Apparatus and method for identifying tube assembly type

Abstract

A method of identifying a tube type. The method includes capturing one or more pixelated images of a cap attached to a tube; identifying a color of one or more pixels of the pixelated image of the cap; identifying one or more gradients of a dimension of the cap; and identifying the tube type based at least on the color of the one or more pixels and the one or more gradients of the dimension of the cap. Apparatuses suitable for implementing the method and other aspects are disclosed.

Description

[0001] Cross-references to related applications

[0002] This application claims the benefit of U.S. Provisional Patent Application No. 62 / 929,070, filed October 31, 2019, entitled “APPARATUS AND METHODS OF IDENTIFYING TUBE ASSEMBLY TYPE”, the disclosure of which is incorporated herein by reference in its entirety for all purposes. Technical Field

[0003] Embodiments of this disclosure relate to apparatus and methods for assembling identification tubes. Background Technology

[0004] Automated testing systems can use one or more reagents to perform clinical chemistry or laboratory tests to identify analytes or other components in biological samples (samples), such as serum, plasma, urine, interstitial fluid, cerebrospinal fluid, etc. For convenience and safety reasons, these samples are almost always contained in sample tubes (e.g., blood collection tubes). Sample tubes may be capped, and in some cases, the cap may include a color and / or shape that provides information about the type of test to be performed, the type of additives contained in the tube (e.g., serum separators, coagulants such as thrombin, or anticoagulants and their specific types, such as EDTA or sodium citrate, or antiglycation additives), and whether the tube is provided with vacuum capability, etc.

[0005] In some automated testing systems, sample containers and samples are digitally imaged and processed, such as using computer-aided digital imaging systems, making it possible to determine the type and color of the cap. During imaging, one or more images of the sample tube (including the cap) and the sample can be captured.

[0006] However, under certain conditions, such automated testing systems may exhibit performance variations and may inappropriately characterize tube and / or cap types. Therefore, there is a need for improved methods and equipment for digital imaging and processing of sample containers and caps. Summary of the Invention

[0007] In a first embodiment, a method for identifying a tube type is provided. The method includes capturing one or more images of a cap attached to a tube, the capture generating a pixelated image of the cap, the pixelated image comprising a plurality of pixels; identifying the color of one or more pixels of the pixelated image of the cap; identifying one or more gradients of a dimension of the cap; and identifying the tube type based at least on the color of one or more pixels and the one or more gradients of the cap dimension.

[0008] In a second embodiment, a method for identifying tube types is provided. The method includes capturing one or more images of a cap, the capture generating a pixelated image of the cap; identifying the color of the cap; identifying the dimensional gradient of the cap; identifying the tube type based at least on the color and dimensional gradient of the cap; and identifying a match between an arrangement of tests and the tube type.

[0009] In a third embodiment, a diagnostic device is provided. The diagnostic device includes an imaging apparatus configured to capture one or more images of a cap assembly including a cap attached to a tube, wherein the one or more images include pixelated images of the cap; and a controller communicatively coupled to the imaging apparatus, the controller including a processor coupled to a memory storing executable program instructions that are executable to perform: determining the color of one or more pixels of the pixelated image of the cap; determining one or more gradients of the cap's dimensions; and identifying a tube type based on at least the following:

[0010] The color of one or more pixels, and

[0011] One or more gradients of the dimensions of the top cover. Attached Figure Description

[0012] The accompanying drawings described below are for illustrative purposes and are not necessarily drawn to scale. Therefore, the drawings and description are intended to be illustrative in nature and not restrictive. The drawings are not intended to limit the scope of this disclosure in any way. The same numerals are used throughout the text to denote the same or similar elements;

[0013] Figure 1A The illustration shows a side view of a first tube assembly according to one or more embodiments of the present disclosure, including a top cap attached to the tube and an indicator. Figure 2A The box indicating the location of the mask image shown;

[0014] Figure 1B The illustration shows a side view of a second tube assembly according to one or more embodiments of the present disclosure, including a top cap attached to the tube and an indicator. Figure 2B The box indicating the location of the mask image shown;

[0015] Figure 2A This is a schematic diagram of a mask image of a first top cover of a first tube assembly according to one or more embodiments of the present disclosure, the first top cover of the first tube assembly including an indicative mask image and Figure 1A The box indicating the position of the first tube assembly;

[0016] Figure 2B This is a schematic diagram of a mask image of a second top cover of a second tube assembly according to one or more embodiments of the present disclosure, the second top cover of the second tube assembly including an indication mask image and Figure 1B The frame indicating the position of the second tube assembly;

[0017] Figure 3A The illustration shows a side view of a portion of a first tube assembly that can be analyzed to determine a multi-row gradient according to one or more embodiments of the present disclosure;

[0018] Figure 3B The illustrations depict one or more embodiments according to the present disclosure. Figure 3A The width curve of the upper part of the first tube assembly;

[0019] Figure 3C The illustrations depict one or more embodiments according to the present disclosure. Figure 3B The first derivative of the graph;

[0020] Figure 4A The illustration shows a side view of a portion of a second tube assembly that can be analyzed to determine a multi-row gradient according to one or more embodiments of the present disclosure;

[0021] Figure 4B One or more embodiments according to this disclosure are illustrated graphically. Figure 4A The width curve of the upper part of the second tube assembly;

[0022] Figure 4C One or more embodiments according to this disclosure are illustrated graphically. Figure 4B The first derivative of the graph;

[0023] Figure 5A A bar graph illustrating the spectrum (wavelength relative to average intensity) through one or more embodiments of the first tube assembly according to the present disclosure is provided.

