Application development environment for biological sample evaluation processing

Abstract

To provide automation of analysis speed, analysis accuracy, and process.SOLUTION: A system and method for developing applications (Apps) for automated assessment and analysis of processed biological samples. Such samples are obtained, combined with nutrient media and incubated. The incubated samples are imaged and image information is classified according to predetermined criteria. The classified image information is then evaluated according to Apps derived from classified historical image information in a data base. The classified historical image information is compared with the classified image information to provide guidance on further processing of the biological sample, and Apps tailored to process provide sample process guidance tailored to the classifications assigned to the image information.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] [CROSS-REFERENCE TO RELATED APPLICATIONS] This application claims the benefit of the filing date of U.S. Provisional Patent Application No. 62 / 568,579, filed October 5, 2017, the disclosure of which is incorporated herein by reference. [Background technology]

[0002] Digital imaging of culture plates for the detection of microbial growth, colony counting and / or identification, etc., is receiving increasing attention. Systems and techniques for imaging plates for microbial detection are described in PCT Publication Nos. WO 2015 / 114121, WO 2016 / 172527, and WO 2016 / 172532, which are incorporated herein by reference in their entireties. Using such techniques (also referred to herein as the Kiestra system), laboratory personnel no longer need to read plates by direct visual inspection. Also, shifting laboratory workflow and decision-making to digital image analysis of culture plates can be more efficient.

[0003] While significant advances have been made in imaging technologies, there remains a need to extend the application of such imaging technologies to support automated workflows and / or automated diagnostic processes. In this regard, it is desirable to develop techniques that can automate the interpretation of culture plate images (e.g., growth identification, species identification, susceptibility testing, antibiotic sensitivity analysis, etc.) and determine next steps to perform based on the automated interpretation. However, given the wide variety of specimen types and biological taxa, developing automated image processing logic (e.g., software) for diagnostic indications can be time-consuming. Summary of the Invention

[0004] Disclosed herein is a system for evaluating biological specimens for the presence of pathogens, their identification, and other related analyses and evaluations. The objectives of the system are speed of analysis, accuracy of analysis, and automation of the process. Such systems typically acquire digital images of the incubated specimen, which is placed on a nutrient medium, and ascertain evidence of microbial growth, which provides an indication of the presence of pathogens in the biological specimen. Such systems, referred to herein as the Kiestra system, include equipment such as a camera and lighting for acquiring one or more images of the incubated specimen, a barcode reader for the specimen container (e.g., a Petri dish containing inoculated plating medium), and the like.

[0005] Such systems communicate with and are controlled by one or more customer-centric imaging applications (imaging-related apps or apps). Such imaging apps can be software that utilizes data derived from a collection of historical specimen images to analyze new specimen images for automated identification and / or diagnosis of disease states. The apps can provide clinical tools for rapid specimen characterization and results reporting. The apps link clinical specimens (from clinical sites) to non-patient-identifying facts about the specimen (e.g., non-patient-identifying specimen origin information such as demographic attributes), process conditions (e.g., incubation time and temperature), process materials or environment (e.g., nutrient media), and / or non-patient-identifying test results about the specimen (e.g., no growth of pathogens, growth or other positive identification of pathogens, enumeration of identified pathogens). These facts and conditions are collectively referred to herein as analytical information. Categorizing such analytical information in this manner allows apps to be developed using only the most relevant historical processing information. As such, each developed app is created for a very narrow purpose and is deployed only when the specimen classification elements correspond to the app classification elements. This data is useful when it is not tied to patient-identifying information (i.e., patient de-identification). In one embodiment, the system automatically de-identifies specimen information and information subject to certain conditions. De-identification involves providing metadata tied to non-patient-identifying classification information (e.g., geographic region where the specimen was obtained, specimen type, etc.), which in effect allows this data to be used in systems and methods that cannot retain confidential patient information. The system may include apps that provide image time series processing, classification / training and test diagnostic / evaluation algorithms, and / or expert systems. Apps may include modules. Here, a module may be understood to be one or more processes or algorithms for a specific purpose, for example, using image metadata (metadata serving as classifications as described herein) and rules. In some cases, multiple modules may be implemented as a single package.In some embodiments, the data may be stored in a database, which may be used for app development, app training, app certification, app testing, and the like.

[0006] The system can provide an image analysis process that can incorporate best practices and addresses relevant specimen types and automated picking.

[0007] To significantly reduce the time to market for valuable imaging apps, a multi-fold or modular system can be used. Such a multi-fold system may include modules such as: a) Define best-practice solutions tied to the specific media used to culture the target microorganism and the target microorganism's taxon, balancing usefulness with development timelines. For example, develop an application that uses a combination of test results and images tied to a specific media or taxon to evaluate new specimens classified with the same media and taxon. This application requires enough information to develop a reliable application, but not so much information that it delays development. b) Adjusting the algorithm development cadence (i.e., pace or speed) to maximize reuse of existing algorithms. c) Establish a clinical collaboration site that continuously generates images and associated metadata from diverse specimen types, biological taxa, and best practice culture media, i.e., build a database of information that can be used to obtain and train specimen information or further train developed applications. This database of fully classified historical specimen information is referred to herein as a data lake. d) Generate a database of images with defined criteria that can be applied to algorithm training and validation / clinical submission. The database may contain information / data representing clinical specimen images and classifications (patient demographics, imaging time and conditions, selection of media type, etc.) with concatenated manual / standard analysis of truth (quantification, identification, interpretation of results). The database infrastructure provides isolated data that can be individually accessed as appropriate for algorithm development, formal verification and validation (V&V), or clinical submission. The generation of this data allows for on-demand prioritization and development of apps for specific specimens or media types. e) Employ strategies for non-selective media and for identification and sister colonies that bucket classifications that reflect colony complexity (i.e., pure colonies, dominant colonies, or complex colonies). This data can be defined, for example, by media type (e.g., chromogenic media (i.e., CHROMagar)) for a given taxon (or a given taxon with a specific characteristic for a particular media type (i.e., hemolysis for BAP)), or other information about the conditions and reagents used to obtain the images and test results. f) Limiting certain Apps to only certain imaging systems / instruments (e.g., using a 25mp camera will result in certain system capabilities, while other imaging instruments will result in other system capabilities). Also, scoping an App for use in certain narrowly defined situations (e.g., specimen type, medium type, taxon type, camera type) and developing the App only using data corresponding to the defined situation will likely provide an App that is more useful and accurate for the specimens it will be used to evaluate.

[0008] Once the specimens, media, and organisms are validated and receive regulatory approval, the Apps described herein can be further developed and refined. For example, a quadrant-limited volume App can be implemented initially for throat swabs and wounds, and later implemented for perianal or other specimen types once more specimens have been processed. This continued training / development of the App results in a robust cadence of new imaging Apps of increasing usefulness. For example, as more images are evaluated, the App "learns" how to distinguish colonies from background in images of plate cultures. Processing image information to distinguish pixels associated with images of colonies from pixels associated with images of background is described in the Kiestra system referenced earlier in this specification.

