CN110675386B - Detection system for group B streptococcus - Google Patents
Detection system for group B streptococcus Download PDFInfo
- Publication number
- CN110675386B CN110675386B CN201910915252.XA CN201910915252A CN110675386B CN 110675386 B CN110675386 B CN 110675386B CN 201910915252 A CN201910915252 A CN 201910915252A CN 110675386 B CN110675386 B CN 110675386B
- Authority
- CN
- China
- Prior art keywords
- colony
- microbial
- streptococcus
- group
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10056—Microscopic image
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Apparatus Associated With Microorganisms And Enzymes (AREA)
- Measuring Or Testing Involving Enzymes Or Micro-Organisms (AREA)
Abstract
The embodiment of the invention discloses a detection system of group B streptococcus, which relates to the field of clinical microbiology inspection, and comprises the following components: a microbial colony culture dish in which a plurality of microbial colonies to be screened are grown; the shooting device is used for shooting the microbial colony culture dish to obtain microbial colony culture dish images containing a plurality of microbial colony images to be screened; the communication equipment is used for sending the microbial colony culture dish image containing the plurality of microbial colony images to be screened to the cloud analysis device; and the cloud analysis device is used for analyzing the plurality of microbial colony images to be screened in the microbial colony culture dish image shot by the shooting equipment and determining whether the group B streptococcus colonies exist in the plurality of microbial colonies to be screened. The embodiment of the invention is based on image recognition of the group B streptococcus, and greatly reduces the labor cost and the equipment cost on the basis of ensuring the screening accuracy.
Description
Technical Field
The embodiment of the invention relates to the field of clinical microbiology inspection, in particular to a detection system for group B streptococcus.
Background
Group B streptococcus (Group B Streptococcus, GBS) is an aerobic gram-positive coccus, a pathogenic bacterium that has been considered invasive infection of newborns since the 70 s of the 20 th century, and can cause severe infections such as sepsis, pneumonia, and meningitis in newborns. The most important risk factor for neonatal progression to invasive infections is the parturient genitourinary tract or gastrointestinal colonisation of GBS, and antibiotic prophylaxis at least 4 hours prior to delivery is effective in avoiding most neonatal GBS infections. The us CDC issued recommendations for preventing neonatal GBS infection, greatly reducing neonatal GBS infection rates. The experts in China also release comments, which emphasize the importance of GBS screening and prevention in pregnant women in perinatal period in China.
GBS screening should be performed at 35-37 weeks of gestation, collecting vaginal lower and rectal swabs, typically by inoculating into Todd-Hewitt broth (containing antibiotics to inhibit the growth of infectious agents), incubating for 24 hours at 35 ℃, then transferring to a golombian blood agar (Columbia Blood Agar, CBA) plate, incubating for 24-48 hours at 35 ℃, observing the presence or absence of suspicious GBS colonies (with narrow beta hemolytic rings, white, smooth colonies with gradual boundary) on the CBA plate, and selecting suspicious colonies for further strain identification by means of 16S rDNA gene sequencing, automated bacterial identifier, or matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS), etc. Other GBS detection methods also include GBS fluorescent quantitative PCR detection, GBS chromogenic medium containing special substrate, and the like.
However, the above identification methods all require skilled technicians to perform preliminary screening on suspicious colonies on the CBA plate, and then further identification is performed by using expensive instruments such as mass spectrometers, automated bacterial identification instruments, fluorescent quantitative PCR instruments or gene sequencers, and the cost of commercial chromogenic media is high, and the primary hospitals are generally not burdened, so that the methods are more suitable for large hospitals with better economic conditions. In the vast primary hospitals and women and young health care hospitals with a large number of pregnant women, the popularization of GBS screening is limited by the two aspects of the technology and the cost.
Disclosure of Invention
The detection system for the group B streptococcus solves the problem that the group B streptococcus is difficult to screen and prevent in perinatal pregnant women due to the limitation of the technology and the cost in the vast primary hospitals and women and child care hospitals.
The detection system for the group B streptococcus provided by the embodiment of the invention comprises:
a microbial colony culture dish in which a plurality of microbial colonies to be screened are grown;
the photographing device is used for photographing the microbial colony culture dish to obtain microbial colony culture dish images containing a plurality of microbial colony images to be screened;
the communication equipment is used for sending the microbial colony culture dish image containing the plurality of microbial colony images to be screened to the cloud analysis device;
and the cloud analysis device is used for analyzing the plurality of microbial colony images to be screened in the microbial colony culture dish image shot by the shooting equipment and determining whether the group B streptococcus colonies exist in the plurality of microbial colonies to be screened.