[0024] Figure 5B A bar graph illustrating the spectrum (wavelength relative to average intensity) through one or more embodiments of the second tube assembly according to this disclosure is provided.

[0025] Figure 5C The illustrations depict the production process according to one or more embodiments of the present disclosure. Figure 5A A photographic image of a portion of the first tube assembly for backlighting;

[0026] Figure 5D The illustrations depict the production process according to one or more embodiments of the present disclosure. Figure 5B A photographic image of a portion of the second tube assembly for rear-facing illumination;

[0027] Figure 6 A table illustrating examples of different tube assemblies with different cap colors and cap shapes but used for the same test, according to one or more embodiments of the present disclosure;

[0028] Figure 7 The illustration shows an LDA (Linear Discriminant Analysis) plot in the HSV color space according to one or more embodiments of the present disclosure, which attempts to separate tube types based solely on color (for a set of tube types across different manufacturers);

[0029] Figure 8 The illustration shows a schematic side view of a diagnostic analyzer according to one or more embodiments of the present disclosure, the diagnostic analyzer being adapted to image tube assemblies and distinguish cap and tube types;

[0030] Figure 9 This is a flowchart of a method for identifying tube types according to one or more embodiments of the present disclosure;

[0031] Figure 10 This is a flowchart of another method for identifying tube types according to one or more embodiments of the present disclosure. Detailed Implementation

[0032] Diagnostic laboratories can use blood collection tubes (e.g., tube assemblies) from different manufacturers to perform multiple tests. Tube assemblies may include tubes with attached caps. Different tube assembly types may have different properties, such as different sizes and / or different chemical additives. For example, many tube assembly types are chemically active, meaning the tubes contain one or more added chemicals that can be used to alter the state of the sample or otherwise assist in its processing, such as by promoting separation or maintaining properties (e.g., anticoagulants, gel separators, or cell-free preservatives).

[0033] For example, in some embodiments, the inner wall of the tube may be coated with one or more additives, or additives may be provided elsewhere in the tube. For example, the types of additives included in the tube may be serum separating agents (e.g., separating gels), coagulants (such as thrombin), anticoagulants (like sodium heparin, sodium lithium, EDTA, potassium EDTA, K2EDTA, K3EDTA, or sodium citrate 1:9 or 1:4, acidic citrate glucose (ACD), sodium polyethanolsulfonate (SPS), etc.), anti-glycolysis additives (such as combinations of sodium fluoride and potassium sodium oxalate), or other additives (such as cell-free preservatives) used to alter, inhibit, or maintain the properties of the sample or to facilitate its processing. Tube manufacturers may associate the color of the cap of the tube assembly with specific types of chemical additives in the tube. Some colors may indicate the presence of a serum separating gel in combination with another additive (such as an anticoagulant or a clot activator).

[0034] Furthermore, different manufacturers may have their own standards that associate characteristics of the tube assembly (such as cap color and cap shape) with specific properties of the tube assembly, which may be related to the contents of the tube or whether the tube is provided with vacuum capability. In some embodiments, manufacturers may include labels indicating the contents of the tube, but these labels may be obscured in some cases. In some embodiments, diagnostic devices may read the labels.

[0035] For example, a manufacturer can associate all tube assemblies with light green caps with tubes containing lithium heparin, configured to test for substances such as glucose, electrolytes such as sodium and potassium, cholesterol, and, for example, some enzymes and markers. Lavender caps can identify tubes containing EDTA and its derivatives (anticoagulants), configured to detect hematological parameters such as white blood cells, platelet counts, and hemoglobin. Other cap colors, such as yellow, gold, light blue, white, gray, orange, red, and black, can be used and can have various meanings. In some embodiments, manufacturers can use different colored caps for the same tube type. The laboratory uses this color information for further tube processing. Furthermore, since some tubes may contain chemically active agents (often lined with substances like coagulants, anticoagulants, or antiglycation compounds), it becomes important to associate which tests can be run on which tube types, as tests are almost always content-specific. Confusion about tube types can lead to undefined or incorrect results, such as a tube for urine being used to process (e.g., test) different bodily fluids, or a tube for coagulation, for example, being used for tests that require anticoagulants.