[0009] Also described herein is a development system that delivers high-value software solutions while reducing time to market and maximizing resource utilization. A key component of this system can include initiating a collaboration program involving selected clinical sites running imaging systems (e.g., the Kiestra system) to collect clinical specimen imaging information categorized by association with metadata for development (algorithm training) and validation (submission preparation) purposes. This collection of data can provide for automated algorithm refinement, improved performance, and reduced development time. Validation uses isolated image collections in a database. These data can also be reused as additional features are implemented. However, this information is only deployed if the data classification corresponds to the app classification / purpose. This approach results in considerable efficiencies, flexibility within the app (i.e., versioning), and high-value resources.

[0010] For non-selective media, classification buckets are created to characterize population complexity and identify cultures as no growth, pure, dominant, or complex / mixed, rather than attempting to identify all sister colonies of all taxa on all media types. Identifying sister colonies is technically very challenging and time-consuming, especially on non-CHROMagar media. Mixed cultures typically require expert knowledge for interpretation and, even with some level of automated image analysis, require review for confirmation and release. By using the pure, dominant, and complex / mixed categories, most specimens can be characterized, and specimens with pure or dominant colony populations can be automatically worked up with high confidence. These colonies can then be selected by the app for automated picking by the picking system. This approach primarily replaces the need to individually characterize and define the specifications of each specimen type (e.g., saliva, wound, swab, etc.) by / for a specific pathogen with a more general classification strategy that efficiently utilizes imaging algorithms to characterize most specimen types.

[0011] Examples of imaging apps are summarized below. Such applications include: MRSA Screening Imaging App, Urine 2.0 Imaging App, Rapid detection imaging app, and Elephant Limited Edition Photography App, may include:

[0012] For example, in response to an MRSA (Methicillin-Resistant Staphylococcus Aureus) screening analysis, the following actions may be included (i.e., for a negative MRSA screening result): i) empirical treatment with certain antibiotic groups not used for resistant Staphylococcus aureus, ii) no patient isolation or special management, iii) proceeding with certain subsequent medical procedures such as surgery, iv) ordering certain further diagnostic tests, and v) guiding the selection of certain potentially effective antibiotic groups for antibiotic susceptibility testing interpretation. Applications developed according to the methods described herein may prompt a user for some or all of the above actions based on historical analysis of previous specimens that share certain predetermined criteria with the specimen being evaluated by the app.

[0013] Actions by the rapid detection imaging app may include: i) communicating the detection of a threshold level of growth indicative of infection to a treating physician hours earlier than standard techniques; ii) initiating a diagnostic to rapidly determine the identity of the growing pathogen (e.g., via MALDI-tof); iii) communicating the pathogen identification to a physician significantly earlier than standard techniques; and iv) initiating a diagnostic for the determination of antimicrobial susceptibility, as well as communicating an antibiotic susceptibility profile significantly earlier than current techniques. These objectives are achieved by using previous images and the results of the analysis of those images to inform the analysis of the current image. An app must be developed that is sufficiently sophisticated and reliable to be used to classify previous images and evaluate current images. Note that if the image analysis performed by the app results in a positive detection of microbial growth, in one embodiment, the app can communicate with the analyzer receiving the specimen during evaluation and identify the colony(s) on the specimen to be picked for further analysis. The app can direct or control downstream processes. Such downstream processing includes: i) preparing a suspension of one or more picked colonies in a predetermined buffer or solution; ii) adjusting the suspended cell suspension to a predetermined cell concentration; iii) spotting the cell suspension onto a substrate (e.g., a MALDI plate); iv) covering the spot with one or more reagents including, for example, a MALDI matrix solution, extraction chemicals; v) dispensing one or more aliquots of the cell suspension into wells of an antimicrobial susceptibility plate for determination of susceptibility profile to a series of antibiotics at various concentrations; and vi) dispensing the suspension for analysis by PCR, sequencing, or other molecular diagnostic test. In an alternative embodiment, the App directs, but does not control, subsequent sample processing / analysis.

[0014] Apps developed according to the methods described herein must be deployed to control processes and evaluate specimens using associated data and analytics. Methods and devices that can be used to obtain associated data and analytics include clinical sites that collect images and reference data used to build datasets for subsequent training and validation of imaging apps. See Figure 1. This typically includes collaboration and resources to identify, initiate, and manage data. Software tools collect data and metadata from clinical sites. Technical resources may be required at the laboratory to generate classifications and assign metadata not typically available in routine clinical workups, such as metadata from image processing.

[0015] A digital camera (a conventional camera with good megapixel resolution, e.g., 5MP, 25MP, etc.) can be implemented for adequate performance for initial growth / no growth and tentative identification (ID). For example, for colonies 5 mm or larger in diameter detected by an imaging module or device that captures digital images of an inoculated plate, this data can be combined with growth / no growth detection to indicate when growth occurs at a colony size sufficient for further processing. The imaging device itself can be an app that analyzes the digital image, such as that described in the Kiestra system mentioned elsewhere herein. According to these systems and methods, colonies are distinguished from the background, and from this image analysis, the density of colonies on the inoculated plate is determined and communicated to the app, which then determines further action in response to the image analysis. This information can then be used by the app to automatically determine which colonies to pick from the image (without an operator reading the plate or identifying the colonies to pick). The app is developed and deployed for a specific plate environment. For example, pure plates (one colony type) and dominant plates (two or more colony types, but mostly one colony type) may represent each colony type shown by the Purity Plate module, with associated rules determining further workup. Plates deemed complex are processed by a different App (or an App that fires a different rule). A predetermined number of colony types can each be designated for automated identification (ID) and antibiotic susceptibility testing (AST) workup in the ID / AST module, which may involve an automated picking system / robot.

[0016] To reduce product development costs and time to market, best practice support can be employed. Specific media (see below) can be used to achieve optimal recovery and performance for your platform (e.g., the BD Kiestra system).

[0017] This system can involve processing plate media readings performed in a laboratory. Plate media is prepared as described elsewhere herein. Media type and sample taxonomic group are examples of specimen classification elements that inform metadata associated with specimen information. Selected apps can provide pseudo-IDs with picking recommendations for pure or only the predominant colony type on the plate. Specimens with multiple clinically important isolates are rare and often complex. Specimens identified as complex can be classified into a mixed category for manual review and action.

[0018] Rather than being specific to an app configured for use with a particular type of specimen or target species suspected to be contained in that specimen, a more general set of algorithms can be designed. These algorithm tools are not described in detail herein but can be developed and deployed by those skilled in the art. The algorithms are validated as a process for deployment in an app designed to evaluate and process a range of specimen types. If such a tool is not limited to a particular specimen type, it can be used to review more plates or as a component of another app.