Preferably, the cloud analysis device includes:
the image recognition device is used for screening a plurality of microbial colony images of suspected group B streptococcus colonies from the plurality of microbial colony images to be screened and marking the plurality of microbial colony images of the suspected group B streptococcus;
and the colony analysis device is used for analyzing the plurality of microbial colony images of the suspected group B streptococcus marked by the image recognition device by utilizing an artificial neural network and determining whether the plurality of microbial colony images of the suspected group B streptococcus marked by the image recognition device are group B streptococcus colony images.
Preferably, the image recognition device is configured to screen a plurality of microbial colony images of suspected group B streptococcus colonies from the plurality of microbial colony images to be screened, and specifically includes:
the bacterial colony characteristic analysis module is used for analyzing each microbial colony image to be screened in the microbial colony culture dish image to obtain a microbial colony characteristic value in each microbial colony image;
the group B streptococcus colony preliminary screening module is used for determining whether each microbial colony image is a suspected group B streptococcus microbial colony image according to the microbial colony characteristic value in each microbial colony image and a preset group B streptococcus colony characteristic threshold range.
Preferably, the microbial colony characteristic values comprise sizes, colors and gradient of color change from center to edge of the microbial colony, and the preset group B streptococcus colony characteristic threshold range comprises a colony size threshold range, a colony color threshold range and a gradient of color change from center to edge of the colony.
Preferably, the image recognition device determines, for each microbial colony, that each microbial colony image is a microbial colony image of suspected group B streptococcus when the size, color, and colony center-to-edge color gradient of the microbial colony are within the colony size threshold range, colony color threshold range, and colony center-to-edge color gradient threshold range, respectively, of the preset group B streptococcus.
Preferably, the artificial neural network is a convolutional neural network comprising a convolutional layer, a modified linear unit layer, a pooling layer and a fully connected layer, and the colony analysis device analyzes the plurality of microbial colony images of the suspected group B streptococcus marked by the image recognition device by using the convolutional neural network comprising the convolutional layer, the modified linear unit layer, the pooling layer and the fully connected layer to determine whether the plurality of microbial colony images of the suspected group B streptococcus marked by the image recognition device are group B streptococcus colony images.
Preferably, the system further comprises a photographing auxiliary device for photographing the microbial colony, the photographing auxiliary device comprising:
the box body is provided with a top plate, a bottom plate and side plates;
the supporting seat is arranged above the bottom plate;
the lamp housing of the bottom light source is arranged on the supporting seat and can rotate relative to the bottom plate;
and the culture dish fixing clamp is arranged on the upper part of the lamp housing and is used for clamping the microorganism colony culture dish.
Preferably, the photographing auxiliary device further comprises a movable arm mounted above the top plate and capable of rotating or bending relative to the top plate, and a photographing fixing clamp mounted at the extending end of the movable arm and used for clamping the photographing equipment.
Preferably, the photographing auxiliary device further comprises: at least one of a plane scattering light source arranged on the lower surface of the top plate, an annular light source arranged on the inner wall of the side plate and an ultraviolet lamp disinfection device arranged on the inner wall of the side plate.
The technical scheme provided by the embodiment of the invention has the following beneficial effects:
the embodiment of the invention is based on image recognition of the group B streptococcus, does not need to be screened for a first time by experienced technicians or purchase expensive instruments and equipment, greatly reduces labor cost and equipment cost on the basis of ensuring screening accuracy, and is suitable for screening and preventing GBS of pregnant women in wide and big primary hospitals and women and young health care hospitals.
Drawings
FIG. 1 is a schematic block diagram of a group B streptococcus detection system according to an embodiment of the present invention;
FIG. 2 is a schematic block diagram of another group B streptococcus detection system provided in accordance with an embodiment of the present invention;
FIG. 3 is a schematic block diagram of the cloud analysis device shown in FIG. 1 or FIG. 2;
FIG. 4 is a schematic view of a photographing auxiliary device of the microbial colony culture dish shown in FIG. 1 or FIG. 2;
FIGS. 5a and 5b are GBS pure colony patterns and GBS colony patterns of clinical samples without fractionation, respectively;
FIG. 6 is a schematic block diagram of a detection system for group B streptococcus according to an embodiment of the present invention;
FIG. 7 is a flow chart of single colony image segmentation (labeling) provided by an embodiment of the present invention;
FIG. 8 is a flow chart of detecting whether individual colonies that have been segmented (labeled) are GBS according to an embodiment of the present invention;
fig. 9 is a flow chart of GBS screening after application of the system according to an embodiment of the present invention.