[0036] Figure 6 A table illustrating examples of different tube assemblies with different cap colors and shapes but used for the same test is provided. Cap colors and shapes for three different tube types and three different cap standards are shown. Figure 6 As shown, for all three cap standards, the first tube type can have the same cap shape (e.g., the first cap shape), but the three cap standards have different cap colors. For example, the first cap standard could be transparent with tan stripes, the second cap standard could be dark red, and the third cap standard could be gray. Figure 6 In one embodiment, the three cap standards of the second tube type can use the same cap shape, a first cap shape, and the same color, orange with brown stripes. The third tube type can use different cap shapes and colors for the three cap standards. For example, the first cap standard can use a first shape and can be green with red stripes, the second cap standard can use a second cap shape and can be red, and the third cap standard can use a second cap shape and can be green.

[0037] Figure 7 The illustration shows an LDA (Linear Discriminant Analysis) plot in the HSV color space, which attempts to distinguish tube types as much as possible based solely on color (for a set of tube types across different manufacturers). Figure 7 The plot points illustrate different cap colors, such as cap colors for different tube types and cap colors from different manufacturers. As shown, not all tube types can be easily distinguished in the HSV color space. For example, in... Figure 6 In the second cap standard, the first tube type has a dark red cap, and the third tube type has a red cap. For the machine, using the HSV color space to distinguish between the dark red cap and the red cap, and therefore between the first and third tube types, can be difficult. Figure 6 In the third-level cap standard, the first tube type is gray, and the third tube type is green. For the machine, using the HSV color space to distinguish between the gray cap and the green cap, and therefore between the first tube type and the third tube type, can be difficult.

[0038] like Figure 7 As shown, because manufacturers may use different standards, simply detecting the cap color often does not provide sufficient information to accurately identify the tube type. As disclosed herein, more accurate tube type identification involves identifying characteristics beyond just the cap color. The embodiments described herein enable differentiation of tube assembly types regardless of cap color to correctly identify certain tube properties, such as vacuum capability. In particular, the embodiments described herein can identify the tube assembly type even when different tube assembly types have similar cap colors.

[0039] More specifically, the embodiments disclosed herein can distinguish tube assembly types, including different tube functions and / or additive components, by detecting various cap characteristics, which may include at least the color of the cap and the geometry of the cap and / or tube.

[0040] The methods and apparatus described herein enable machines such as imaging devices or quality inspection modules to correctly identify and robustly distinguish between various tube assembly types, even when the cap color itself may not provide sufficient information for some identifications. The methods and imaging apparatus described herein can utilize artificial intelligence, deep learning techniques, statistical models, and / or other techniques such as discriminative models to classify tube types based on the color, shape, and / or other characteristics of the tube assembly's cap.

[0041] In some embodiments, the statistical and / or discriminative models may include support vector machines (SVMs), decision trees, convolutional neural networks (CNNs), etc., which may include previously trained models. The processes described herein may be executed by a processor coupled to memory, where executable program instructions are stored in the memory and are executable on a suitable processor to perform the methods. In some embodiments, the SVM is a supervised learning model with one or more correlation learning algorithms that analyze data for classification and regression analysis.

[0042] In a first broad aspect, embodiments of this disclosure provide imaging methods and apparatus configured and operable to determine physical characteristics of a cap and / or tube, such as one or more geometric features indicating the shape and color of the cap and / or tube. In some embodiments, gradients of one or more dimensions of the cap may be determined and used. Machines employing machine learning techniques (e.g., imaging devices or quality inspection modules) may use these characteristics of the cap and / or tube to identify and / or classify different tube assemblies.

[0043] The methods and apparatus disclosed herein can distinguish between different tube types using cap geometry, cap color, and other characteristics of the cap and tube. The methods and apparatus can identify characteristics of the tube assembly using front and / or back illumination of the tube assembly. For example, in some embodiments, a light source such as a plated light source can illuminate the front surface of the tube assembly and capture an image of reflected light from the front surface. In other embodiments, one or more light sources such as a plated light source can back-illuminate the tube assembly, and an imaging device can capture an image of light passing through the tube assembly, enabling characterization of the cap's translucency. In some cases, front illumination can provide improved discrimination at least for the cap color.

[0044] One or more images of the tube assembly are captured to generate one or more pixelated images of the tube assembly, wherein each pixelated image includes a plurality of pixels forming the image. The one or more images can be analyzed to extract geometric features of the tube and / or cap, which can be used to identify the type of tube to which the cap is attached. For example, the features of the tube and / or cap can be input into a discriminative model such as a linear support vector machine to identify the cap. In some embodiments, the tube type can be identified at least partially based on the identification of the cap.

[0045] In some embodiments, further processing of the image in the region identified as the cover may use a color determination algorithm to extract or identify the color of the cover. In some embodiments, the color value representing each pixel of the cover may be identified. For example, by using a computer algorithm, the mean value (mean color) of the color values ​​of all pixels or pixel patches may be determined. The color components of the mean color may then be determined. In some embodiments, the mean color may be converted to an HSV (hue, saturation, value) color space to produce color hue, saturation, and value components. Figure 7 The example provided is a tube component mapped to the HSV color space. These color components provide three dimensions or vectors that can be input into a discrimination model to help clearly identify the tube type. Other color representation methods can be used, such as HSL, RGB (red, green, blue), AdobeRGB, YIQ, YUV, CIELCAB, CIELUV, ProPhoto, sRGB, Luma plus Chroma, CMYK, etc.