[0019] The system develops a set of tools that provide a full range of capabilities that can be later supplemented by acquiring more specimen processing data (e.g., imaging data) and adding it to a database with associated classifications. These tools evaluate plates and provide specific results independent of specimen type. A compilation of these results by specimen and patient demographic attributes is processed by the app to provide an indication of specimen volume. This simplifies implementation and accelerates time to market while providing a level of flexibility to users. The development of a more general toolbox of image analysis algorithms (along with classification metadata and rules called modules) that enable the app can be rapidly matured or optimized (e.g., for specific specimens) by deploying artificial intelligence algorithms—e.g., neural networks, artificial neural networks, or deep learning algorithms—that use the classified specimen data in the database to make decisions about specimens of interest. As part of an exponential strategy, artificial intelligence can be employed to "automatically" determine what image attributes are and what algorithms to use, best providing the desired module capabilities. Individual modules may be valuable enough to launch as apps, or multiple modules may be packaged into an app. In some versions, a set of modules can be differentiated to perform the following sample analyses based on the obtained sample information and its classification: (a) growth / no growth, (b) semi-quantitative, (c) provisional ID, (d) pure, dominant, complex, (e) antibiotic sensitivity (Kirby-Bauer test), and (f) sister colony location.

[0020] Examples of applications (Apps) may include Key ID, Rapid Detection, CHROMagar ID 2.0, Elephant Limitation, and Pseudo ID, which can be summarized as follows:

[0021] Key ID is a tool to identify a specific species on a specific plate where any number of colonies are present. Any number of colonies specific to the Key ID app are designated in pure or mixed cultures. Examples are: 1. MRSA screening on MRSA II, 2. Group A beta streptococci on BAP (blood agar plates) and 3. Streptococcus pneumoniae on BAP, is.

[0022] The Rapid Detection App provides a rapid detection process. Growth at any read point can be used as a flag for growth to be detected as early as possible on any plate. The user selects the read point that serves as the flag, recognizing the tradeoff: earlier read points may be less reliable but produce results more quickly, while later read points may be more reliable but require more time for detection. The action taken depends on the read point selected and the rule being invoked. Early detection can occur as early as four hours, although incubation periods of six, eight, or longer hours are contemplated. Incubation times for specific analytes are readily ascertainable by those skilled in the art. The systems and methods described herein are not limited to any particular incubation time. The Rapid Detection App is useful for rapid positivity of critical analytes.

[0023] For example, growth on a BAP plate is detected in 8 hours. A rule might take these results and, if the sample type is CSF (cerebrospinal fluid), alert the lab as soon as the app determines that the sample shows growth on the plate, since CSF is typically sterile. This determination requires the app to issue some kind of alert in response to determining that the CSF sample is positive for a pathogen. For other sample types, such as saliva, early detection is of little use because nearly all cultures contain normal flora. In these cases, no rules are written for samples classified this way, and no action by the app based on rapid detection is taken. Therefore, different automated processes can be triggered depending on the plate type and the type of detection.

[0024] Another embodiment is a CHROMagar ID App for quantifying urine or other specimen types. The app obtains classified image information, which allows the app to tentatively identify pure or predominant colony types on the CHROMagar orientation. Mixed plates are also identified, but the app may invoke different rules depending on whether the plate is complex. A rules engine can be used to separate processed specimen images into categories. Certain categories allow for automated reporting of results. These categories are described in detail elsewhere herein.

[0025] The Elephant Limited Quantity and Preliminary ID app is a tool that evaluates all positives and classifies each as (1) no growth, (2) pure, (3) predominant, or (4) complex, as well as assessing the overall quantity on the plate. Pure or predominant isolates are tentatively identified, and colonies are identified for picking. In this case, all cultures are analyzed and placed into categories for automated reporting or sent for review. Plates with significant pathogens can be sent to other systems (e.g., picking or testing) for picking without customer / clinician intervention. The table in Figure 3 lists the target organisms tentatively identified on the referenced plates. The table in Figure 4 lists commonly inoculated plates for a particular specimen (best practice).

[0026] An environment (e.g., Figure 1) may represent a software development workflow. This development effort focuses on completing the workflow and implementing the detection of the organism group in Figure 3 on the target medium in Figure 4. The flowchart in Figure 2 outlines a general strategy for algorithm and software development for one or more apps supporting the analytical workflow of digital image evaluation. Images of inoculated and incubated culture plates are all evaluated by the modules outlined in Figure 2 and described below. Each module evaluates a specific result or a discrete group of results and is largely independent of specimen type. However, metadata associated with the specimen (used to classify the specimen) can also direct further processing (e.g., incubation instructions, imaging instructions, etc.). Such metadata can be read from barcodes on specimen containers (i.e., plates or Petri dishes). Once results are available, expert system processing evaluates these results and recommends or takes appropriate action. For example, if an app indicates that a specimen should be evaluated by AST, the expert system provides guidance rules to the AST panel. This guidance typically takes into account regulatory or guidance positions (i.e., by FDA or CLSI) and limitations (i.e., restrictions) on the proven capabilities of the AST testing platform. The expert system is provided with a set of base rules. Even if the app itself is not an expert system, the app can itself invoke rules based on information it learns about images of interest for specimens of interest. These rules can be edited or additional rules developed by the user specific to their organization.

[0027] As plates are evaluated and multiple results are combined at the specimen level, another set of rules guides automated reporting, user, Laboratory Information System (LIS) or Laboratory Information Management System (LIMS) alerts, and sends important isolates to a worklist or picking system for ID / AST testing, etc.

[0028] Targeting organisms on culture media, independent of specimen type, allows for more efficient and rapid development of complete systems. For example, E. coli can be considered the same genus and species regardless of the specimen source. Regulatory approval of some Apps for some specimen types may be challenging due to the uncommon nature of positive results for some specimens (i.e., CSF and other sterile sites) and therefore the lack of sufficient historical specimen processing / imaging data to develop clinically reliable Apps. However, over a period of time, with the benefit of images from diverse sources (i.e., clinical trials, regulatory submissions) and the test results associated with these images, Apps can be developed even for rare specimens.

[0029] One embodiment is a method of processing a biological sample, the method comprising obtaining the biological sample, combining the biological sample with a nutrient medium, incubating the biological sample, obtaining a digital image of the incubated biological sample, classifying the digital image according to analytical criteria selected from the group including specimen origin information, clinical sample criteria, process materials, and process conditions, obtaining data from historical digital images of the biological sample incubated on the nutrient medium that share at least one of the analytical criteria assigned to the digital image, and outputting instructions to a user for further processing of the biological sample using the historical digital image data.

[0030] The specimen origin information includes geographic information about the biological specimen source and the biological specimen type. The process material includes the type of nutrient medium. The historical digital image data is classified by at least one of specimen type, biological taxon, or culture medium type. The method further includes analyzing the digital image data and determining from the analyzed data whether the digital image reflects microbial growth. In response to determining that the digital image does not exhibit microbial growth, the method outputs an indication of no microbial growth. In response to determining that there is an indication of microbial growth, the method performs the steps of determining whether the specimen is a sterile specimen and identifying one or more coordinates of a microbial colony in the image based on the indication of microbial growth. In response to determining that the specimen is sterile, the method further includes the steps of indicating the specimen is a high positive and sending instructions to further process the specimen. Examples of further processing include identification (ID) testing, antibiotic susceptibility testing, or both. Coordinates of objects in the image classified as colonies are communicated to a module, which forwards these coordinates to a picking device that picks the colonies from the biological specimen. The module may be an app or a combination of apps described herein. The module transfers the biological specimen to a picking device and picks colonies from the biological specimen, where the transfer and picking processes are controlled by the module or the module issues instructions to perform such processes. In response to determining that the specimen is not sterile, the module compares the historical image data with the image data to identify a specific predetermined species of microorganism in the digital image of the incubated biological specimen. The comparison process is performed by the module, and if the module determines that a specific predetermined species of microorganism is present in the image data, the module reports the identification of the specific predetermined species. Once such a determination is made, the module flags the specimen for further review.

