second image corresponding to (referred to as [Mathematics]) is recognized as an effective solution. Both of these have a kernel size that matches the uncertainty d of the repositioning distance. Have.

[0117] [Mathematics] In the case of, the contrast is 0, and in other cases, the contrast is calculated using the following formula [Mathematics] and [Mathematics] between [Mathematics] is estimated from the value closest to. [Mathematics]

[0118] [Improvement of SNR] Under normal lighting conditions, photon shot noise (statistical fluctuations in the arrival rate of photons reaching the sensor) limits the SNR of the detection system. Recent sensors have a full well capacity of approximately 1700 to approximately 1900 electrons per effective square micron. Therefore, when imaging an object on the plate, the main concern is not the number of pixels used to image the object, but the area covered by the object within the sensor space. Increasing the area of the sensor improves the SNR for the imaged object. The image quality can be improved by capturing the image using lighting conditions where photon noise dominates the SNR (without saturating the sensor (the maximum number of photons that can be recorded for each pixel per frame)). Image averaging techniques are commonly used to maximize the SNR. Since the SNR in dark regions is much lower than that in bright regions, these techniques are used to handle images with large brightness (or color) differences, as shown by the following equation. However, I is the average current generated by the electron stream in the sensor. When color is perceived due to differences in absorption and reflection of substances and light across the electromagnetic spectrum, the reliability of the captured color depends on the ability of the system to record luminance with a high SNR. Image sensors (e.g., CCD sensors, CMOS sensors, etc.) are known to those skilled in the art. 、When imaging an object on the plate, the main concern is not the number of pixels used to image the object, but the area covered by the object within the sensor space. Increasing the area of the sensor improves the SNR for the imaged object. The image quality can be improved by capturing the image using lighting conditions where photon noise dominates the SNR (without saturating the sensor (the maximum number of photons that can be recorded for each pixel per frame)). Image averaging techniques are commonly used to maximize the SNR. Since the SNR in dark regions is much lower than that in bright regions, these techniques are used to handle images with large brightness (or color) differences, as shown by the following equation. However, I is the average current generated by the electron stream in the sensor. When color is perceived due to differences in absorption and reflection of substances and light across the electromagnetic spectrum, the reliability of the captured color depends on the ability of the system to record luminance with a high SNR. Image sensors (e.g., CCD sensors, CMOS sensors, etc.) are known to those skilled in the art.

[0119] The image quality can be improved by capturing the image using lighting conditions where photon noise dominates the SNR (without saturating the sensor (the maximum number of photons that can be recorded for each pixel per frame)). Image averaging techniques are commonly used to maximize the SNR. Since the SNR in dark regions is much lower than that in bright regions, these techniques are used to handle images with large brightness (or color) differences, as shown by the following equation. However, I is the average current generated by the electron stream in the sensor. When color is perceived due to differences in absorption and reflection of substances and light across the electromagnetic spectrum, the reliability of the captured color depends on the ability of the system to record luminance with a high SNR. Image sensors (e.g., CCD sensors, CMOS sensors, etc.) are known to those skilled in the art.

Number

Number

[0120] However, I is the average current generated by the electron stream in the sensor. When color is perceived due to differences in absorption and reflection of substances and light across the electromagnetic spectrum, the reliability of the captured color depends on the ability of the system to record luminance with a high SNR. Image sensors (e.g., CCD sensors, CMOS sensors, etc.) are known to those skilled in the art. When color is perceived due to differences in absorption and reflection of substances and light across the electromagnetic spectrum, the reliability of the captured color depends on the ability of the system to record luminance with a high SNR. Image sensors (e.g., CCD sensors, CMOS sensors, etc.) are known to those skilled in the art. When color is perceived due to differences in absorption and reflection of substances and light across the electromagnetic spectrum, the reliability of the captured color depends on the ability of the system to record luminance with a high SNR. Image sensors (e.g., CCD sensors, CMOS sensors, etc.) are known to those skilled in the art. When color is perceived due to differences in absorption and reflection of substances and light across the electromagnetic spectrum, the reliability of the captured color depends on the ability of the system to record luminance with a high SNR. Image sensors (e.g., CCD sensors, CMOS sensors, etc.) are known to those skilled in the art. This will not be described in detail in this specification.

[0121] To overcome the imaging limitations of conventional SNR, the imaging system performs an analysis of the imaging rate at the time of image acquisition and, based on the analysis, adjusts the illumination conditions and exposure time in real time. This process is described in Patent Document 1, which is hereby incorporated by reference in its entirety, and is generally referred to as supervised high-quality imaging (SHQI). The system can also customize the imaging conditions for various brightness regions of the plates within different color channels. This process is incorporated herein by reference and is generally referred to as supervised high-quality imaging (SHQI). This process is described in Patent Document 1, which is hereby incorporated by reference in its entirety, and is generally referred to as supervised high-quality imaging (SHQI). The system can also customize the imaging conditions for various brightness regions of the plates within different color channels. For a given pixel x, y of an image, the SNR information of the pixel acquired in the current frame N can be combined with the SNR information of the same pixel acquired in a previously or subsequently acquired frame (e.g., N-1, N+1). As an example, the combined SNR is represented by the following equation.