[0046] In some embodiments, features related to the lid geometry can be input into the discriminative model. For example, geometric features related to the lid's dimensions, such as gradients (e.g., row gradients and / or column gradients), can be input into the discriminative model. As used herein, the gradient is the rate of change (first derivative) of the lid's dimensions. For example, the algorithm can scan (e.g., raster scan) an image of the lid from top to bottom or bottom to top to determine the lid's width as a function of vertical position (e.g., along the y-axis), where the determined width is calculated along the horizontal or x-axis.

[0047] For example, the algorithm can scan the image of the lid from top to bottom, analyze the shape of the lid, and store the absolute value of the width and the first-order numerical derivative along the y-axis of the lid image. The first-order numerical derivative can be calculated using the following equation (1), which is for a single axis:

[0048] Equation (1)

[0049] in:

[0050] It is the two-dimensional value of u at position indices i and j.

[0051] i is the x-axis position index.

[0052] j is the y-axis position index.

[0053] It is the numerical derivative of u at positions i and j with respect to the y-axis, and

[0054] It is a change in dimension along the vertical axis.

[0055] Equation (1) generates the row gradient. The maximum value of the row gradient, called RG-max, can be calculated. The value of RG-max is a function of the most drastic change in the canopy width and is related to the most drastic change in the canopy width, and can be a vector input to the discriminant model. In other embodiments, other geometric features of the canopy can be analyzed and input into the discriminant model. For example, the dimension or shape profile along the top surface of the canopy can be analyzed and input into the discriminant model. In other embodiments, the gradient of the canopy height can be analyzed and input into the discriminant model, i.e., the rate of change of the canopy height as a function of the width.

[0056] In some embodiments, the material of the top cover can be analyzed to obtain another distinguishing characteristic that can be input into a discriminative model. Analyzing the top cover material may include calculating a measure of the top cover's opacity or translucency. For example, the algorithm may use one or more backlit images of the top cover. In some embodiments, backlit images of the top cover can be captured at high exposure times across multiple spectra (wavelengths) of light, and the images can be analyzed by the algorithm. In some embodiments, three visible spectra (RGB) can be used. For example, a backlit red channel image may be exposed to approximately 10309. A backlit green channel image can be exposed to approximately 20615. Furthermore, the blue channel image with backlighting can be exposed to approximately 10310. Statistics for each wavelength channel can be calculated and fed into the discriminative model. For example, the mean of a high-exposure image for each wavelength in RGB can be calculated. Using these three means (R-mean, G-mean, and B-mean), the discriminative model can use a multidimensional discriminative feature space for cap identification. For example, a 7-dimensional discriminative feature space (H, S, V, RG-max, R-mean, G-mean, B-mean) can be used for cap identification. However, any combination of color feature spaces can be used.

[0057] exist n 3D feature space (in this embodiment) nIn (=7), a discriminative model (e.g., a discriminator) can be trained to correctly identify the cap and thus the tube type. An example of a discriminator is a linear support vector machine (SVM), which draws a decision hyperboundary around each cap and / or tube type in a high-dimensional feature space. The cap and / or tube type can then be identified. In some embodiments, additional features, such as cap height, cap diameter, tube diameter, and / or tube height, or other vision-based features, can be included in the discriminative model as additional dimensional or shape inputs. Cap weight, such as at the cap removal station after the imaging stage, can also be utilized. In other embodiments, additional back-illuminated light or non-visible light (e.g., IR or near-IR) can be used to add a more robust discriminative model to fully utilize the dimensional complexity of the color space.

[0058] The models and algorithms described herein can associate a specific cap type with its appropriate pipe type, rather than solely relying on the cap's color. These models and algorithms can perform these tasks without operator input; that is, the discrimination can be fully automated. The following description provides examples of implementing the methods and devices described above to distinguish pipe types based on the cap attached to the pipe. In other embodiments, the characteristics of the pipe assembly can be analyzed, and the models and algorithms can determine the pipe type based on this analysis.

[0059] This article references Figures 1A to 8 Further details about the methods and equipment are described.