[0031] The module compares the historical image data with the digital image of the incubated biological specimen to determine the amount of microbial growth. Here, the comparing and determining steps are performed in a module in communication with the imaging device that obtained the digital image of the incubated biological specimen. According to this method, the module or app determines whether the microbial growth is one of pure colonies, dominant colonies, or complex colonies by transmitting the digital image of the incubated biological specimen to a module that determines the growth level as a vector of three probabilities. In response to determining that the colonies are pure, the module reports that the plate is pure. If the growth on the pure plate exceeds a predetermined threshold growth, the biological specimen is identified as a high positive, and the module communicates this information to a user of the system. The module, receiving the colony coordinates from the imaging device, communicates the coordinates to a picking device that picks the colony from the biological specimen. This method may also include transferring the biological specimen to the picking device and picking the colony from the biological specimen. Here, the transferring and picking steps are controlled, requested, or required by the module. If the module determines that growth does not exceed a predetermined threshold, the module provides a tentative identification of the colony based on a comparison of the image of the colony provided to the module with historical image data accessed by the module. In response to this determination, the module performs the further step of reporting the tentative ID to the user.

[0032] If the module determines that growth does not exceed a predetermined threshold, the module performs the further step of reporting to the user that the complex sample does not meet or exceed a positive growth threshold. In response to determining that the colony is dominant, the module provides a tentative identification of the colony based on a comparison of the image of the colony provided to the module with historical image data accessed by the module. The module performs the further step of reporting the tentative ID to the user.

[0033] If the module determines that the growth exceeds a predetermined threshold growth, the module identifies the sample as a high positive. The method also includes alerting a user to the high positive. The method may also include identifying coordinates of the high positive. The method may also include communicating coordinates of the colony to a module that communicates the coordinates to a picking device that picks the colony from the biological sample. The method may also include controlling the module or issuing instructions to move the biological sample to the picking device and picking the colony from the biological sample, wherein the moving and picking steps are controlled by the module. The method may also include alerting a user that further review is required. If the module determines that the growth does not exceed the predetermined threshold, the module performs the further step of reporting a temporary ID to the user. In response to determining that the colony is complex, the method further includes reporting that the plate is complex from the module. The method may also include determining, by the module, whether the growth exceeds a predetermined threshold growth. If the module determines that the growth exceeds a predetermined threshold growth, the method further includes alerting a user that further review is required. [Brief explanation of the drawings]

[0034] [Figure 1] FIG. 1 is a block diagram illustrating an example system environment for developing a biometric image processing application. [Figure 2] FIG. 1 is an example flow diagram of an example App and its process implementation in a development environment according to aspects of the present disclosure. [Figure 3] 1 is a table showing examples of target organisms with tentative identification and associated media. [Figure 4] 1 is a table showing examples of target media with associated specimen types. [Figure 5]1 is a schematic diagram of a process that may be implemented in some versions of the present technology, such as in processing system 101 of FIG. 1, where the processing system can access data and / or algorithms from the data lake described herein to develop apps using training, validation, and / or clinical submission processes. These processes can also utilize information derived from the data lake (e.g., updated data and algorithms). DETAILED DESCRIPTION OF THE INVENTION

[0035] The present disclosure provides instruments and methods in an environment for developing imaging applications for identifying and analyzing biological specimens, such as microbial growth. Many of the methods described herein can be fully or partially automated, such that they are integrated as part of a fully or partially automated laboratory workflow.

[0036] This application provides a description of the design and implementation of a system that facilitates the delivery of automated imaging capabilities to bioimaging systems, such as the BD Kiestra™ system. The imaging capabilities of such systems are enabled by a set of hardware, software, analysis algorithms, and clinical rules. An example of one such commercially available system includes one or more digital cameras (e.g., 4MP) with multiple illumination configurations that generate optimized and standardized images based on an appropriate platform. The system described herein can be implemented with other optical systems for imaging microbiological samples. Many such commercially available systems exist and will not be described in detail herein. One example would be the BD Kiestra™ ReadA Compact intelligent incubation and imaging system. The Kiestra™ ReadA Compact is an automated incubator with a built-in camera and plate movement system that enables automated imaging of plates. The ReadA Compact is commercially available. The ReadA Compact also includes integrated plate import and export devices that interface the incubator with other operating equipment. Thus, in some embodiments, in response to analysis of the digital images by one or more Apps, the Apps can issue instructions and control incubation of associated analytes during evaluation by the Apps. Other exemplary systems include those described in PCT Publication No. WO 2015 / 114121 and U.S. Patent Application Publication No. 2015 / 0299639, which are incorporated herein by reference in their entireties. Such optical imaging platforms are well known to those skilled in the art and will not be described in detail herein.

[0037] A series of apps for the system can provide analysis of images from most specimens generated at various predetermined times in ReadA Compact, etc. The system can enable downstream actions on no-growth and / or negative plates, including automatic release of these plates, and automatic characterization of colonies for limited ID and AST analysis.

[0038] At a high level, imaging analysis tools (imaging apps) can be deployed to enable the collection of various results and / or actionable clinical results that provide considerable value to laboratories. These apps utilize one or more image analysis algorithms (modules) and a set of rules that inform how to apply the module to a particular medium type and / or particular specimen. Certain apps also have associated expert systems that overlay further sets of rules, typically regulatory / clinical guidance, on the more basic app decisions to inform recommendations for action and interpretation of results.

[0039] Developing an imaging app can utilize an iterative approach. A data lake is developed using either specimen information acquisition or images of clinical specimens processed as part of routine clinical laboratory practice, or both. Algorithms are developed to model conclusions / commands / outputs, as shown in Figure 2, which can be derived from truth and classification information associated with specimen images evaluated by the app. Verification and validation (V&V) of the app is performed using a predefined collection of images, along with certain metadata requirements from the data lake, marked for V&V analysis. The app is then validated and trained by subjecting the app to clinical studies, which include processing specimens to obtain images of those specimens and using the app to evaluate the specimen images. The use of a specific clinical site for the clinical study is not required.

[0040] However, the exponential strategy can be utilized with several components by implementing a biometric image processing application development system 100 as shown in Figure 1. The system 100 includes one or more of the following: a) A toolbox application collection for a processing system 101 or the like with one or more processors, with general-purpose algorithms and modules that enable apps and can be rapidly matured or optimized (e.g., for a type of specimen) by applying artificial intelligence algorithms such as neural networks, artificial neural networks, and other deep learning algorithms. The artificial intelligence implementation can "automatically" determine what the image attributes are and what algorithms to use to best provide the desired module capabilities. b) Critical to the development and deployment of the Apps described herein is a developed database 102 (denoted herein as a data lake) containing images of clinical specimens and image conditions with associated manual / standard analysis of "truth" (i.e., facts about the images, such as colony quantification, ID, result interpretation, etc.), and possibly patient demographic data (appropriately anonymized). These images retain classification information in the form of metadata, so that only relevant image data is used to develop a particular App. The data lake can be populated with the cooperation of clinical laboratories, such as imaging systems 104, which can optionally feed data into the database over a network.