[0122] For a given pixel x, y of an image, the SNR information of the pixel acquired in the current frame N can be combined with the SNR information of the same pixel acquired in a previously or subsequently acquired frame (e.g., N-1, N+1). For a given pixel x, y of an image, the SNR information of the pixel acquired in the current frame N can be combined with the SNR information of the same pixel acquired in a previously or subsequently acquired frame (e.g., N-1, N+1). For a given pixel x, y of an image, the SNR information of the pixel acquired in the current frame N can be combined with the SNR information of the same pixel acquired in a previously or subsequently acquired frame (e.g., N-1, N+1). As an example, the combined SNR is represented by the following equation. For a given pixel x, y of an image, the SNR information of the pixel acquired in the current frame N can be combined with the SNR information of the same pixel acquired in a previously or subsequently acquired frame (e.g., N-1, N+1). As an example, the combined SNR is represented by the following equation.

Equation

[0123] After updating the image data with a new acquisition, the acquisition system can predict the best next acquisition time to maximize the SNR according to environmental constraints (e.g., the minimum SNR required per pixel within the object area). For example, by merging the information of two images captured under optimal illumination conditions in bright and dark conditions, when the SNR in the dark region increases by only √11 in only two acquisitions, taking the average of five images captured in the non-saturated state increases the SNR in the dark region (10% of the maximum luminance) by only √5. After updating the image data with a new acquisition, the acquisition system can predict the best next acquisition time to maximize the SNR according to environmental constraints (e.g., the minimum SNR required per pixel within the object area). After updating the image data with a new acquisition, the acquisition system can predict the best next acquisition time to maximize the SNR according to environmental constraints (e.g., the minimum SNR required per pixel within the object area). For example, by merging the information of two images captured under optimal illumination conditions in bright and dark conditions, when the SNR in the dark region increases by only √11 in only two acquisitions, taking the average of five images captured in the non-saturated state increases the SNR in the dark region (10% of the maximum luminance) by only √5. After updating the image data with a new acquisition, the acquisition system can predict the best next acquisition time to maximize the SNR according to environmental constraints (e.g., the minimum SNR required per pixel within the object area). For example, by merging the information of two images captured under optimal illumination conditions in bright and dark conditions, when the SNR in the dark region increases by only √11 in only two acquisitions, taking the average of five images captured in the non-saturated state increases the SNR in the dark region (10% of the maximum luminance) by only √5. After updating the image data with a new acquisition, the acquisition system can predict the best next acquisition time to maximize the SNR according to environmental constraints (e.g., the minimum SNR required per pixel within the object area). For example, by merging the information of two images captured under optimal illumination conditions in bright and dark conditions, when the SNR in the dark region increases by only √11 in only two acquisitions, taking the average of five images captured in the non-saturated state increases the SNR in the dark region (10% of the maximum luminance) by only √5. After updating the image data with a new acquisition, the acquisition system can predict the best next acquisition time to maximize the SNR according to environmental constraints (e.g., the minimum SNR required per pixel within the object area). For example, by merging the information of two images captured under optimal illumination conditions in bright and dark conditions, when the SNR in the dark region increases by only √11 in only two acquisitions, taking the average of five images captured in the non-saturated state increases the SNR in the dark region (10% of the maximum luminance) by only √5.

[0124] [Modeling of Images] Depending on the situation, when calculating the spatial or temporal contrast between pixels of one or more images, the pixel information for a given image may not be available or may be degraded. Unavailability may occur, for example, when an image of the plate was not captured within the time before cell growth (e.g., the plate was not imaged at or immediately after time point t ). Degradation of signal information may occur, for example, when an image was captured at time point t 0 but the pixels of the captured image do not accurately reflect the imaging plate before bacterial growth . Such inaccuracies may be caused by temporary artifacts (e.g., condensation temporarily formed under the plate due to thermal shock when the plate was first placed in the incubator 0 ) that do not reappear within subsequent time-series images of the plate . In such situations, an unavailable or degraded image (or certain pixels of an image) can be replaced with a model image of the plate or improved using this model image . The model image can provide pixel information that reflects how the plate is expected to be examined at a particular point in time of the unavailable or degraded image . For the model image at time point t , the model can be a planar or standard image of the plate and can be constructed arithmetically using 3D imaging and modeling techniques

[0125] . The model can provide pixel information that reflects how the plate is expected to be examined at a particular point in time of the unavailable or degraded image . For the model image at time point t , the model can be a planar or standard image of the plate and can be constructed arithmetically using 3D imaging and modeling techniques . The model can provide pixel information that reflects how the plate is expected to be examined at a particular point in time of the unavailable or degraded image . For the model image at time point t 0 , the model can be a planar or standard image of the plate and can be constructed arithmetically using 3D imaging and modeling techniques . The model can generate a model that is as realistic as possible by using physical design parameters (e.g., ​, diameter, height, partitions for accommodating a plurality of culture media, plastic materials, etc.), and culture medium para meters (e.g., type of culture medium, composition of culture medium, height or thickness of culture medium, etc.), and lighting parameters ters (e.g., angle of light source, color of light source or (plural wavelengths), color of background, etc.), and positioning para meters (e.g., position of plate in imaging chamber) can each be included .

[0126] In the case of a degraded image, the improvement of the degraded image is such that (e.g., the dew condensation below blocks some light from passing through the plate, whereby the section of the plate with dew condensation becomes slightly less transparent) the pixel information is not as sharp as the other parts of the image . When this happens, the degraded pixel features of the image can be sharpened by using signal recovery . The recovery of the signal can include obtaining luminance information for other parts of the image , identifying the median luminance value of the luminance information, and then replacing the luminance information for the less sharp regions of the image with the median luminance value of the remaining image.