[0060] Now for reference Figure 1A and Figure 1B . Figure 1A The illustration shows a side view of a first tube assembly 100A, which includes a first top cap 102A attached to the top of a first tube 104A. Figure 1BThe illustration shows a side view of a second tube assembly 100B, which includes a second cap 102B attached to the top of a second tube 104B. For example, a first tube assembly 100A may be from a first manufacturer, and a second tube assembly 100B may be from a second manufacturer, or may simply be a different tube type from the same manufacturer. The first tube assembly 100A may include one or more chemical additives (not shown) located in the first tube 104A for use in performing a first test, and the second tube assembly 100B may include one or more chemical additives (not shown) located in the second tube 104B for use in performing a second test. Referring to the first tube assembly 100A, the chemical additives may be added to the first tube 104A as a spray coating, as small balls or localized blocks, or elsewhere in the first tube 104A, or on the underside or underside of the first cap 102A. The chemical additives may be in similar locations in the second tube assembly 100B. In some embodiments, the first test may be the same as the second test, and in other embodiments, the first test may be different from the second test.

[0061] The first cap 102A may have a first color, and the second cap 102B may have a second color. Some caps may have more than one color, such as one color on one part of the cap and a second color on another part of the cap. In some embodiments, the first color may be so similar to the second color that a computer algorithm based solely on color may be unable to distinguish between the first and second colors. In other embodiments, the first color may be different from the second color. The methods and apparatus described herein can analyze the aforementioned dimensional and / or shape features and color to distinguish the first cap 102A and the second cap 102B, and thus distinguish the tube type of the first tube assembly 100A and the tube type of the second tube assembly 100B.

[0062] One or more images of the first tube assembly 100A and the second tube assembly 100B can be captured by the imaging device 800. Figure 8 An image can be a pixelated image composed of multiple rows and columns of pixels, where each pixel comprises multiple pixels. See also... Figure 2A and Figure 2B ,in Figure 2A The illustration shows a schematic diagram of a mask image of the first top cover 102A removed from the image of the first tube 104A, and Figure 2B The illustration shows a schematic image of a mask image of a second top cover 102B removed from a second tube 104B. In some embodiments, an algorithm executed on a computer can classify pixels in the image as top covers, and this classification can be used to isolate those pixels classified as top covers for further processing, such as... Figure 2A and Figure 2B As shown in the diagram. Figure 2A and Figure 2B The surrounding box indicates the mask image and Figure 1A The first tube assembly 100A and Figure 2B The second tube assembly 100B (including the same frame) is aligned in position. The size (width and height) of the frame and the position of the frame can be determined by the size and position required for any desired cap type provided on a tube of any size.

[0063] Figure 2A The pixel locations in the image (shown by shaded lines) can be used to analyze the original color image to determine the color of the first cover 102A. Determining the color of the first cover 102A can be performed by calculating the mean color, median color, mode, or other color values ​​of the pixels. More advanced unsupervised machine learning methods, such as k-means clustering, can also be used to create color clusters. The mean color can be calculated in the HSV color space to generate three color feature dimensions (H-hue, S-saturation, and V-value (luminance)) of the HSV color model, which are then input into the discriminative model of the first tube assembly 100A. The same process can be applied to... Figure 2B The pixels in the image (shown in shaded areas) are used to generate three color feature dimensions for the second top cover 102B, which are then input into the discrimination model of the second tube assembly 100B.

[0064] By analyzing the geometric features of a first tube assembly 100A, including a first top cover 102A and / or a first tube 104A, and a second tube assembly 100B, including a second top cover 102B and / or a second tube 104B, additional dimensions can be obtained as input to the discriminative model. An algorithm executed on a computer can identify steep color intensity transitions to identify edges, and then analyze dimensional gradients, such as row gradients of the first top cover 102A, and may include portions of the first tube 104A. Similarly, the algorithm can analyze dimensional gradients, such as row gradients of the second top cover 102B, and may include portions of the second tube 104B.

[0065] right Figure 3A For reference, Figure 3A The illustration shows a portion of the first tube assembly 100A, which can be analyzed to determine geometric gradients, such as row gradients. Also shown... Figure 4A For reference, Figure 4A The illustration shows a portion of the second tube assembly 100B, which can be analyzed to determine geometric gradients, such as row gradients. (Reference) Figure 3A The algorithm can scan a portion of the first tube 104A and the first top cover 102A from top to bottom (or from bottom to top) to determine numerical values ​​indicating the shape of the portion of the first tube 104A and the first top cover 102A. Figure 3AIn this embodiment, the values ​​are various widths of the top portion of the first tube 104A and the first top cover 102A. These values ​​can be varied based on the image color intensity that identifies the edges of the first top cover 102A and the first tube 104A. Optionally, a segmentation routine can be used to identify the boundaries of the top cover in pixel space.

[0066] For further reference Figure 3B , Figure 3B The diagram is illustrated graphically along Figure 3A The first tube component 100A shown is used as a function of the vertical dimension (Y dimension) and various width dimensions (x dimension). For example... Figure 3A and Figure 3B As shown, the first tube assembly 100A has a width W31 at a vertical position V31 and a width W32 at a vertical position V32. The first tube assembly 100A has a width W33 at a vertical position V33, which is the width of the first tube 104A. These widths can represent the unique geometric characteristics of the first tube assembly 100A in terms of its width as a function of its vertical position.