[0041] Where the data lake is stored is primarily a design variable. The data lake can be stored locally or in the cloud. The data in the data lake can be partitioned. For example, the data in the data lake can be segmented according to how the data is accessed and / or used. In one embodiment, one segment of the data in the data lake can be for algorithm training, another segment can be used for data verification or validation, and yet another segment can be used for clinical submission.

[0042] The system's database can store and provide classification and even link data for these clinical images, which can be individually retrieved on demand as needed for algorithm development, formal verification and validation, or clinical submission. Apps can be established at an international standardization level, including laboratory protocols, media types, imaging times, quantitative scores, etc. Resources can be supplemented with the generation of images and reference data (specimens and / or spiked / contrived samples) for specimens / organisms / conditions rarely seen in clinical settings. This proactive strategy for data generation allows for near-on-demand prioritization and development of specific apps for specific specimens or media types. This system architecture allows for the integration of new modules into existing laboratory software to facilitate the release of new imaging diagnostic capabilities (apps / modules / packages). This can be achieved by standardizing interfaces between existing / legacy imaging software systems and new apps / modules / packages. In this regard, the App / module / package may be an add-on to system software, such as system software for an automated imaging system (e.g., in a processing system that controls any one or more of plate / sample conveyors, incubators, cameras, pickers, and / or associated robotics for moving such samples / plates within such automated laboratory cells / instruments, etc.). Certain specimens that test negative for pathogens can be spiked with known pathogens, and images of the incubated, pathogen-spiked specimens can then be characterized as described herein. Images of the pathogen-spiked specimens can then be used as a training set for an App that can be used to evaluate and treat low-frequency pathogens.

[0043] This approach allows for maximum flexibility in prioritization and app development cadence. Providing app performance metrics early also helps better understand and recognize app value and solution-level synergies. The data lake can be generated using one or more hospital systems that meet a list of Corporate Clinical Development (CCD) standards (i.e., technical and medical information). The implementation provides the ability to store and categorize data, queries, and audit the database; laboratory protocols that define programs and processes; and training, monitoring, compliance, and quality metrics, as is common in clinical trials and can occur as part of the data lake's development rather than at the end of the typical product development process. This approach can implement dedicated clinical laboratory resources to interpret certain plate results outside of standard protocols and link images with analytical results against the data lake. The data lake must be developed in a specialized process that can be independent of normal / routine laboratory operations, specifically run for certain specimens, plate types, or image acquisition time points. Thus, in some versions, the data lake database may include images of clinical specimens with concatenated manual / standard analysis of truth (e.g., quantification, ID, interpretation of results) and metadata related to classification for the images (e.g., patient demographics, time and conditions of imaging, selection of media type, etc.).

[0044] Establishing both the algorithm and the data lake may involve a certain level of standardization. This also defines what any given app will validate and the analysis / commands / outputs that can ultimately be obtained. Given the diversity of media types, media suppliers, and incubation times used in laboratories around the world, a "best practices" approach can be used to begin this endeavor. Additional conditions can be added later by stocking the data lake with appropriate data. An example media x specimen matrix is summarized in Figure 3. The table in Figure 3 includes 12 media types. Note that blood agar, trypticase soy broth (TSA), and Columbia are grouped together as one media type. XLD is xylose lysine deoxycholate agar, SS agar is Salmonella-Shigella agar, CNA is Columbia-nalidixic acid agar (CNA), and CLED is cystine-lactose-electrolyte-deficient agar. The listed media are known to those skilled in the art, as are the microorganisms known to be identifiable on the listed media. Therefore, standardization may include laboratory protocol, media type, imaging time, streak pattern, quantification score, etc., which is a classification applied to data for database development and image analysis with reference to database / data lake / historical image information. This focuses validation efforts, minimizes development time, and enables inter-laboratory metrics, data sharing, etc. The use of a data lake as training data, validation data, clinical submission data, etc. for developing apps is shown in Figure 5.

[0045] To ensure accuracy of the database, data addition to the database may involve independent human image analysis of images performed by several people or manual plate analysis by technicians. Human readings can be compared to clinical laboratory reports. In some cases, further image review may be involved when readings conflict. Database entry may involve de-identifying patient information from image-related data. Image review for data entry may involve human scoring of growth on plates for pure, dominant, complex, and no growth, i.e., quadratically limited amounts.

[0046] Example software modules for the toolbox In a typical example, the data lake may contain media plate images coupled with laboratory-determined quantitation (e.g., no growth, +, ++, +++). The data lake may also contain identification (ID) of organisms determined to be important to the specimen class or type (e.g., those deemed important by a trained clinical microbiologist). The data lake may also contain image-based metadata and, potentially, patient demographic information. To facilitate algorithm development, identification of organisms not typically identified as pathogens (i.e., normal flora) may also be required. Once added, some images are utilized to train and test appropriate algorithms for app development.

[0047] At a high level, deploying imaging analysis tools enables the collection of diverse and / or useful clinical results that add substantial value to laboratories. These modules (Mods) consist of one or more image analysis algorithms and a set of rules that inform how to apply the module with specific media types and specific specimens. Some modules, such as Screening for MRSA, can be implemented as apps, while others are often packaged with other modules to provide synergistic capabilities (e.g., quadrant limit and purity are often packaged as apps). Some apps may also have associated expert systems that overlay additional rule sets, typically overlaying regulatory / clinical guidance, that inform recommendations for action and interpretation of results (e.g., KB zones, as described herein). Based on technical and clinical considerations, one or more apps can be bundled as components of a launch package, depending on the needs of various clinical laboratories. Some apps are also expected to have versions (e.g., FDA-cleared UCA V1.8 becomes UCA 2.0).

[0048] Some examples of algorithms and modules (collections of synergistic algorithms) are general-purpose (generally work across analytes, pathogens, and media) and can be considered in relation to the process summarized below and shown in Figure 2. As mentioned previously, categorizing the functionality of various applications helps guide the development of detection applications for various species and media.

[0049] 1. Multiplication App / Module 1010A The Growth App 1010 (see FIG. 2) can be concerned with the answer to a simple question: Is there any growth detectable on this plate at this particular incubation time? The answer to this question is a probability of growth ranging from 0 to 1. The Growth App can be a module that targets any medium, independent of dispense volume or streaking pattern. Growth can be detected as early as possible from a predetermined imaging point. Rules dictate whether to issue an alert based on specimen type and / or medium. In some versions, Gram stain results can also be integrated into the app / module, if desired. In some cases, this may be implemented as an early detection or early growth app / module. Growth can be detected as early as possible from a predetermined imaging point (e.g., a detection window of 4 to 14 hours or more). Rules dictate whether to issue an alert based on specimen type and / or medium. In some versions, Gram stain results can also be integrated into this, if desired.