Detailed Description
The following detailed description of the preferred embodiments of the present invention is provided in conjunction with the accompanying drawings, and it is to be understood that the preferred embodiments described below are merely illustrative and explanatory of the invention, and are not restrictive of the invention.
FIG. 1 is a schematic block diagram of a system for detecting group B streptococcus according to an embodiment of the present invention, as shown in FIG. 1, the system may include:
a microbial colony culture dish, such as a CBA plate or the like, in which a plurality of microbial colonies to be screened are grown;
photographing equipment, such as a camera, a mobile phone camera and the like, for photographing the microbial colony culture dish to obtain a microbial colony culture dish image containing a plurality of microbial colony images to be screened;
communication equipment, such as a WiFi module, a 3, 4, 5G module and the like, is used for sending the microbial colony culture dish image containing the plurality of microbial colony images to be screened to a cloud analysis device;
and the cloud analysis device is used for analyzing the plurality of microbial colony images to be screened in the microbial colony culture dish image shot by the shooting equipment and determining whether the group B streptococcus colonies exist in the plurality of microbial colonies to be screened.
In this embodiment, the photographing apparatus and the communication apparatus may be implemented by a camera module and a communication module of a mobile terminal (for example, a doctor's mobile phone). Therefore, after the cloud analysis device analyzes and determines whether the group B streptococcus colony exists in the plurality of microbial colonies to be screened, the mobile terminal (for example, a doctor mobile phone) can obtain the result from the cloud analysis device, so that a doctor can view the result on the mobile terminal.
Fig. 2 is a schematic block diagram of another group B streptococcus detection system according to an embodiment of the present invention, as shown in fig. 2, where on the basis of the embodiment of fig. 1, the system may further include a doctor computer, a microbial colony culture dish image including a plurality of microbial colony images to be screened, which is shot by a doctor mobile phone, may be directly uploaded to a cloud analysis device by a communication device on the doctor mobile phone, or may be uploaded to the cloud analysis device by a communication device on the doctor mobile phone via the doctor computer, and similarly, both the doctor mobile phone and the doctor terminal may obtain results analyzed and determined by the cloud analysis device from the cloud analysis device.
The system provided in the embodiments of fig. 1 and fig. 2 uses the photographing device to photograph a microbial colony culture dish growing a plurality of microbial colonies to be screened, obtains a microbial colony culture dish image containing a plurality of microbial colony images to be screened, and uses the communication device to send the microbial colony culture dish image containing the plurality of microbial colony images to be screened to the cloud analysis device, the cloud analysis device analyzes the plurality of microbial colony images to be screened in the microbial colony culture dish image obtained by the photographing device, determines whether a group B streptococcus colony exists in the plurality of microbial colonies to be screened, and returns the result of whether a group B streptococcus colony exists in the plurality of microbial colonies to be screened to the communication device for the communication device to transmit to a doctor for checking. Therefore, primary screening of experienced technicians is not needed in primary hospitals and women and child healthcare hospitals, expensive instruments and equipment are not needed to be purchased, and as long as a doctor shoots a plurality of microbial colony culture dish images to be screened and uploads the microbial colony culture dish images to the cloud, analysis and detection results of the cloud can be obtained, so that labor cost and equipment cost of GBS screening of pregnant women in primary hospitals and women and child healthcare hospitals are greatly reduced.
Fig. 3 is a schematic block diagram of the cloud analysis apparatus shown in fig. 1 or fig. 2, and as shown in fig. 3, the cloud analysis apparatus may include:
the image recognition device is used for screening a plurality of microbial colony images of suspected group B streptococcus colonies from the plurality of microbial colony images to be screened and marking the plurality of microbial colony images of the suspected group B streptococcus;
and the colony analysis device is used for analyzing the plurality of microbial colony images of the suspected group B streptococcus marked by the image recognition device by utilizing an artificial neural network and determining whether the plurality of microbial colony images of the suspected group B streptococcus marked by the image recognition device are group B streptococcus colony images.
In one embodiment, the image recognition apparatus of fig. 3 may include:
the bacterial colony characteristic analysis module is used for analyzing each microbial colony image to be screened in the microbial colony culture dish image to obtain a microbial colony characteristic value in each microbial colony image;
the group B streptococcus colony preliminary screening module is used for determining whether each microbial colony image is a suspected group B streptococcus microbial colony image according to the microbial colony characteristic value in each microbial colony image and a preset group B streptococcus colony characteristic threshold range.