[0067] For further reference Figure 3C , Figure 3C The diagram illustrates the first tube assembly 100A. Figure 3B The first derivative of the dimension plot (width dimension relative to vertical dimension). Equation (1) can be applied to Figure 3B The graph, to generate Figure 3C The diagram. Note that the top row of the first top cover 102A can be ignored during processing, so the first derivative does not approach infinity when equation (1) is applied to the top row. Figure 3C The y-axis of the graph is called Width. Refers to RG-max1. The maximum width is Figure 3C As shown in the diagram, it can be input into the discrimination model of the first tube component 100A as another model input besides color.

[0068] See also Figure 4B , Figure 4B The graph illustrates the function of the vertical dimension (Y dimension). Figure 4A The second tube assembly 100B shown has various widths (x-dimensions). For example... Figure 4A and Figure 4B As shown, the second tube assembly 100B has a width W41 at a vertical position V41 and a width W42 at a vertical position V42. The second tube assembly 100B has a width W43 at a vertical position V43, which is the width of the second tube 104B. These widths represent the unique geometric characteristics of the second tube assembly 100B.

[0069] For further reference Figure 4C , Figure 4C The diagram shows... Figure 4B The first derivative of the second tube assembly 100B is plotted on the graph. Equation (1) can be applied to... Figure 4B The graph, to generate Figure 4C The diagram shows that the top row of the second top cover 102B can be ignored during processing, so the first derivative does not approach infinity when calculating the derivative of the first row. Figure 4C The y-axis of the graph is called width. The maximum peak value of the width (called RG-max2) is in Figure 4C As shown in the diagram, it can also be input as another model input into the discrimination model of the second tube assembly 100B. In some embodiments, One or more minimum width values ​​can be used as model input along with color.

[0070] Based on the foregoing, the gradient of the first tube assembly 100A and the second tube assembly 100B can be used to identify and / or distinguish at least some features of the tube assemblies without considering the colors of the first top cover 102A and the second top cover 102B. For example, it can be used to distinguish vacuum capabilities by judging the top cover gradient. In some embodiments, the gradient of the top cover width of the first top cover 102A and the second top cover 102B can be used to distinguish at least some features of the first tube assembly 100A and the second tube assembly 100B, respectively.

[0071] Another distinguishing characteristic of the first tube assembly 100A and the second tube assembly 100B may be the top cover material and / or the tube material, which can be determined by calculating a measure of opacity or translucency. To measure this characteristic, the portion of the tube assembly including the top cover can be backlit, and an image of the tube assembly can be captured, for example, at a high exposure time, and can also be captured using illumination of multiple spectra. In some embodiments, three visible spectra (red (R), green (G), and blue (B)) can be used to backlit the first tube assembly 100A and the second tube assembly 100B. However, other colors can be used, and even colors including UV and IR backlighting can be used. For example, invisible light can be used to illuminate the first tube assembly 100A and the second tube assembly 100B.

[0072] refer to Figure 5A , Figure 5A The illustration shows a first top cover 102A passing through the first tube assembly 100A and being conveyed by an imaging device (e.g., Figure 8 A plot of an example spectrum captured by (107A). Also refer to... Figure 5B , Figure 5B The illustration shows an example spectrum passing through the second top cover 102B of the second tube assembly 100B and captured by the imaging device. Figure 5A and Figure 5B The image is plotted on an intensity scale from 0 to 255 (e.g., mean intensity), where 0 represents no light received and 255 represents no light blocked by the tube assembly. In some embodiments, the backlit red channel image can be exposed to approximately 10309. A backlit green channel image can be exposed to approximately 20615. Furthermore, the blue channel image with backlighting can be exposed to approximately 10310. Other exposures can be used.

[0073] For further reference Figure 5C , Figure 5C The illustrations show photographic images of a portion of the first tube assembly 100A on top of another. The top photographic image (labeled 1) is a portion of the first tube assembly 100A captured using foreground illumination with the full RGB color spectrum. The second through fourth monochrome photographic images illustrate the transparency of the tube portions to light of different wavelengths (e.g., colors). For example, the second monochrome photographic image (labeled 2) is a portion of the first tube assembly 100A captured using red background illumination. The third monochrome photographic image (labeled 3) is a portion of the first tube assembly 100A captured using green background illumination. The fourth monochrome photographic image (labeled 4) is a portion of the first tube assembly 100A captured using blue background illumination.

[0074] Also refer to Figure 5D , Figure 5D The illustration shows photographic images of a portion of the second tube assembly 100B above another. The top photographic image is a portion of the second tube assembly 100B captured using foreground illumination in the full RGB color spectrum (labeled 1). The second monochrome photographic image (labeled 2) is a portion of the second tube assembly 100B captured using red background illumination. The third monochrome photographic image (labeled 3) is a portion of the second tube assembly 100B captured using green background illumination. The fourth monochrome photographic image (labeled 4) is a portion of the second tube assembly 100B captured using blue background illumination.