[0050] 2.Key ID App / Module 1020 The Key Id module 1020 (see FIG. 2) may aim to identify species that are likely to grow on a given medium. For any requested Key ID organism, this module may provide a list of colony locations (including probabilities) per Key ID in order of decreasing probability. These colonies may then be picked manually or by an automated picking system.

[0051] In some versions, the system may include pathogen-based screening and key pathogen modules (or modules). These modules can result in the detection of specific pathogens, groups of pathogens, or pathogens with specific properties. A screening app can be implemented to enable the identification of specific pathogens on CHROMagar, which can be used for patient management and pathogen characterization, e.g., MRSA, ESBL, CPE, VRE, etc. CHROMagar allows for a collection of pathogens to be preliminarily identified, i.e., CHROMagar orientation for Gram-negative (GN) and Gram-positive (GP) bacteria. Pathogen-specific media can be used for specimens. For example, SS medium can be used for Salmonella and Shigella in feces. Certain organisms can be preliminarily identified or flagged on more general media based on, for example, hemolytic properties on blood agar or unique morphological characteristics on specific media. A possible collection of modules, along with their pathogen and media capabilities, is summarized in the table in Figure 4.

[0052] 3. Elephant Limited App / Module 1030 If growth is detected based on a streaking pattern, e.g., a BD Kiestra™ InoqulA™ quadrant streaking pattern, the quadrant quantification module 1030 (see Figure 2) provides a growth level of mild, moderate, or severe. BD Kiestra™ InoqulA™ automates the processing of both liquid and non-liquid bacteriology specimens, facilitating a streamlined workflow, enabling standardized processes, and ensuring consistently high-quality streaking for inoculation of solid growth media. Growth levels are returned as a vector of three probabilities (mild, moderate, or severe) ranging from [0, 1] and summing to 1. For example, this module can evaluate all plates to determine whether there is no growth or various amounts of growth (e.g., +, ++, +++) and whether any growth is pure, dominant, or complex. A determination of no growth can optionally trigger an automated or batch release. In some cases, growth quantification (e.g., +, ++, +++) can be determined by three or more colonies in any particular quadrant.

[0053] With respect to growth type, images / plates can be characterized as pure, dominant, and complex. This can be based on the minimum number of isolated colonies of each type. For example, a dominant growth can have three or more (or just two) colony types present, with one type outnumbering the other type(s) (e.g., 10-fold). Complex can have three or more colony types present, with no dominant isolate or the isolates not identifiable within a provisional ID table, such as the example in Figure 3. Complex plates can be automatically flagged / subjected to manual interpretation. Pure and dominant plate types can be automatically subjected to further automated processing, such as to represent each colony type represented with rules that guide further workup (e.g., picking, ID, and / or AST).

[0054] For example, a Colony Forming Units (CFU) / mL quantification App / Module 1040 (see FIG. 2) can be implemented. Based on InoqulA™ streaking pattern #4 (monoplate) or #6 (biplate), this module can give growth levels of <1, 1-9, 10-99, 100-999, and >1000 CFU / medium on the plate. Growth levels can be returned as a vector of 5 probabilities ranging from [0, 1] and summing to 1. To get the equivalent units of CFU / ml (bucket), the dispense volume may need to be taken into account.

[0055] Generally, in some versions, the quantification module can determine whether growth is due to a single growing organism, a dominant organism, or a mixture of organisms. A pure organism can be considered an organism that accounts for 99% or more of the observable / imageable growth. A dominant organism can be an organism that accounts for (90%, 99%) of the observable / imageable growth. Purity levels range from [0, 1] and can be returned as a vector of probabilities (e.g., three probabilities: pure, dominant, complex) that sum to 1. In the case of pure or dominant growth, up to five colony locations for the dominant organism are given in decreasing probabilities.

[0056] An example of a response for a given quantification is as follows: 1) An image of a specimen on a plate medium is judged to have a mixed flora above a predetermined threshold (100,000 CFU / mL). The App's response to this judgment is to flag the plate as complex because it has a mixed flora above the threshold amount and recommend a manual review of the plate. Because this judgment is not made with respect to the medium or taxon, this is an App of broad applicability and is not limited to deployment with respect to a specific medium or taxon on the plate. 2) The images are evaluated and determined to show no growth at 24 hours. If the sample is classified as a critical sample, the app issues a preliminary report of no growth and either recommends or controls reincubation of the plate for an additional 24 hours. If subsequent images do not detect growth at 48 hours, the app sends the user a final report of no growth after 48 hours. The app either recommends or controls plate discard. 3) The image of the specimen on the plate medium is determined to have pure growth above a predetermined threshold (100,000 CFU / mL) and a colony size above 0.5 mm. The App's response is to order or control the picking of colonies for ID and AST testing. The App flags the specimen for review by a technician and sends a report of the specimen exceeding 100,000 CFU / mL to the physician associated with the specimen. 4) The image of the specimen is determined by the App to indicate the presence of MRSA. The App sends a report that MRSA was detected and adds the specimen to a positive MRSA worklist. If the App determines that the MRSA colony size is above a threshold (e.g., greater than 0.5 mm), the App issues a command or controls the sending of the specimen for ID and AST testing. As described elsewhere herein, ID and AST involve specimen workup and evaluation, respectively. Therefore, ID and AST systems and devices are typically downstream of the incubation / imaging device (e.g., Kiestra™ ReadA Compact). 5) If the image of the specimen classified as ESBL is determined to show no growth, the App will issue a final report stating that no ESBL isolates were detected and will order or control the disposal of the plate. 6) The image of the specimen classified as saliva is identified as having a mixed flora (and thus a complex plate) above a threshold amount. If the app determines the plate is complex, it instructs a technician to review the plate. Note that different thresholds for mixed flora that trigger a request for manual review can be developed depending on the specimen classification. 7) An image of a specimen classified as critical on blood agar is determined to show growth. In this case, the Critical Specimens App is triggered, alerting the physician associated with the specimen and adding the specimen to the Critical Specimens Worklist. If the App determines that the colonies are greater than a threshold size (i.e., greater than 0.5 mm) and classifies the specimen as being on MacConkey agar, the App sends the specimen to automated picking for MALDI and Gram-negative AST. The App also sends a report indicating that a Gram-negative specimen has been isolated. 8) The image of the specimen shows a colony count greater than 100,000 CFU / mL and classifies the image as pure and having a colony size greater than a predetermined threshold (e.g., greater than 0.5 mm). In response, the App automatically picks the specimen for AST, adds the specimen to a list of positive technician reviews, and sends a report to the physician associated with the specimen indicating that greater than 100,000 CFU / mL colonies were detected in the specimen. Additionally, if the AST results indicate that the picked colonies are carbapenem-resistant, the App performs molecular confirmation testing. 9) The image of the specimen is determined to have a severe dominant growth. In response, the App automatically picks the growth and prepares a suspension to perform MALDI on the picked sample. If the colony is identified as E. coli by MALDI, the App further evaluates the sample for Gram-negative AST (using either the MALDI suspension or a new pick of the colony).

[0057] In some versions, growth detection can be implemented as two modules: one module evaluates the plate to determine growth / no growth, and the other module evaluates the amount of growth (+, ++, +++) for three or more colonies in any particular quadrant. For example, the first module evaluates images of a critical, normally sterile specimen. If this evaluation indicates growth at a predetermined time point and the colony size is determined to be greater than a predetermined threshold size, the app identifies the coordinates of a representative colony and issues a command to an imaging device (e.g., ReadA) to move the plate to an instrument that automatically picks the identified colonies. The picked colonies are resuspended in solution to a predetermined density for further testing, for example, in a molecular diagnostic instrument or test (e.g., PCR, sequencing).

[0058] 4. Temporary ID App / Module 1050 In the case of pure or predominant growth, the Pseudo-ID module 1050 (see FIG. 2) can identify the predominant organism likely to grow on a given medium using a set of identification algorithms based on training with the data lake. This module can provide / output the name of the highest ranked (e.g., probability) organism (or group of organisms) and up to five colony locations ranked by highest to lowest probability for this identification. These algorithms allow for the identification of specific species on specific medium types, where any number of colonies are present and deemed clinically significant.

[0059] For example, the media and colony identities may be those shown in the table in Figure 3. Rules may be included in the module to perform workup of these colonies. Specific examples of ID Apps, such as the Urine Culture App (UCA) and Chrom ID App, may allow for media-required orientation from urine for these organisms and preliminary ID on CHROMagar. Rules may apply the ability to auto-report / auto-release (or batch) and downstream workup (e.g., auto-picking, testing, etc.). In some cases, rules may flag plates as high positives for quick review and further processing via a worklist or auto-picking, etc.

[0060] 4.1 Purity Plate App / Module In some versions, the system can implement a purity plate module. Pure, dominant, and complex plates may each require a minimum number of isolated colonies. Dominant growths typically have three or more colony types, with one colony type more than 10 times more numerous than the others. Complex types typically have three or more colony types, with no dominant isolate or an unidentifiable isolate (see Provisional ID Table below). Complex plates are typically interpreted manually. The module can then classify plate images according to whether the plate is pure, dominant (though it may be slightly contaminated), and / or complex.

[0061] 4.2 Auto-select ID / AST module Pure and dominant plates can represent each colony type represented by the purity plate Mod, with associated rules guiding further workup as shown in the example above. A predetermined number of each colony type can be designated for automated identification (ID) and AST workup in the ID / AST module, which may include an automated picking system / robot.

[0062] 5. Kirby-Bauer (KB) Zone of Inhibition Diameter Measurement App Module Some embodiments can use any measurement app. Such an app can provide zone of inhibition measurements by utilizing existing imaging capabilities and AST expert systems. Optionally, these measurements can be coupled with the expert system to provide interpretation. This version of the app also provides the opportunity for early zone of inhibition measurements for specific drug / organism combinations and zone of inhibition on media plated directly from positive blood cultures. Such algorithms can be based on metadata and / or images from the data lake.

[0063] In some versions, the app leverages existing imaging capabilities and AST expert systems to provide zone of inhibition measurements and Abx disk identification. Optionally, these measurements can be coupled with an expert system to provide guidance on antibiotic susceptibility profiles for pathogens isolated from patients and guidance on treatment / response. In some versions, implementations of this app can provide early zone of inhibition measurements for specific drug / organism combinations and zone of inhibition on media plated directly from positive blood cultures. Some apps may include expert systems (interpretations) and can be greatly facilitated with the required metadata and images, along with a data lake that is appropriately stocked, monitored, and audited.

[0064] System 100 may include any one or more of the imaging-related modules / apps described above, although in some versions, a particular segmentation of module functionality may be implemented by a discrete collection of the following modules / apps: (a) Elephant Limitation, (b) No Growth Detection, (c) Purity Plates: # Colony Types (e.g., Pure, Predominant, Complex), (d) Screening (e.g., CHROMagar, i.e., MRSA), (e) Important Pathogens, (f) Early Growth Detection, (g) Automated Selection ID / AST, and (h) Zone of Inhibition Measurement Kirby-Bauer.

[0065] Imaging App and Launch Package By providing a combination of algorithms / modules / rules, a data lake, and the ability to extract predetermined subsets of data, system 100 may enable rapid algorithm maturation and on-demand app development. As an example, apps can be developed based on specimen type and implemented by a collection of apps. The strategy for implementing a collection of apps is influenced by many factors: software launch cadence support; the value of individual apps versus integrated apps; the availability of certain algorithms or specimen / plate / organism types in the data lake; etc.

[0066] An example can be considered in relation to the table below:

[0067] [Table 1]

[0068] In this specimen-based example, a series of five different modules pertains to one specimen type. Apps validated for a specific specimen type are one way to package functionality, but an exponential approach allows for other options. For example, surveillance apps can be launched with specific target organisms (MRSA, Streptococcus, Shigella), quantitative apps can be packaged by media type (sample volume on blood agar independent of specimen), etc. However, in this sense, certain apps will have minimal value for certain specimens (i.e., sample volume on non-selective media containing saliva, given high normal flora levels). Note that apps can be defined by specific rules and regions with specific functionality that are limited to the geographic region from which the specimen under evaluation was obtained. Table 1 identifies functions specific to clinical requirements, particularly for the United States (US) and Europe (EU).

[0069] Thus, potential apps can be segmented into two high-level buckets. The first bucket collection can be considered screening and key identification apps. Such apps typically target specific organisms on CHOMagar or high-value pathogens on, for example, blood agar. Each of these is discrete and can be prioritized for development and launched individually as desired, with minimal impact on other apps or specimen types. Similarly, Kirby-Bauer Zone of Inhibition apps are generally independent of next-generation apps ("Next Gen Apps"), which have associated algorithms that can be prioritized independently. Furthermore, as new CHROMagar apps (i.e., vancomycin-resistant Enterococcus (VRE)) become available, they can be added to this list by adding them to the data lake using appropriate specimens. If the target isolate is relatively infrequent, the data lake can be supplemented with artificial (spiked) samples. Further examples of screening and key ID apps are shown in the table below.

[0070] [Table 2] TIFF0007720365000003.tif233169 TIFF0007720365000004.tif168170

[0071] From Table 2, it can be observed that the App is highly specific and the output can depend on the specimen classification (i.e., specimen type, US region or European region, etc.). Table 2 also shows, at a high level, the type of data used to train the App.

[0072] A second bucket collection of apps can be prioritized and grouped for launch by several criteria using a more general algorithm. An example of these apps is outlined in Table 3 below. Essentially, each cell in Table 3 represents an app. Cells in the table below that share the same value are capabilities appropriate for that specimen type, and the values in the shared cells are reasonably packaged together in a common module. In this model, there are eight additional launch packages / modules.

[0073] [Table 3]

[0074] An example imaging module can be considered with reference to the table below.