[0127] [Application Example] The following disclosure is mainly based on tests conducted in various dilutions of physiological saline to simulate the standard urine reported volume (CFU / ml bucket group). The suspension for each isolate was adapted to a 0.5 McFarland turbidity standard solution and placed in a BD Urine Vacutainer tube (model number 364951) at an estimated 1×10 , 6 , 1×10 5 , 5×10 4 , 1×10 4 , 1×10 3 and 1×10 2 CFU / ml suspension It was used to prepare dilutions. The specimen tubes were streaked with standard urine patterns, i.e., #4 Zigzag (dispensing 0.01 ml per plate), and processed using Kiestra InoqulA (WCA1).

[0128] The plates were processed using ReadA Compact (35 °C, non-CO 2 ) and imaged every 2 hours for the first 24 hours and every 6 hours for the next 24 hours, for a total of 48 hours of culture. The culture time was input as the first reading at 1 hour, with an acceptable margin of ±15 minutes. The next readings from 2 to 24 hours were set every 2 hours with an acceptable margin of ±30 minutes. The readings from 24 to 48 were set every 6 hours with an acceptable margin of ±30 minutes. After a pure feasibility investigation, this was changed to have no acceptable margin. By doing this, image acquisition was improved in the desired range of 18 to 24 hours.

[0129] In other cases, over a 48-hour period, images can be obtained at 2-hour intervals for the first 24 hours and at 6-hour intervals for the next 24 hours. In this case, over a 48-hour period, a total of 17 images are obtained, including the image obtained at time point t (0 hours). 0

[0130] All acquired images were spectrally balanced for geometric and chromatic aberrations of the lens using the pixel size of known objects, the normalized illumination conditions, and a high signal-to-noise ratio for each band in pixel units. Cameras suitable for use in the methods and systems described herein are known to those skilled in the art and are not described in detail herein. ​ Do not. As an example, capture an image of a 90 mm plate using a 4 megapixel camera such that colonies are within a range of 100 μm in diameter with appropriate contrast, when there are up to 30 colonies / mm 2 of local density (> 10 5 CFU / plate) should be countable.

[0131] The following media were used to evaluate the contrast of colonies grown thereon.

[0132] TSA II 5% sheep blood (BAP): A non-selective medium widely used for urine culture.

[0133] BAV: Used for colony counting and presumptive ID based on colony morphology and hemolysis

[0134] MacConkey II agar medium (MAC): A selective medium for the most common Gram-negative UTI pathogens MAC is used for differentiation of lactose that produces colonies. MAC also inhibits the swarming of Proteus. BAP and MAC are widely commonly used for urine culture. Some media are not recommended for use in colony counting due to partial inhibition of some Gram-negative bacteria.

[0135] Colistin nalidixic acid agar medium (CNA): A selective medium for the most common Gram-positive UTI pathogens CNA is not as commonly used as MAC for urine culture, but is useful for colony identification when excessive growth of Gram-negative colonies occurs.

[0136] CHROMAgar Orientation (CHROMA): Widely used for urine culture ​​​​The non-selective medium used. CHROMA is used for the counting and ID of colonies based on the color and morphology of colonies. Escherichia coli and Enterococcus spp. are identified by this medium and do not require confirmatory testing. CHROMA is not used as much as BAP due to cost. In the case of mixed samples, CLED medium was also used.

[0137] Cysteine-Lactose-Electrolyte-Deficient (CLED) agar medium: Used for colony counting and presumptive ID of urinary pathogens based on lactose fermentation.

[0138] Specimen processing Becton Dickinson Kiestra™ InoqulA™ was used to automate the processing of bacteriological specimens, enable standardization, and ensure consistent high-quality streaking. The Becton Dickinson Kiestra™ InoqulA™ specimen processor uses magnetic rolling bead technology and streaks agar plates using a customizable pattern. The magnetic rolling beads are 5 mm in diameter.

[0139] Figure 17 shows an example of the performance of the contrast collection algorithm when retrieving Morganella growth in both CHROMagar (upper figure) and blood agar medium (lower figure) media. Both CHROMagar and BAP are non-selective growth media commonly used in microbiological research. Each of the central images in Figure 17 represents an imaging plate illuminated by light from above the plate (top illumination). The left image represents the spatial contrast corresponding to the central image. The spatial contrast is based on a kernel with a median of 1 millimeter. Finally, the right image represents the temporal contrast of the central image. Trust uses various illumination conditions (different color channels, lighting settings, etc.) at the time of capturing each image, and is edited from images captured at 0 hours (t ) to 12 hours (t 0 ) of the culture. x ) .

[0140] As shown in FIG. 17, local contrast has limitations when dealing with translucent colonies, especially when the edge transition is small. This is because when using the spatial contrast algorithm alone, it may only be able to select regions in the most contrasty confluent state from the image, since there is a possibility that it cannot capture the object. From FIG. 17, the effectiveness of temporal contrast in separated colonies is also clear.

[0141] Also, FIG. 17 (especially the image of the blood agar medium at the bottom) emphasizes the problem of using spatial contrast alone to detect growth in a transparent medium. The ink printed on the transparent case has very prominent edges, so it has prominent spatial contrast. As a result, it becomes very difficult to see any other spatial contrast, because the colonies do not have edges made to be equally prominent. Therefore, the use of temporal contrast in combination with spatial contrast is very beneficial in the present disclosure.

[0142] Finally, the result of the contrast determination described above can automate the method for high-speed detection and identification of colonies in the imaged medium. The automated method provides a significant advantage over comparable manual methods.