[0075] from Figure 5A and Figure 5B The diagram shows that the portion of the first tube assembly 100A labeled 1 is almost completely transparent across all wavelengths (R, G, B), and the portion of the second tube assembly 100B labeled 1 blocks a significant amount of light across all three wavelengths. In some embodiments, the mean of each wavelength in the high-exposure image is calculated and referred to as R-mean, G-mean, and B-mean, respectively. In some embodiments, the median of each wavelength in the high-exposure image is calculated.

[0076] When calculating all the above eigenvalues, a 7-dimensional discriminative feature space (H, S, V, RG-max, R-mean, G-mean, B-mean) can exist for each tube component. n In the 3D feature space (in this embodiment) n =7), a discriminative model can be trained to correctly identify various tube types. An example of a discriminator is a linear SVM, which draws a decision hyperboundary around each tube type in this high-dimensional feature space. Based on the aforementioned model, even if the first cap 102A and the second cap 102B have exactly the same or similar colors, the first tube assembly 100A and the second tube assembly 100B can be distinguished. By utilizing more optical features—such as cap opacity or translucency, cap weight, cap vertical height, diameter, (one or more) cap gradients or other vision-based geometric features, or additional image types—such as different backlighting or illumination using invisible light (e.g., IR or near IR), even more powerful discriminative models can be used to fully exploit the dimensional complexity of the space. Any combination of color space features and geometric features can be used for cap type discrimination.

[0077] The apparatus and methods described herein enable diagnostic laboratories to distinguish between many different tube types that may pass through the laboratory. Relying solely on cap color can lead to unreliable results due to varying manufacturer and / or regional standards. Based on the foregoing, when cap color alone is insufficient to differentiate between various tube types, the apparatus and methods disclosed herein improve the discrimination of various tube types and facilitate differentiation between them. This is advantageous because it allows diagnostic devices or machines to determine the type of tube (and therefore its corresponding characteristics) within the diagnostic device or equipment without requiring any manual input from the operator; i.e., it can be fully automated.

[0078] In some embodiments, some corresponding features may be manually entered by the operator or obtained via another sensor. Technical features contributing to the advantages of the apparatus and methods described herein may include a high-dimensional feature vector for each tube type using data collected from one or more airborne image capture devices or sensors, and a discriminant model in high-dimensional space to correctly determine the tube type. Using such a high-dimensional discriminant model can accelerate sample processing workflows and correctly identify mismatches between ordered tests and the tube type used. Therefore, checks can be performed to ensure that the correct tube type is used based on its chemical additives or geometric properties, such as those indicated by the cap.

[0079] Figure 8An imaging device 800 is illustrated, configured to capture one or more images of a first tube assembly 100A consisting of a first top cover 102A and a first tube 104A, wherein the one or more images may be pixelated images composed of pixels, with each image comprising a plurality of pixels. The imaging device 800 may also be referred to as a diagnostic device. The imaging device 800 includes a front light source, such as a light source 805A, which may be a light plate. In some embodiments, more than one light source may be used, such as a light plate positioned on either lateral front side of the imaging device 807. Another front light source, such as a light plate, may be provided on another lateral front side of the imaging device 807 (as shown outside the paper) at the same longitudinal position as light source 805A. In some embodiments, one or more back light sources, such as light source 805B, may be provided. For example, light source 805B may be one or more light plates that provide light for determining opacity. Other arrangements are also possible.

[0080] Imaging apparatus 800 may further include a controller 809 communicatively coupled to imaging apparatus 807 and light sources 805A, 805B. Controller 809 may be any suitable computer, including a memory 812 and a processor 810 suitable for storing and executing executable program instructions in the form of a discrimination model. Controller 809 may send signals to light sources 805A, 805B at appropriate times to provide front lighting and / or back lighting depending on the type of image being captured (backlighting or front lighting). Controller 809 may identify the first top cover 102A and the first tube 104A using any suitable segmentation program (such as a CNN or other trained model).

[0081] Once the first cap 102A and its representative area are identified, its color can be determined by a color model in any suitable color space, such as the HSV color model. Other suitable multi-element color spaces or models can be used, such as HSL, RGB (red, green, blue), Adobe RGB, YIQ, YUV, CIELCAB, CIELUV, ProPhoto, sRGB, Luma plus Chroma, CMYK, or other suitable color space models. Similarly, once the first cap 102A and the first tube 104A are identified, a dimension determination procedure or routine can determine the width and / or height of the first tube 104A and / or the gradient of that dimension (e.g., width as a function of cap height and / or height as a function of width). These values ​​can be input into any suitable identification model (e.g., a discriminative model) to determine the type of the first tube assembly 100A.

[0082] Now for reference Figure 9 , Figure 9This is a flowchart of method 900 for identifying a tube type. Method 900 includes, in 902, capturing one or more images of a cap (e.g., first cap 102A) attached to a tube (e.g., first tube 104A), the capture generating a pixelated image of the cap, the pixelated image comprising a plurality of pixels. Method 900 includes, in 904, identifying the color of one or more pixels of the pixelated image of the cap. Method 900 includes, in 906, identifying one or more gradients of the dimensions of the cap. Method 900 includes, in 908, identifying the tube type based at least on the color of one or more pixels and one or more gradients of the dimensions of the cap.