[0075] [Table 4]

[0076] EUCAST is the European Committee on Antimicrobial Susceptibility Testing. In one example of a process integrated with one or more apps for sample evaluation and process control, a BD Kiestra™ InoqulA is used to inoculate samples onto plate media. The samples are streaked onto the media using a predetermined pattern tracked as part of the metadata via a barcode. The streaked sample is transported to a BD Kiestra™ ReadA Compact, where the sample is incubated and imaged at a time determined by the app. The images obtained by the ReadA at the specified time are analyzed by the app to determine whether the sample on the plate is pure. Based on the determination, further workup of the sample is performed. The images are evaluated to identify the coordinates of selected colonies. The app can send these coordinates to an instrument (or a technician). The app can issue a command to send the sample to an instrument that picks the colonies. The app can further coordinate or control the picking of the colonies and the transport of the picked colonies to another platform that performs pathogen identification. In one example, identification is performed by MALDI. As described elsewhere herein, samples are evaluated by MALDI by placing the picked sample in a suspension and inoculating the suspension onto a MALDI plate. The app can also coordinate or control the transfer of the colony suspension to the BD Kiestra™ InoqulA. Here, using a "spread pattern," the suspension can be inoculated onto another type of culture medium (e.g., Mueller Hinton) and then transferred to the AST testing instrument, where a predetermined antibiotic disc (e.g., BD BBL™ Sensi-Discs™) is placed on the culture. The plate with the inoculated specimen and antibiotic disc is then sent to the ReadA Compact under the coordination and control of the app. The ReadA acquires images and provides them to the app, which then sends the results to an expert system for analysis of the resulting antibiotic disc inhibition zones and interpretation of the results.The expert system then sends the results of the analysis to the clinical laboratory staff.

[0077] Although the invention herein has been described with reference to particular embodiments, it is to be understood that these embodiments are merely illustrative of the principles and applications of the present invention. It is therefore to be understood that numerous modifications can be made to the exemplary embodiments and other arrangements can be devised without departing from the spirit and scope of the invention as defined by the appended claims.

Claims

1. 1. A method for processing a biological sample, comprising: Obtaining a biological sample; combining the biological sample with a nutrient medium; incubating the biological sample; obtaining a digital image of the incubated biological sample; applying analytical criteria to classify the digital image as specimen origin information, clinical sample criteria, process materials, process conditions, or a combination thereof; obtaining data from historical digital images of biological specimens previously incubated on a nutrient medium that share at least one of the analysis criteria assigned to the digital images; outputting instructions to a user for further processing of the biological specimen based on the historical digital image data; and A method comprising:

2. The method of claim 1 , wherein the specimen origin information includes geographic information about the source of the biological sample and the type of the biological sample.

3. The method of claim 1 , wherein the process material comprises a type of nutrient medium.

4. The method of claim 1 , wherein the historical digital image data is categorized by at least one of specimen type, biological taxon, or culture medium type.

5. The method of claim 1 , further comprising analyzing the digital image data and determining from the analyzed data whether the digital image reflects microbial growth.

6. 6. The method of claim 5, wherein, in response to determining that the digital image does not exhibit microbial growth, an indication of no microbial growth is output.

7. 6. The method of claim 5, further comprising determining whether the specimen is a sterile specimen in response to determining that there is an indication of microbial growth, and identifying one or more coordinates of a microbial colony in the image based on the indication of microbial growth.

8. 8. The method of claim 7, wherein, in response to determining that the specimen is sterile, the method indicates that the specimen is high positive and sends instructions to further process the specimen.

9. 9. The method of claim 8, wherein the further processing is selected from the group consisting of an identity test, an antibiotic susceptibility test, or both.

10. 10. The method of claim 9, further comprising communicating the coordinates of the colony to a module that communicates the coordinates to a picking device that picks the colony from the biological sample.

11. 11. The method of claim 10, further comprising transferring the biological sample to the picking device and picking the colony from the biological sample, wherein the transferring and picking steps are controlled by the module.

12. 11. The method of claim 10, wherein, in response to determining that the specimen is not sterile, the historical image data is compared with the image data to identify a particular predetermined species of microorganism in the digital image of the incubated biological specimen.

13. 13. The method of claim 12, wherein the comparison is performed by the module, and if the module determines that a particular predetermined species of the microorganism is present in the image data, the module reports an identification of the particular predetermined species.

14. 14. The method of claim 13, further comprising flagging the specimen for further review.

15. 8. The method of claim 7, further comprising comparing the historical image data with a digital image of the incubated biological specimen and determining the amount of microbial growth, wherein the comparing and determining steps are performed in a module in communication with an imaging device that obtained the digital image of the incubated biological specimen.

16. 8. The method of claim 7, further comprising determining that the microbial growth is one of a pure colony, a dominant colony, or a complex colony by transmitting a digital image of the incubated biological sample to a module that determines the growth level as a vector of three probabilities.

17. Upon determining that the colony is pure, reporting from said module that the plate is pure; determining, with the module, whether the growth exceeds a predetermined threshold growth; 17. The method of claim 16, further comprising:

18. If the module determines that the growth exceeds a predetermined threshold growth, identifying the sample as a high positive; alerting the user to said high positive value; identifying the coordinates of the high positives; transmitting the coordinates of the colony to a module that transmits the coordinates to a picking device that picks the colony from the biological sample; transferring the biological sample to the picking device and picking the colony from the biological sample; In addition, the moving process and picking process are controlled by the module; 20. The method of claim 17, further comprising:

19. 20. The method of claim 18, wherein if the module determines that the growth does not exceed the predetermined threshold, the module provides a tentative identification of the colony based on a comparison of the image of the colony provided to the module with the historical image data accessed by the module, and the module performs the further step of reporting the tentative identification to a user.

20. 17. The method of claim 16, wherein, in response to determining that the colony is dominant, the module provides a tentative identification of the colony based on a comparison of an image of the colony provided to the module with the historical image data accessed by the module, and the module performs the further step of reporting the tentative identification to a user.

21. 21. The method of claim 20, further comprising determining, by the module, whether the growth exceeds a predetermined threshold growth.

22. if the module determines that the proliferation exceeds the predetermined threshold proliferation, identifying the sample as a high positive; alerting the user to said high positive value; identifying the coordinates of the high positives; transmitting the coordinates of the colony to a module that transmits the coordinates to a picking device that picks the colony from the biological sample; transferring the biological sample to the picking device and picking the colony from the biological sample; In addition, the moving process and picking process are controlled by the module; Alerting users that further review is required; and 22. The method of claim 21 further comprising:

23. 22. The method of claim 21, wherein if the module determines that the growth does not exceed the predetermined threshold, the module performs the further step of reporting the temporary ID to a user.

24. In response to determining that the colony is complex, reporting from the module that the plate is complicated; determining, with the module, whether the growth exceeds a predetermined threshold growth; 20. The method of claim 17, further comprising:

25. 25. The method of claim 24, further comprising alerting a user that further review is required if the module determines that the growth exceeds the predetermined threshold growth.

26. 20. The method of claim 18, wherein if the module determines that the growth does not exceed the predetermined threshold, the module performs the further step of reporting to a user that the complex sample does not meet or exceed the predetermined threshold growth.

Images