The microbial colony characteristic values comprise sizes, colors and gradient of color change from the center of the colony to the edge of the colony of the microorganism, and the corresponding threshold range of the colony characteristic of the preset group B streptococcus comprises a threshold range of the size of the colony of the preset group B streptococcus, a threshold range of the color of the colony and a threshold range of the gradient of color change from the center of the colony to the edge of the colony. Thus, for each microbial colony, the image recognition device can determine that each microbial colony image is a microbial colony image of suspected group B streptococcus when the size, color and center-to-edge color gradient of the microbial colony are within the preset group B streptococcus colony size threshold range, the colony color threshold range, and the center-to-edge color gradient threshold range, respectively.
In one embodiment, the artificial neural network employed by the colony analysis device of FIG. 3 may be a convolutional neural network including a convolutional layer, a modified linear unit layer, a pooling layer, and a fully-connected layer. Assume that the CNN is a trained CNN, i.e., the weights of the layers are known. To ensure the accuracy of the colony analysis device in detecting GBS mixed in normal flora, the following operations are performed in this embodiment:
1. input layer: 200×200×3, i.e., 200 pixels×200 pixels×3 dimensions (i.e., RGB three dimensions).
The value of each pixel in each dimension (e.g., the dimension corresponding to red) is the product of the pixel red value and the darkness;
2. convolution layer: 3 x 100, i.e., the convolutional layer employs 100 convolution kernels of 3*3.
Each convolution kernel focuses on an image feature, such as a vertical edge, a horizontal edge, a color, a texture, etc., so that 100 convolution kernels are a feature extractor set of the whole image, and thus, image feature data a×100 can be obtained after convolution (the size of a is related to the convolution step size), that is, image feature data of 100 groups a×a.
3. And correcting the linear unit layer to reset the negative number in the image characteristic data of a×a×100 to 0, and obtaining the image characteristic data of a×a×100.
4. And the pooling layer performs dimension reduction processing on the image characteristic data of a x 100 by using a 2 x 2 filter, preferably a maximum pooling method, and retains the information with scale invariance.
And 5, repeatedly performing convolution, linear correction and dimension reduction processing until the image characteristic data of 2 x 100 are obtained, namely 100 groups of image characteristic data of 2 x 2.
6. Complete tie layer: 2 x 100, that is, the full connection layer adopts 100 x 2 convolution kernels of 2 x 2, and processes the image feature data of 2 x 100 to obtain 1*2 data, that is, the GBS value and the GBS value, and sends the data to the output layer.
7. The output layer may be considered a classifier that ultimately decides whether or not to be a GBS based on the value of the GBS and the value of the non-GBS.
In practice, the image feature data of 2×2×100 may be arranged into column data of 1×1×400, each value participates in voting, and whether the GBS exists or not is determined, so that a voting list formed by the voting result of whether the GBS exists or not corresponding to each value may be obtained, and finally whether the GBS exists or not is determined according to the voting list formed by the voting result of whether the GBS exists or not corresponding to each value.
Those of ordinary skill in the art will appreciate that the functional modules/units in the cloud analysis apparatus disclosed above may be implemented as software, firmware, hardware, and suitable combinations thereof. In a hardware implementation, the division between the functional modules/units mentioned in the above description does not necessarily correspond to the division of physical components; for example, one physical component may have multiple functions, or one function or step may be performed cooperatively by several physical components. Some or all of the physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on computer readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). The term computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data, as known to those skilled in the art. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital Versatile Disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by a computer. Furthermore, as is well known to those of ordinary skill in the art, communication media typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media.
On the basis of the embodiment, the system further comprises a photographing auxiliary device for photographing the microbial colony for obtaining better photographing effect and higher picture definition. FIG. 4 is a schematic diagram of a photographing auxiliary device of the microbial colony culture dish shown in FIG. 1 or FIG. 2, as shown in FIG. 4, including:
a case 1 having a top plate, a bottom plate, and side plates;
the supporting seat 10 is arranged above the bottom plate;
a bottom light source 2, the lamp housing of which is mounted on the supporting seat 10 and can rotate relative to the bottom plate;
and the culture dish fixing clamp 4 is arranged on the upper part of the lamp housing and is used for clamping the microbial colony culture dish.
Wherein, the through-hole that is used for shooing is offered at the roof middle part of box 1.
The lamp housing of the bottom light source 2 can be provided with a handle 3 for pushing the lamp housing to change the angle, so that the angle of the bottom light source 2 relative to the bottom plate can be adjusted.
The structure of the supporting seat can adopt a connecting component in the prior art, and the description is omitted here.
The culture dish fixing clamp 4 may be at least one group of clamps according to the prior art, and will not be described herein.