[0143] Figure 18 shows a pair of flowcharts comparing the timeline of an automated inspection process 1800 (e.g., routine 200 of FIG. 2) with the timeline of a comparable, manually executed inspection process 1805. Each process begins by receiving (1810, 1815) a specimen for inspection at a laboratory. Next, each process proceeds to culturing 1820, 1825, during which the specimen can be imaged multiple times. In the automated process, an automated evaluation 1830 is performed generally after 12 hours of culturing, and after that time, it is possible to clearly determine (1840) whether there is no growth (or normal growth) in the specimen. As shown by the results in FIG. 17, the use of time contrast in the automated process significantly improves the ability to detect colonies even after only 12 hours have elapsed. In contrast, a manual evaluation 1835 in the manual process cannot be performed until approximately 24 hours have elapsed since the start of the culturing process. Only after 24 hours have elapsed can it be clearly determined (1845) whether there is no growth (or normal growth) in the specimen. with the timeline of a comparable, manually executed inspection process 1805 and shows a pair of flowcharts. Each process begins by receiving (1810, 1815) a specimen for inspection at a laboratory. Next, each process proceeds to culturing 1820, 1825, during which the specimen can be imaged multiple times. In the automated process, an automated evaluation 1830 is performed generally after 12 hours of culturing, and after that time, it is possible to clearly determine (1840) whether there is no growth (or normal growth) in the specimen. As shown by the results in FIG. 17, the use of time contrast in the automated process significantly improves the ability to detect colonies even after only 12 hours have elapsed. In contrast, a manual evaluation 1835 in the manual process cannot be performed until approximately 24 hours have elapsed since the start of the culturing process. Only after 24 hours have elapsed can it be clearly determined (1845) whether there is no growth (or normal growth) in the specimen. and shows a pair of flowcharts. Each process begins by receiving (1810, 1815) a specimen for inspection at a laboratory. Next, each process proceeds to culturing 1820, 1825, during which the specimen can be imaged multiple times. In the automated process, an automated evaluation 1830 is performed generally after 12 hours of culturing, and after that time, it is possible to clearly determine (1840) whether there is no growth (or normal growth) in the specimen. As shown by the results in FIG. 17, the use of time contrast in the automated process significantly improves the ability to detect colonies even after only 12 hours have elapsed. In contrast, a manual evaluation 1835 in the manual process cannot be performed until approximately 24 hours have elapsed since the start of the culturing process. Only after 24 hours have elapsed can it be clearly determined (1845) whether there is no growth (or normal growth) in the specimen. 1820, 1825, during which the specimen can be imaged multiple times. In the automated process, an automated evaluation 1830 is performed generally after 12 hours of culturing, and after that time, it is possible to clearly determine (1840) whether there is no growth (or normal growth) in the specimen. As shown by the results in FIG. 17, the use of time contrast in the automated process significantly improves the ability to detect colonies even after only 12 hours have elapsed. In contrast, a manual evaluation 1835 in the manual process cannot be performed until approximately 24 hours have elapsed since the start of the culturing process. Only after 24 hours have elapsed can it be clearly determined (1845) whether there is no growth (or normal growth) in the specimen. In the automated process, an automated evaluation 1830 is performed generally after 12 hours of culturing, and after that time, it is possible to clearly determine (1840) whether there is no growth (or normal growth) in the specimen. As shown by the results in FIG. 17, the use of time contrast in the automated process significantly improves the ability to detect colonies even after only 12 hours have elapsed. In contrast, a manual evaluation 1835 in the manual process cannot be performed until approximately 24 hours have elapsed since the start of the culturing process. Only after 24 hours have elapsed can it be clearly determined (1845) whether there is no growth (or normal growth) in the specimen. and after that time, it is possible to clearly determine (1840) whether there is no growth (or normal growth) in the specimen. As shown by the results in FIG. 17, the use of time contrast in the automated process significantly improves the ability to detect colonies even after only 12 hours have elapsed. In contrast, a manual evaluation 1835 in the manual process cannot be performed until approximately 24 hours have elapsed since the start of the culturing process. Only after 24 hours have elapsed can it be clearly determined (1845) whether there is no growth (or normal growth) in the specimen. As shown by the results in FIG. 17, the use of time contrast in the automated process significantly improves the ability to detect colonies even after only 12 hours have elapsed. In contrast, a manual evaluation 1835 in the manual process cannot be performed until approximately 24 hours have elapsed since the start of the culturing process. Only after 24 hours have elapsed can it be clearly determined (1845) whether there is no growth (or normal growth) in the specimen. even after only 12 hours have elapsed. In contrast, a manual evaluation 1835 in the manual process cannot be performed until approximately 24 hours have elapsed since the start of the culturing process. Only after 24 hours have elapsed can it be clearly determined (1845) whether there is no growth (or normal growth) in the specimen. In contrast, a manual evaluation 1835 in the manual process cannot be performed until approximately 24 hours have elapsed since the start of the culturing process. Only after 24 hours have elapsed can it be clearly determined (1845) whether there is no growth (or normal growth) in the specimen. until approximately 24 hours have elapsed since the start of the culturing process. Only after 24 hours have elapsed can it be clearly determined (1845) whether there is no growth (or normal growth) in the specimen. Only after 24 hours have elapsed can it be clearly determined (1845) whether there is no growth (or normal growth) in the specimen. whether there is no growth (or normal growth) in the specimen.

[0144] Figure 19 shows a timeline for determining colony semi - quantification. At 2000, a plate 2005 is inoculated. At 2010, a reference image of the newly inoculated plate 2005 is obtained. This is the background image described above in this specification. During a predetermined period, culturing proceeds, and after this period, at 2015, another image of the plate 2005 is obtained. This image may show an object or artifact change indicating microbial growth. At 2000, a plate 2005 is inoculated. At 2010, a reference image of the newly inoculated plate 2005 is obtained. This is the background image described above in this specification. During a predetermined period, culturing proceeds, and after this period, at 2015, another image of the plate 2005 is obtained. This image may show an object or artifact change indicating microbial growth. At 2010, a reference image of the newly inoculated plate 2005 is obtained. This is the background image described above in this specification. During a predetermined period, culturing proceeds, and after this period, at 2015, another image of the plate 2005 is obtained. This image may show an object or artifact change indicating microbial growth. This is the background image described above in this specification. During a predetermined period, culturing proceeds, and after this period, at 2015, another image of the plate 2005 is obtained. This image may show an object or artifact change indicating microbial growth. During a predetermined period, culturing proceeds, and after this period, at 2015, another image of the plate 2005 is obtained. This image may show an object or artifact change indicating microbial growth. For evidence of colonization, it is compared with the images obtained in 2010. If there is no sign of microbial growth, the plate is cultured and after a predetermined time, another image is obtained in 2020. In step 2020, a change in the object or artifact is detected as evidence of microbial growth. From the perspective of image analysis, the imaged object is identified (based on the difference between the object and its adjacent surroundings), and then growth is detected from the image by identifying the change in the object over time. As described in more detail above, both these differences and changes take the form of "contrast". In addition to detecting growth, the image analysis in 2020 can further include determining whether the object identified as a colony has an appropriate biomass for inclusion in the analysis. Figure 19 shows that in step 2020, sufficient biomass has not been identified and further culturing is required. In 2025, evidence of colonies is shown in the image, but these colony objects need to be further evaluated to not only quantify the colonies but also determine the nature of the colonies (i.e., whether they are sisters of other identified colony objects, pure (i.e., single microbial species) colonies, or mixed (i.e., multiple microbial species) colonies). The confluence state of the colonies and the method for its determination are described in the above section that explained the segmentation of the object. If no growth is found after a predetermined time, or only a small amount of growth is found, in 2022, plate 2005 is released as sterile. The final report is likely to indicate no significant growth or report normal flow-through growth. From the perspective of image analysis, the imaged object is identified (based on the difference between the object and its adjacent surroundings), and then growth is detected from the image by identifying the change in the object over time. As described in more detail above, both these differences and changes take the form of "contrast". In addition to detecting growth, the image analysis in 2020 can further include determining whether the object identified as a colony has an appropriate biomass for inclusion in the analysis. Figure 19 shows that in step 2020, sufficient biomass has not been identified and further culturing is required. In 2025, evidence of colonies is shown in the image, but these colony objects need to be further evaluated to not only quantify the colonies but also determine the nature of the colonies (i.e., whether they are sisters of other identified colony objects, pure (i.e., single microbial species) colonies, or mixed (i.e., multiple microbial species) colonies). The confluence state of the colonies and the method for its determination are described in the above section that explained the segmentation of the object. If no growth is found after a predetermined time, or only a small amount of growth is found, in 2022, plate 2005 is released as sterile. The final report is likely to indicate no significant growth or report normal flow-through growth. In 2025, evidence of colonies is shown in the image, but these colony objects need to be further evaluated to not only quantify the colonies but also determine the nature of the colonies (i.e., whether they are sisters of other identified colony objects, pure (i.e., single microbial species) colonies, or mixed (i.e., multiple microbial species) colonies). The confluence state of the colonies and the method for its determination are described in the above section that explained the segmentation of the object. If no growth is found after a predetermined time, or only a small amount of growth is found, in 2022, plate 2005 is released as sterile. The final report is likely to indicate no significant growth or report normal flow-through growth. If no growth is found after a predetermined time, or only a small amount of growth is found, in 2022, plate 2005 is released as sterile. The final report is likely to indicate no significant growth or report normal flow-through growth. If no growth is found after a predetermined time, or only a small amount of growth is found, in 2022, plate 2005 is released as sterile. The final report is likely to indicate no significant growth or report normal flow-through growth.

[0145] Biomass is determined by relating the object to the biomass. Step 2: The size of the sample is determined and compared to a predefined object size that represents a measure of biomass. In 025, when the object is formed from a pure sample, the object is covered by When the area of ​​the culture medium is equal to or greater than the first criterion, the biomass is selected for downstream testing. If the biological sample is not a pure sample, the inoculated sample dish may be further cultured. In step 2025, all the objects are selected based on the images obtained. If not, the inoculated culture plate is further cultured and then re-inoculated in step 2030. The size of the object associated with the colony in the image is determined based on a second predetermined size. If the size of the object, and therefore the biomass, exceeds the second criterion, At least a portion of the biomass is selected. The second predetermined criterion is that i) the object is pure ii) the reference area covered by the biomass if the object is from a sample; or the diameter of the biomass in the case of an impure sample. The predetermined reference biomass is greater than the first reference. This is step 20 in FIG. By comparing the size of the object at 25 with the size of the object at step 2030 It can be seen that the object in step 2030 is smaller than the object in step 2025. Significantly large, but you are confident that the object / biomass is from a pure sample , the objects identified in step 2025 can be selected.