[0083] Now for reference Figure 10 , Figure 10 This is a flowchart of method 1000 for identifying tube types. Method 1000 includes, in 1002, capturing one or more images of a top cover (e.g., a first top cover 102A), which generates a pixelated image of the top cover. Method 1000 includes, in 1004, identifying the color of the top cover. The method includes, in 1006, identifying the dimensional gradient of the top cover. Method 1000 includes, in 1008, identifying the tube type based at least on the color and dimensional gradient of the top cover. The method includes, in 1008, identifying a match between the arranged test and the tube type. If the tube type is suitable for the arranged test, controller 809 may indicate on the display that "correct tube type used" or report that there is no error in the tube type used. If the indication does not match, the system may report an error, such as by indicating on the display or otherwise indicating that "incorrect tube type" was used.

[0084] While this disclosure is susceptible to various modifications and alternatives, specific embodiments and methods thereof have been illustrated by way of example in the accompanying drawings and described in detail herein. However, it should be understood that this is not intended to limit this disclosure to the particular system or method disclosed, but rather, the invention is intended to cover all modifications, equivalents, and alternatives falling within the scope of the claims.

Claims

1. A method for identifying pipe type, comprising: Capture one or more images of the top cap attached to the tube, the capture generating a pixelated image of the top cap, the pixelated image comprising a plurality of pixels; The color of one or more pixels in the pixelated image of the top cover; Identify one or more gradients in the width or height dimension of the top cover; and Pipe type should be identified based on at least the following: The color of the one or more pixels, and One or more gradients in the width or height dimension of the top cover.

2. The method according to claim 1, wherein, The identification tube type is based on a discriminant model.

3. The method according to claim 2, wherein, One or more inputs to the discriminant model include the width and / or height of the pipe.

4. The method according to claim 2, wherein, One or more inputs to the discriminative model include one or more gradients in the width or height dimension of the dome.

5. The method according to claim 2, wherein, The input to the discrimination model includes the weight of the top cover.

6. The method according to claim 2, wherein, The input to the discrimination model includes images captured using invisible light illumination, including IR or near-IR.

7. The method according to claim 2, wherein, The input to the discriminative model includes a combination of color feature spaces.

8. The method according to claim 2, wherein, The input to the discriminative model includes a combination of a color feature space and a discriminative feature space, wherein the discriminative feature space includes one or more gradients of the width or height dimension of the cap.

9. The method according to claim 2, wherein, The input to the discriminant model includes a 7-dimensional discriminant feature space, which includes hue (H), saturation (S), value (V), maximum row gradient (RG-max), R-mean, G-mean, and B-mean.

10. The method according to claim 2, wherein, The discriminant model is a linear support vector machine.

11. The method according to claim 1, wherein, The color of one or more pixels of the pixelated image that identifies the top cover includes the use of the top cover's opacity or semi-transparency and backlighting.

12. The method according to claim 1, wherein, Identifying the color of the one or more pixels involves calculating the average color of multiple pixels in the pixelated image of the top cover.

13. The method according to claim 1, wherein, The color that identifies the one or more pixels is determined by a color space model in a multi-element color space, the color space model being selected from one or more of the following: HSV, HSL, RGB, Adobe RGB, YIQ, YUV, CIELCAB, CIELUV, ProPhoto, sRGB, Luma plus Chroma, CMYK or other suitable color space models.

14. The method according to claim 1, wherein, Identifying the color of the one or more pixels includes calculating the mean color of a plurality of pixels on top of the pixelated image and identifying the hue (H), saturation (S), and value (V) components of the mean color in the HSV color space.

15. The method according to claim 1, wherein, One or more gradients in the width or height dimension of the top cover include one or more width gradients.

16. The method according to claim 15, wherein, The one or more width gradients include the maximum value of the width gradient.

17. The method according to claim 1, wherein, One or more gradients in the width or height dimension of the lid include multiple gradients in the lid width or lid height.

18. The method of claim 1, further comprising identifying a match or mismatch between the scheduled test and the tube type.

19. The method of claim 1, further comprising: Material of the sign top cover; Pipe type should be identified based on at least the following: The material of the top cover The color of the top cover, and Gradient of width or height dimension.

20. A diagnostic device, comprising: An imaging device configured to capture one or more images of a tube assembly including a top cap attached to the tube, wherein the one or more images include one or more pixelated images of the top cap. and A controller communicatively coupled to the imaging apparatus, the controller including a processor coupled to a memory storing executable program instructions that are executable for: Determine the color of one or more pixels in one or more pixelated images of the top cover; Determine one or more gradients in the width or height dimension of the roof; and Pipe type should be identified based on at least the following: The color of one or more pixels, and One or more gradients in the width or height dimension of the top cover.

Images