According to the embodiment, the supporting seat 10 is arranged above the bottom plate of the box body 1, the bottom light source 2 opposite to the photographing hole is supported by the supporting seat 10, so that light rays emitted by the bottom light source 2 are transmitted upwards from the bottom of the microbial colony culture dish arranged above the bottom light source 2, microbial colonies in the microbial colony culture dish are set off and are lightened, different types of hemolysis rings (alpha hemolysis and beta hemolysis) of the microbial colonies are easy to distinguish, especially narrow beta hemolysis rings of GBS colonies can be easily identified, and therefore, the microbial colonies can be clearly photographed by using photographing equipment, and further, accurate analysis of the bacterial colonies is facilitated.
On the basis of the structure of the photographing auxiliary device, the photographing auxiliary device may further comprise: the plane scattering light source 6 is arranged on the lower surface of the top plate, the annular light source 5 is arranged on the inner wall of the side plate, and at least one ultraviolet lamp disinfection device is arranged on the inner wall of the side plate, so that reflection of light on the surface of the microbial colony culture dish is prevented to the greatest extent, and the effect similar to that of an shadowless lamp is achieved.
The light sources of the embodiment can adopt light sources capable of emitting natural light, cold light and warm light, such as an LED lamp and the like, and the brightness and the light color of the light sources can be adjusted by adopting a circuit capable of realizing brightness and light color adjustment in the prior art, so that the light intensity of a plurality of light sources from top to bottom in the box body can be adjusted, the light color can be selected, the photographing requirements of different types of culture mediums (such as a CBA flat plate, a Chinese blue flat plate, a chocolate flat plate and the like) or different colony colors (white, gray, yellow and the like) can be met, and the optimal photographing effect can be achieved.
Because of the problem of blurred photographed images caused by shake of hands when the photographing apparatus is held by hand, the present embodiment may further install a movable arm 7 above the top plate, which is rotatable or bendable with respect to the top plate, and install a photographing fixture 8 at an extending end of the movable arm 7 for clamping the photographing apparatus, based on the structure of the photographing auxiliary device.
Wherein, the angle of the fixed clamp 8 of shooing can be changed relative to the roof to can carry out multi-angle, three-dimensional shooting to colony characteristics such as arch, recess according to the different circumstances of colony.
Three common types of convolutional neural networks (Convolutional neural network, CNN) (namely the traditional CNN, alexNet, autoencoder) are adopted to classify and identify the most common 18 bacterial colony images in clinic, the result of bacterial automation identifier and mass spectrum identification is a gold standard, and the overall coincidence rate of three types of CNN algorithms is 73%. This primarily demonstrates the feasibility of bacterial classification and identification by colony image recognition. However, the bacterial colonies identified in the above study were single colony images of passaged parturients, as shown in FIG. 5a, rather than images of clinical specimens directly inoculated onto CBA plates. In practice, the swabs used for GBS screening typically contain normal flora of the female genital tract, with suspected GBS colonies often intermixed therein, as shown in fig. 5 b. Identification of such colony images is therefore more complex and challenging.
To this end, the present invention contemplates a system for screening pregnant women for GBS based on image recognition technology, fig. 6 is a schematic block diagram of a detection system for group B streptococcus provided by an embodiment of the present invention, as shown in fig. 6, which may include:
1. CBA flat plate
2. Photographing auxiliary device
The photographing auxiliary device fixes the CBA plate, so that the narrow beta hemolytic ring around the GBS colony is more obvious.
3. Photographing device
And setting the optimal position of the fixed photographing equipment above the photographing auxiliary device so as to achieve optimal photographing effect and definition.
4. Cloud server
The image recognition device may run image recognition software to first segment (tag) valid individual colony images from the entire petri dish image so that the colony analysis device compares and detects individual colonies with standard GBS pure colonies in the "machine learning" database to determine if the individual colonies that are segmented (tagged) are GBS, as follows.
(1) First, the image recognition apparatus performs image segmentation (labeling) using a threshold segmentation method.
Specifically, three parameters of colony size (size), colony color (Hue), and gradient of color change from center to edge (gradient) of the colony were set. The thresholds of the three parameters are adjusted and set according to the GBS colony characteristics, and the colonies meeting the requirements can be automatically segmented (marked). The specific parameter setting process should use known GBS colonies to perform pre-experimental determination, namely, all GBS colonies and similar colonies can be marked to the maximum extent by optimizing the optimal parameter range aiming at the characteristics of the GBS colonies, and a foundation is laid for the next judgment.