[0146] If a biological specimen is determined to exhibit quantitatively significant growth, When the target colony to be processed is detected, one or more colonies can be identified as candidate colonies to be selected for analysis in 2025 or 2030. Selecting colonies can be a fully automated process, in which each of the selected colonies is sampled and tested. Alternatively, selecting colonies can be a partially automated process, in which multiple candidate colonies are automatically identified and presented visually to an operator in a digital image, whereby the operator can input a selection of one or more candidates for sampling and further testing. Sampling of the selected or picked colonies can itself be automated by the system. The time at which colonies can be selected with respect to 2025 and 2030 depends on the information about the observed colonies at the time first observed in 2025. If the colonies evolve from what is considered a pure sample (such as a deep scratch or cut), the colonies can be selected immediately if there is sufficient quantity to observe and collect them. Time is crucial when processing such samples, and eliminating one or more additional culture cycles can mean providing test results hours earlier. If the sample is estimated to be impure, additional culture cycles are required so that the colonies can grow further to assist in the identification and quantification of these colonies. colonies can be identified as candidate colonies to be selected for analysis. The selection of colonies can be a fully automated process, in which each of the selected colonies is sampled and tested. Alternatively, selecting colonies can be a fully automated process, in which each of the selected colonies is sampled and tested. Alternatively, selecting colonies can be a fully automated process, in which each of the selected colonies is sampled and tested. Alternatively, selecting colonies can be a partially automated process, in which multiple candidate colonies are automatically identified and presented visually to an operator in a digital image, whereby the operator can input a selection of one or more candidates for sampling and further testing. Sampling of the selected or picked colonies can itself be automated by the system. The time at which colonies can be selected with respect to 2025 and 2030 depends on the information about the observed colonies at the time first observed in 2025. If the colonies evolve from what is considered a pure sample (such as a deep scratch or cut), the colonies can be selected immediately if there is sufficient quantity to observe and collect them. Time is crucial when processing such samples, and eliminating one or more additional culture cycles can mean providing test results hours earlier. If the sample is estimated to be impure, additional culture cycles are required so that the colonies can grow further to assist in the identification and quantification of these colonies. selecting colonies can be a partially automated process, in which multiple candidate colonies are automatically identified and presented visually to an operator in a digital image, whereby the operator can input a selection of one or more candidates for sampling and further testing. Sampling of the selected or picked colonies can itself be automated by the system. The time at which colonies can be selected with respect to 2025 and 2030 depends on the information about the observed colonies at the time first observed in 2025. If the colonies evolve from what is considered a pure sample (such as a deep scratch or cut), the colonies can be selected immediately if there is sufficient quantity to observe and collect them. Time is crucial when processing such samples, and eliminating one or more additional culture cycles can mean providing test results hours earlier. If the sample is estimated to be impure, additional culture cycles are required so that the colonies can grow further to assist in the identification and quantification of these colonies. selecting colonies can be a partially automated process, in which multiple candidate colonies are automatically identified and presented visually to an operator in a digital image, whereby the operator can input a selection of one or more candidates for sampling and further testing. Sampling of the selected or picked colonies can itself be automated by the system. The time at which colonies can be selected with respect to 2025 and 2030 depends on the information about the observed colonies at the time first observed in 2025. If the colonies evolve from what is considered a pure sample (such as a deep scratch or cut), the colonies can be selected immediately if there is sufficient quantity to observe and collect them. Time is crucial when processing such samples, and eliminating one or more additional culture cycles can mean providing test results hours earlier. If the sample is estimated to be impure, additional culture cycles are required so that the colonies can grow further to assist in the identification and quantification of these colonies. selecting colonies can be a partially automated process, in which multiple candidate colonies are automatically identified and presented visually to an operator in a digital image, whereby the operator can input a selection of one or more candidates for sampling and further testing. Sampling of the selected or picked colonies can itself be automated by the system. The time at which colonies can be selected with respect to 2025 and 2030 depends on the information about the observed colonies at the time first observed in 2025. If the colonies evolve from what is considered a pure sample (such as a deep scratch or cut), the colonies can be selected immediately if there is sufficient quantity to observe and collect them. Time is crucial when processing such samples, and eliminating one or more additional culture cycles can mean providing test results hours earlier. If the sample is estimated to be impure, additional culture cycles are required so that the colonies can grow further to assist in the identification and quantification of these colonies. selecting colonies can be a partially automated process, in which multiple candidate colonies are automatically identified and presented visually to an operator in a digital image, whereby the operator can input a selection of one or more candidates for sampling and further testing. Sampling of the selected or picked colonies can itself be automated by the system. The time at which colonies can be selected with respect to 2025 and 2030 depends on the information about the observed colonies at the time first observed in 2025. If the colonies evolve from what is considered a pure sample (such as a deep scratch or cut), the colonies can be selected immediately if there is sufficient quantity to observe and collect them. Time is crucial when processing such samples, and eliminating one or more additional culture cycles can mean providing test results hours earlier. If the sample is estimated to be impure, additional culture cycles are required so that the colonies can grow further to assist in the identification and quantification of these colonies. selecting colonies can be a partially automated process, in which multiple candidate colonies are automatically identified and presented visually to an operator in a digital image, whereby the operator can input a selection of one or more candidates for sampling and further testing. Sampling of the selected or picked colonies can itself be automated by the system. The time at which colonies can be selected with respect to 2025 and 2030 depends on the information about the observed colonies at the time first observed in 2025. If the colonies evolve from what is considered a pure sample (such as a deep scratch or cut), the colonies