Illustrating: setting the diameter of the colony to be 1-5 mm, setting the color gradient from the center of the colony to the edge to be 50-80 (namely, setting the color gradient from 0-100, wherein 0 represents that the color of the center of the colony is consistent with that of the edge, and 100 represents that the center of the colony is completely different from the edge) within the color tone change range from pure white to gray. Individual colonies within the set threshold values for all three parameters are segmented (labeled) and colonies that do not meet any one or more of the three parameters are not labeled. The labeled colonies were stored as several individual colony pictures, as shown in the individual colony image segmentation (labeling) flow chart of fig. 7.
This example screens out a portion of non-GBS colonies by image segmentation (labeling).
(2) Then, the colony analysis apparatus detects whether the single colony separated (labeled) is GBS by using the CNN method, and a flow chart for detecting whether the single colony separated (labeled) is GBS is shown in FIG. 8.
The overall architecture of CNN is composed of a number of different layers (layers) that transform data of an input layer into data of an output layer by performing different functions. The scale of each picture in the Input layer (Input layer) is represented by three-dimensional numbers, i.e. "length, width, RGB colors". The input layer is an image of a single colony, and the parameters of length, width and RGB used in the embodiment are 200, 3 in order to highlight the characteristics of gradient boundary and narrow beta hemolysis ring of the GBS colony. The second layer is a convolutional layer (Convolutional layer) whose "length x width x number of layers" parameter is set to 3 x 100. The third layer is a modified linear cell layer (Rectified linear unit layer) that brings all negative cells in the convolutional layer to zero. Since the convolution layer generates a lot of data, the fourth layer is a Pooling layer (Pooling layer) for reducing the amount of data calculated and further fitting without losing the original effective information, and the parameters of this layer are set to 2×2×100. The process is then cycled back from the fourth layer to the second layer, and the original 200×200×3 colony pictures are converted into a number of 2×2×100 datasets. The fifth layer is a fully connected layer (Fully connected layer), and the 2×2×100 data set is first split into 1×1×400 data sets (the data sets contain the result of previous operations in the cells) according to small cells, and the 1×1×400 data sets are arranged in a row, and voting (vote) is performed according to a numerical value, for example, a cell having a numerical value of 1 or more is "GBS" and a cell having a numerical value of <1 is "not GBS". The final voting result is the Output layer (Output layer). The determination of whether a single colony image is GBS is completed. According to the embodiment, whether GBS is mixed in normal flora of a vaginal/rectal swab sample of a pregnant woman can be detected, and the coincidence rate of the detection system can reach 80% -85% by taking the results of a bacterial automatic identifier and mass spectrum identification as gold standards.
Fig. 9 is a flowchart of GBS screening after applying the system according to the embodiment of the present invention, as shown in fig. 9, the detection flow of the system is as follows: 1. the pregnant woman genital tract swab is inoculated into Todd-Hewitt broth, transferred to a CBA plate the next day, and then the CBA plate incubated for 24-48 hours at 35 ℃ is placed into a photographing auxiliary device; 2. fixing photographing equipment (smart phone and the like) provided with GBS screening software on the top of a photographing auxiliary device; 3. adjusting light rays, angles and the like, and shooting clear colony pictures; 4. uploading the pictures to a cloud database through software; 5. dividing the picture at the cloud end, comparing the picture with a standard GBS colony picture in a database, and detecting; 6. the terminal obtains a detection result, namely whether GBS colonies are detected in the CBA plate or not; 7. reporting to the doctor/pregnant woman the GBS screening result.
The rise of artificial neural networks (Artificial Neural Network, ANN) has provided new possibilities for GBS screening, i.e. by colony image recognition of bacteria. As an important component of artificial intelligence (Artificial Intelligence) and Machine Learning (Learning), ANN has emerged in the last 50 th century, and recently, ANN has been increasingly applied to many fields of medicine with the development of big data and the great improvement of computer performance. In the field of clinical microbiological testing, the growth characteristics of bacterial colonies in a culture medium (including colony size, color, appearance, edge morphology, gloss, hemolysis rings, etc.) are important bases for classifying and identifying bacteria. The pregnant woman GBS is screened based on the ANN image recognition technology, has the advantages of low cost, simplicity and convenience in operation and high accuracy, a shooting auxiliary device is matched with a shooting device (a smart phone, an intelligent pad, a camera and the like), a clear colony picture is shot, then the colony picture is uploaded to a cloud database for comparison, and the GBS detection result can be obtained by the terminal in real time. The GBS screening system based on image recognition improves screening accuracy, objectivity and repeatability, does not need experienced technicians to perform primary screening, does not need expensive instruments and equipment to purchase, greatly reduces detection cost, has wide applicable crowd, and is particularly suitable for screening the GBS of the pregnant women in wide primary hospitals and women and young health care hospitals.