can be selected immediately if there is sufficient quantity to observe and collect them. Time is crucial when processing such samples, and eliminating one or more additional culture cycles can mean providing test results hours earlier. If the sample is estimated to be impure, additional culture cycles are required so that the colonies can grow further to assist in the identification and quantification of these colonies. selecting colonies can be a partially automated process, in which multiple candidate colonies are automatically identified and presented visually to an operator in a digital image, whereby the operator can input a selection of one or more candidates for sampling and further testing. Sampling of the selected or picked colonies can itself be automated by the system. The time at which colonies can be selected with respect to 2025 and 2030 depends on the information about the observed colonies at the time first observed in 2025. If the colonies evolve from what is considered a pure sample (such as a deep scratch or cut), the colonies can be selected immediately if there is sufficient quantity to observe and collect them. Time is crucial when processing such samples, and eliminating one or more additional culture cycles can mean providing test results hours earlier. If the sample is estimated to be impure, additional culture cycles are required so that the colonies can grow further to assist in the identification and quantification of these colonies. selecting colonies can be a partially automated process, in which multiple candidate colonies are automatically identified and presented visually to an operator in a digital image, whereby the operator can input a selection of one or more candidates for sampling and further testing. Sampling of the selected or picked colonies can itself be automated by the system. The time at which colonies can be selected with respect to 2025 and 2030 depends on the information about the observed colonies at the time first observed in 2025. If the colonies evolve from what is considered a pure sample (such as a deep scratch or cut), the colonies can be selected immediately if there is sufficient quantity to observe and collect them. Time is crucial when processing such samples, and eliminating one or more additional culture cycles can mean providing test results hours earlier. If the sample is estimated to be impure, additional culture cycles are required so that the colonies can grow further to assist in the identification and quantification of these colonies. selecting colonies can be a partially automated process, in which multiple candidate colonies are automatically identified and presented visually to an operator in a digital image, whereby the operator can input a selection of one or more candidates for sampling and further testing. Sampling of the selected or picked colonies can itself be automated by the system. The time at which colonies can be selected with respect to 2025 and 2030 depends on the information about the observed colonies at the time first observed in 2025. If the colonies evolve from what is considered a pure sample (such as a deep scratch or cut), the colonies can be selected immediately if there is sufficient quantity to observe and collect them. Time is crucial when processing such samples, and eliminating one or more additional culture cycles can mean providing test results hours earlier. If the sample is estimated to be impure, additional culture cycles are required so that the colonies can grow further to assist in the identification and quantification of these colonies. selecting colonies can be a partially automated process, in which multiple candidate colonies are automatically identified and presented visually to an operator in a digital image, whereby the operator can input a selection of one or more candidates for sampling and further testing. Sampling of the selected or picked colonies can itself be automated by the system. The time at which colonies can be selected with respect to 2025 and 2030 depends on the information about the observed colonies at the time first observed in 2025. If the colonies evolve from what is considered a pure sample (such as a deep scratch or cut), the colonies can be selected immediately if there is sufficient quantity to observe and collect them. Time is crucial when processing such samples, and eliminating one or more additional culture cycles can mean providing test results hours earlier. If the sample is estimated to be impure, additional culture cycles are required so that the colonies can grow further to assist in the identification and quantification of these colonies. selecting colonies can be a partially automated process, in which multiple candidate colonies are automatically identified and presented visually to an operator in a digital image, whereby the operator can input a selection of one or more candidates for sampling and further testing. Sampling of the selected or picked colonies can itself be automated by the system. The time at which colonies can be selected with respect to 2025 and 2030 depends on the information about the observed colonies at the time first observed in 2025. If the colonies evolve from what is considered a pure sample (such as a deep scratch or cut), the colonies can be selected immediately if there is sufficient quantity to observe and collect them. Time is crucial when processing such samples, and eliminating one or more additional culture cycles can mean providing test results hours earlier. If the sample is estimated to be impure, additional culture cycles are required so that the colonies can grow further to assist in the identification and quantification of these colonies. selecting colonies can be a partially automated process, in which multiple candidate colonies are automatically identified and presented visually to an operator in a digital image, whereby the operator can input a selection of one or more candidates for sampling and further testing. Sampling of the selected or picked colonies can itself be automated by the system. The time at which colonies can be selected with respect to 2025 and 2030 depends on the information about the observed colonies at the time first observed in 2025. If the colonies evolve from what is considered a pure sample (such as a deep scratch or cut), the colonies can be selected immediately if there is sufficient quantity to observe and collect them. Time is crucial when processing such samples, and eliminating one or more additional culture cycles can mean providing test results hours earlier. If the sample is estimated to be impure, additional culture cycles are required so that the colonies can grow further to assist in the identification and quantification of these colonies. selecting colonies can be a partially automated process, in which multiple candidate colonies are automatically identified and presented visually to an operator in a digital image, whereby the operator can input a selection of one or more candidates for sampling and further testing. Sampling of the selected or picked colonies can itself be automated by the system. The time at which colonies can be selected with respect to 2025 and 2030 depends on the information about the observed colonies at the time first observed in 2025. If the colonies evolve from what is considered a pure sample (such as a deep scratch or cut), the colonies can be selected immediately if there is sufficient quantity to observe and collect them. Time is crucial when processing such samples, and eliminating one or more additional culture cycles can mean providing test results hours earlier. If the sample is estimated to be impure, additional culture cycles are required so that the colonies can grow further to assist in the identification and quantification of these colonies.