Although the present invention has been described in detail hereinabove, the present invention is not limited thereto and various modifications may be made by those skilled in the art in accordance with the principles of the present invention. Therefore, all modifications made in accordance with the principles of the present invention should be understood as falling within the scope of the present invention.
Claims (6)
1. A detection system for screening group B streptococcus based on image recognition technology, the system comprising:
a microbial colony culture dish in which a plurality of microbial colonies to be screened are grown;
the photographing auxiliary device is internally clamped with the microorganism colony culture dish and adjusts the angles of the light rays and the microorganism colony culture dish so as to achieve the optimal photographing effect;
the mobile terminal is clamped by the photographing auxiliary device, the angle of the mobile terminal can be changed through the photographing auxiliary device to achieve the optimal photographing effect, the photographing module photographs the microbial colony culture dish clamped in the photographing auxiliary device to obtain the microbial colony culture dish image with the optimal photographing effect and containing a plurality of microbial colony images to be screened, and the communication module sends the microbial colony culture dish image with the optimal photographing effect and containing the plurality of microbial colony images to be screened to the cloud analysis device;
the image recognition equipment of the cloud analysis device is used for screening out a part of bacterial colonies of non-group B streptococcus from a plurality of microbial colony images to be screened in the microbial colony culture dish image by adopting a threshold segmentation method according to the size, the color and the gradient of the color change from the center to the edge of each microbial colony image to obtain a plurality of microbial colony images of suspected group B streptococcus bacterial colonies and marking the plurality of microbial colony images of suspected group B streptococcus;
the colony analysis equipment of the cloud analysis device is used for analyzing the microbial colony image of each suspected group B streptococcus by utilizing an artificial neural network for detecting the group B streptococcus to obtain a detection result of whether the microbial colony image of each suspected group B streptococcus is the group B streptococcus colony image or not, so that the mobile terminal can acquire the detection result.
2. The system of claim 1, wherein the image recognition device is configured to screen a plurality of images of microbial colonies suspected of group B streptococcus colonies from the plurality of images of microbial colonies to be screened, and specifically comprises:
the bacterial colony characteristic analysis module is used for analyzing each microbial colony image to be screened in the microbial colony culture dish image to obtain a microbial colony characteristic value in each microbial colony image, wherein the microbial colony characteristic value comprises the size, the color and the color change gradient from the center to the edge of the microbial colony;
the group B streptococcus colony preliminary screening module is used for determining whether each microbial colony image is a microbial colony image of suspected group B streptococcus according to the microbial colony characteristic value in each microbial colony image and a preset group B streptococcus colony characteristic threshold range, wherein the preset group B streptococcus colony characteristic threshold range comprises a preset group B streptococcus colony size threshold range, a preset group B streptococcus colony color threshold range and a colony center-to-edge color change gradient threshold range.
3. The system of claim 2, wherein for each microbial colony, the image recognition device determines that each microbial colony image is a microbial colony image of suspected group B streptococcus when the size, color, and colony center-to-edge color change gradient of the microbial colony are within the colony size threshold range, colony color threshold range, colony center-to-edge color change gradient threshold range, respectively, of the preset group B streptococcus.
4. The system of claim 1, wherein the artificial neural network is a convolutional neural network comprising a convolutional layer, a modified linear cell layer, a pooling layer, and a fully-connected layer, and wherein the colony analysis device analyzes the number of microbial colony images of the suspected group B streptococcus marked by the image recognition device to determine whether the number of microbial colony images of the suspected group B streptococcus marked by the image recognition device are group B streptococcus colony images using the convolutional neural network comprising a convolutional layer, a modified linear cell layer, a pooling layer, and a fully-connected layer.
5. The system of claim 1, wherein the photographing assisting device comprises:
the box body is provided with a top plate, a bottom plate and side plates;
the supporting seat is arranged above the bottom plate;
the lamp housing of the bottom light source is arranged on the supporting seat and can rotate relative to the bottom plate;
the culture dish fixing clamp is arranged at the upper part of the lamp housing and is used for clamping the microorganism colony culture dish;
the movable arm is arranged above the top plate and can rotate or bend relative to the top plate;
and the photographing fixing clamp is arranged at the extending end of the movable arm and used for clamping the mobile terminal.