[0147] Referring back to FIG. 18, the use of an automated process enables faster AST and MA Enable the inspection of LDI. Such an inspection 1850 in an automated process can be started immediately after the initial evaluation 1830, and the results can be acquired (1 860) and reported (1875) by the 24-hour mark. In contrast, such an inspection 855 in a manual process often cannot be started until approaching the 36-hour mark, and it takes an additional 8 to 12 hours to be able to review (1865) and report

[0148] (1875) the data. In summary, the manual inspection process 1805 is shown to take up to 48 hours, requires an 18- to 24-hour incubation period, and only then can the plates be evaluated for growth, and there is no way to further track how long the samples have been incubated. In contrast, the automated inspection process 1800 can detect even with a relatively poor contrast between colonies (compared to the background and each other), and because the microbiologist can image and incubate without having to track the timing, the samples can be identified and prepared for further testing (e.g., AST, MALDI) before only requiring 12 to 18 hours of incubation, and the entire process can be completed within about 24

[0149] hours. Therefore, the automated process of the present disclosure, with the aid of the contrast processing described herein, results in a faster Without departing from the spirit and scope of the invention as defined by the claims of the present invention numerous changes may be made to the exemplary embodiments and other configurations may be devised It should be understood that this is possible.

Claims

1. 1. An automated method for assessing microbial growth in plate media, comprising: Providing a culture medium disposed within a substantially optically transparent container and inoculated with a biological sample. and, incubating the inoculated culture medium in an incubator; The substantially optically transparent container with the inoculated culture medium is placed in a digital imaging device. Steps and At the first time point (t ０ A first section having a plurality of pixels of the inoculated culture medium in obtaining a digital image of the the first digital to the substantially optically transparent container having the inoculated culture medium; determining coordinates of a plurality of pixels within the pixel image; The substantially optically transparent container with the inoculated medium is captured by the digital imaging device. The inoculated culture medium is then removed and placed in the incubator for further incubation. Top and After further incubation, the substantially optically transparent container with the inoculated culture medium is then placing the image sensor in a digital imaging device; At the second time point (t ｘ A second pixel having a plurality of pixels of the inoculated culture medium. obtaining a digital image of the The coordinates of a plurality of pixels in the second digital image are compared with the coordinates of a plurality of pixels in the first digital image. so that the coordinates of the corresponding pixels in the aligning the first digital image with the second digital image; A plurality of pixels in the second digital image are compared with a corresponding pixel in the first digital image. comparing with a plurality of pixels; Identifying pixels that have changed between the first digital image and the second digital image. determining whether a difference between the first digital image and the second digital image changes; The pixels that are not in the step indicate the background. Any of the identified pixels in the second digital image are pixels that represent background. determining whether a predetermined level of reference contrast is present relative to the cell; Identifying one or more objects in the second digital image, The object has the predetermined level of reference contrast relative to the pixels representing the background, having a plurality of pixels not separated from each other by a scene pixel; Associating the identified objects with biomass; Determining whether the biological sample is identified as a pure sample and identifying the biological sample as a pure sample. and further determining whether the biomass exceeds a first criterion when the The first criterion is a predetermined amount of the culture medium covered by the identified object. an area of ​​the identified object, and if the area of ​​the identified object exceeds the first criterion, selecting at least a portion of the biomass for analysis; If the biological sample is not a pure sample, the inoculation is performed in the substantially optically transparent container. Further culturing the culture medium. The method includes:

2. If no target is identified after a predetermined period of time, the method comprises: Claim 1: The substantially optically transparent container is flagged as a plate with no growth.

2. The method according to claim 1.

3. After the further incubation step, a third digital image is obtained having a plurality of pixels. R, The method comprises: Identifying additional pixels that have changed from the second digital image to the third digital image. determining a the identified pixel and the other pixel identified in the third digital image. Which of the pixels has a predetermined level of reference contrast relative to the pixels showing the background? determining whether the metric has a last Identifying one or more objects in the third digital image, The object has the predetermined level of reference contrast relative to the pixels representing the background, having a plurality of pixels not separated from each other by a scene pixel; Associating the identified objects with biomass; Determining whether the biomass exceeds a second standard, the second standard comprising: The criteria are: i) the amount of biomass that can be obtained by the target being a pure sample; and ii) if the object is due to an impure sample, a diameter of the biomass; If the biomass exceeds the second criterion, then a smaller amount of the biomass is selected for analysis. a step of selecting at least a portion of the second reference biomass, the second reference biomass being smaller than the first reference biomass; Bigger steps and The method of claim 1 further comprising:

4. The second data is generated based on a position of a reference mark on the optically transparent container.

2. The method of claim 1 further comprising the step of registering a first digital image with a second digital image. The method described.

5. The third data is generated based on a position of a reference mark on the optically transparent container.

4. The method of claim 3 further comprising the step of registering the first digital image with the second digital image. The method described.

6. The reference mark is an optically detectable, off-center mark on a bottom surface of the substantially optically transparent container; Dots and an end of an optically detectable label provided on a side of the substantially optically transparent container; a center of an optically detectable label on said substantially optically transparent container; The method according to claim 4 or 5, wherein the compound is selected from the group consisting of:

7. Obtaining a plurality of first digital images at the first time according to a predetermined set of lighting conditions. wherein each of the first digital images is obtained under a different lighting condition. Each illumination condition is determined by measuring the illumination source of the substantially optically transparent container having the inoculated culture medium. and a specified orientation of the substantially optically transparent container in the digital imaging device. and a specified underlying background color.

8. The digital imaging device includes: a substantially optically transparent container having an inoculated culture medium and a top surface of the substantially optically transparent container having an inoculated culture medium and a bottom surface of the inoculated culture medium; a provided illumination source; The inoculated medium is placed on the bottom surface of the substantially optically transparent container. a provided illumination source; oriented toward a side of the substantially optically transparent container having the inoculated medium. With a lighting source The method of claim 1 , comprising:

9. The specified orientation of the substantially optically transparent container is with respect to an illumination source above. and the specified orientation of the substantially optically transparent container is relative to a sideways illumination source. the specified background color is black, The specified orientation of the substantially optically transparent container is relative to an underlying illumination source. the specified background color is white, The method according to claim 8.

10. The illumination sources include an illumination source emitting red wavelengths, an illumination source emitting green wavelengths, and an illumination source emitting blue wavelengths. and an illumination source emitting

11. The method of claim 1 , wherein the object is a colony of microorganisms.

Images