6. The system of claim 5, wherein the photographing assisting device further comprises: at least one of a plane scattering light source arranged on the lower surface of the top plate, an annular light source arranged on the inner wall of the side plate and an ultraviolet lamp disinfection device arranged on the inner wall of the side plate.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910915252.XA CN110675386B (en) | 2019-09-26 | 2019-09-26 | Detection system for group B streptococcus |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910915252.XA CN110675386B (en) | 2019-09-26 | 2019-09-26 | Detection system for group B streptococcus |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110675386A CN110675386A (en) | 2020-01-10 |
CN110675386B true CN110675386B (en) | 2023-08-18 |
Family
ID=69079083
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910915252.XA Active CN110675386B (en) | 2019-09-26 | 2019-09-26 | Detection system for group B streptococcus |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110675386B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111286523A (en) * | 2020-02-20 | 2020-06-16 | 中国水产科学研究院黄海水产研究所 | Three-primary-color identification and counting method for bacteria |
JP2022046265A (en) * | 2020-09-10 | 2022-03-23 | 合同会社H.U.グループ中央研究所 | Bacterial species identification support method, multi-colony learning model generation method, bacterial species identification support device and computer program |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101680021A (en) * | 2007-06-04 | 2010-03-24 | 株式会社N-Tech | Method for counting colonies |
CN105491279A (en) * | 2015-11-19 | 2016-04-13 | 北京工业大学 | Image acquisition, recognizing and counting method for colonies |
CN106367330A (en) * | 2016-11-19 | 2017-02-01 | 厦门大学 | Shooting identification device and method for morphology of microorganisms |
CN108520206A (en) * | 2018-03-22 | 2018-09-11 | 南京大学 | A kind of fungi microscopic image identification method based on full convolutional neural networks |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CA2985854C (en) * | 2015-04-23 | 2023-10-17 | Bd Kiestra B.V. | A method and system for automated microbial colony counting from streaked sample on plated media |
-
2019
- 2019-09-26 CN CN201910915252.XA patent/CN110675386B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101680021A (en) * | 2007-06-04 | 2010-03-24 | 株式会社N-Tech | Method for counting colonies |
CN105491279A (en) * | 2015-11-19 | 2016-04-13 | 北京工业大学 | Image acquisition, recognizing and counting method for colonies |
CN106367330A (en) * | 2016-11-19 | 2017-02-01 | 厦门大学 | Shooting identification device and method for morphology of microorganisms |
CN108520206A (en) * | 2018-03-22 | 2018-09-11 | 南京大学 | A kind of fungi microscopic image identification method based on full convolutional neural networks |
Non-Patent Citations (1)
Title |
---|
基于改进水平集的菌落图像智能计数算法;张力新等;天津大学学报(自然科学与工程技术版)(第01期);88-93 * |
Also Published As
Publication number | Publication date |
---|---|
CN110675386A (en) | 2020-01-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US8787633B2 (en) | System and method of organism identification | |
JP6777726B2 (en) | Colony contrast collection | |
CN103518224B (en) | Method for analysing microbial growth | |
CN110675386B (en) | Detection system for group B streptococcus | |
CN107580715A (en) | Method and system for automatic counting microbial colonies | |
CN113723573A (en) | Tumor tissue pathological classification system and method based on adaptive proportion learning | |
CN101933321A (en) | Image sensor apparatus and method for scene illuminant estimation | |
CN112215795A (en) | Intelligent server component detection method based on deep learning | |
CN106367330B (en) | A kind of microbial morphology shooting identification apparatus and method | |
CN111492065A (en) | System and method for identifying gram type of bacteria | |
CN113450317A (en) | Immunofluorescence image detection method for gynecological clinical microbial infection | |
CN111492064A (en) | Method for identifying yeast or bacteria | |
CN110702615B (en) | Color correction method for color digital transmission microscope | |
CN112241713A (en) | Vaginal microorganism identification method and device based on pattern identification and deep learning | |
JP2022514427A (en) | Systems and methods for monitoring bacterial growth of bacterial colonies and predicting colony biomass | |
CN112651305B (en) | Method for identifying microorganism species | |
CN110726536B (en) | Color correction method for color digital reflection microscope | |
Zhang et al. | An effective and robust method for automatic bacterial colony enumeration | |
US20230100268A1 (en) | Quantifying biotic damage on plant leaves, by convolutional neural networks | |
CN112461762B (en) | Solution turbidity detection method, medium and image processing system based on HSV model | |
CN115239766A (en) | Pig detection and tracking method and device with rapid adaptability | |
Kaur et al. | A novel methodology for automatic bacterial colony counter | |
CN116433568A (en) | LED lamp strip defect detection method and system based on depth support vector description | |
Rangappa et al. | Automatic counting and classification of silkworm eggs using deep learning | |
CN111435986A (en) | Method for acquiring source image database, training device and electronic equipment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |