CN113793336B - Method, device and equipment for detecting blood cells and readable storage medium - Google Patents
Method, device and equipment for detecting blood cells and readable storage medium Download PDFInfo
- Publication number
- CN113793336B CN113793336B CN202111358569.1A CN202111358569A CN113793336B CN 113793336 B CN113793336 B CN 113793336B CN 202111358569 A CN202111358569 A CN 202111358569A CN 113793336 B CN113793336 B CN 113793336B
- Authority
- CN
- China
- Prior art keywords
- image
- blood cell
- blood cells
- cell image
- specified multiple
- 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
Images
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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- 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
- G06T2207/10061—Microscopic image from scanning electron microscope
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30024—Cell structures in vitro; Tissue sections in vitro
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Abstract
The invention provides a method, a device, equipment and a readable storage medium for detecting blood cells, which relate to the technical field of cell detection and comprise the steps of obtaining blood cell images; the blood cell image is obtained through an electron microscope, and the blood cell image comprises at least one group of red blood cells and white blood cells; judging whether the blood cell image is clear under a specified multiple; and respectively identifying the white blood cells and the red blood cells in the clear blood cell pattern under the specified multiple to obtain identification results, wherein the identification results comprise the identification results of the white blood cells and the identification results of the red blood cells. Compared with the traditional method, the method adds the K-Means clustering algorithm with image statistical characteristic prior to separate the intercellular space and the red blood cells, and the algorithm is insensitive to illumination and color tone and has more robustness in clustering.
Description
Technical Field
The invention relates to the technical field of cell detection, in particular to a method, a device, equipment and a readable storage medium for detecting blood cells.
Background
In the experimental examination for observing blood cells by using a microscope, the step of interactive answering exists, an examinee is required to observe a clear object image under a specified microscope multiple and capture the image, the positions of red blood cells and white blood cells are respectively marked according to the image capture, and whether the marking result is correct or not is judged. Therefore, an algorithm capable of automatically positioning various cell positions of a screenshot submitted by a test taker is required to be provided as a criterion of a result of interactive answer of the test taker.
However, whether the screenshot submitted by the examinee meets the requirement or not is judged, and the situations of fuzziness and times using errors are eliminated. The traditional image processing mode is HOG + fuzzy judgment: HOG mainly extracts image features and judges whether the image features accord with the features of images with specified multiples; the fuzzy judgment is mainly used for calculating and analyzing the image gradient, a corresponding threshold value is set, and the calculation result is greater than the corresponding threshold value, so that the image outline is clear and visible; if the threshold value is not exceeded, the image cannot detect obvious contours, and a certain degree of blurring exists.
For satisfactory screenshots, it is also necessary to distinguish red blood cells, white blood cells, and intercellular spaces in the image from each other. The traditional image processing method is often distinguished by colors, and the specific process is as follows: the image is converted from RGB color space to HSV color space, and the three are distinguished by the inherent difference of HSV color values among red blood cells, white blood cells and intercellular spaces.
Disclosure of Invention
It is an object of the present invention to provide a method, device, apparatus and readable storage medium for detecting blood cells that ameliorate the above problems. In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
in a first aspect, the present application provides a method of detecting blood cells, comprising:
obtaining a blood cell image; the blood cell image is obtained through an electron microscope, and the blood cell image comprises at least one group of red blood cells and white blood cells;
judging whether the blood cell image is clear under a specified multiple;
and respectively identifying the white blood cells and the red blood cells in the clear blood cell pattern under the specified multiple to obtain identification results, wherein the identification results comprise the identification results of the white blood cells and the identification results of the red blood cells.
Optionally, determining whether the blood cell image is clear at a specified fold comprises: acquiring an image definition judgment standard, wherein the image definition judgment standard comprises at least two definition classifications;
performing definition classification calculation on the blood cell image according to an image classification model to obtain a definition classification result;
judging whether the definition classification result is a preset definition classification type or not; if yes, the blood cell image is clear under a specified multiple; if not, the blood cell image is not clear at the indicated fold.
Optionally, obtaining the recognition result of the white blood cells comprises:
adjusting the blood cell image which is clear under the specified multiple to a preset image size;
and inputting the adjusted blood cell image into a target detection network to obtain at least one frame-shaped area, wherein the frame-shaped area comprises at least one leukocyte.
Optionally, obtaining the recognition result of the red blood cells comprises:
characterizing initial clustering centers of red blood cell portions in the blood cell images that are clear at the specified times by a first vector, and characterizing initial clustering centers of gap portions in the blood cell images that are clear at the specified times by a second vector, wherein the gap portions are portions of the blood cell images that do not include red blood cells and white blood cells;
inputting the first vector and the second vector into a k-means clustering model for calculation to obtain a clustering result image, wherein the clustering result image comprises two regions with different colors; the two areas with different colors are a red blood cell area and a gap part area respectively.
In a second aspect, the present application also provides a device for detecting blood cells, comprising: the device comprises an acquisition module, a judgment module and an identification module:
an acquisition module: for obtaining blood cell images; the blood cell image is obtained through an electron microscope, and the blood cell image comprises at least one group of red blood cells and white blood cells;
a judging module: used for judging whether the blood cell image is clear under a specified multiple;
an identification module: the blood cell recognition device is used for respectively recognizing the white blood cells and the red blood cells in the blood cell pattern which is clear under the specified multiple to obtain recognition results, and the recognition results comprise recognition results of the white blood cells and recognition results of the red blood cells.
Optionally, the determining module further includes an obtaining unit, a first calculating unit, and a determining unit:
an acquisition unit: the image definition judging standard comprises at least two definition classifications;
the first calculation unit: the blood cell image classification module is used for carrying out definition classification calculation on the blood cell image according to an image classification model to obtain a definition classification result;
a judging unit: the definition classification result is used for judging whether the definition classification result is a preset definition classification type or not; if yes, the blood cell image is clear under a specified multiple; if not, the blood cell image is not clear at the indicated fold.
Optionally, the identification module further includes an adjustment unit and a detection unit:
an adjusting unit: the blood cell image which is clear under the specified multiple is adjusted to a preset image size;
a detection unit: and the system is used for inputting the adjusted blood cell image into a target detection network to obtain at least one frame-shaped area, and the frame-shaped area comprises at least one leukocyte.
Optionally, the identification module further comprises a characterization unit and a second calculation unit:
a characterization unit: for characterizing an initial cluster center of a red blood cell portion in the blood cell image that is clear at a specified multiple by a first vector, and characterizing an initial cluster center of a gap portion in the blood cell image that is clear at a specified multiple by a second vector, the gap portion being a portion of the blood cell image that does not include red blood cells and white blood cells;
a second calculation unit: the first vector and the second vector are input into a k-means clustering model to be calculated to obtain a clustering result image, and the clustering result image comprises two regions with different colors; the two areas with different colors are a red blood cell area and a gap part area respectively.
In a third aspect, the present application also provides an apparatus for detecting blood cells, comprising:
a memory for storing a computer program;
a processor for carrying out the steps of the method for detecting blood cells when said computer program is executed.
In a fourth aspect, the present application also provides a readable storage medium having stored thereon a computer program which, when being executed by a processor, carries out the above-mentioned steps of the method based on detecting blood cells.
The invention has the beneficial effects that: compared with the traditional method, the convolutional neural network can give consideration to both the image feature extraction capability and the processing speed, and has more robustness on fuzzy detection; the target detection model is added to identify and position the leucocytes, and compared with the traditional identification method, the processing speed and the positioning precision are higher; compared with the traditional method, the K-Means with the prior contains image statistical characteristics, is insensitive to illumination and color tone, and has more robustness in clustering.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by the practice of the embodiments of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.
FIG. 1 is a schematic flow chart of a method for detecting blood cells according to an embodiment of the present invention;
FIG. 2 is a schematic structural diagram of an apparatus for detecting blood cells according to an embodiment of the present invention;
FIG. 3 is a schematic structural diagram of an apparatus for detecting blood cells according to an embodiment of the present invention.
In the figure, 701, an acquisition module; 702. a judgment module; 703. an identification module; 7021. an acquisition unit; 7022. a first calculation unit; 7023. a judgment unit; 7031. an adjustment unit; 7032. a detection unit; 7033. a characterization unit; 7034. a second calculation unit; 801. a processor; 802. a memory; 803. a multimedia component; 804. an input/output (I/O) interface; 805. a communication component.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures. Meanwhile, in the description of the present invention, the terms "first", "second", and the like are used only for distinguishing the description, and are not to be construed as indicating or implying relative importance.
The main disadvantages of the prior art are the following:
1. the expression capability of HOG feature extraction is limited, and the judgment capability of HOG is greatly influenced by illumination and variable color tone of an experimental environment;
2. the traditional fuzzy detection method has a good motion fuzzy detection effect, but due to the fact that a microscope is used in the experiment, focusing fuzzy situations which are not suitable for the traditional method exist;
3. HSV color detection is extremely sensitive to hue and illumination intensity and is not robust.
By adding the convolutional neural network, the image feature extraction capacity and the processing speed can be effectively improved, meanwhile, the fuzzy detection is more robust, the white blood cells are identified and positioned through the target detection model, and the processing speed and the positioning precision are higher. The cell gap and the red blood cells are separated through a K-Means clustering algorithm with image statistical characteristic prior, the influence of illumination and color tone is reduced, and clustering is more robust.
Example 1:
this example provides a method for detecting blood cells, which may be human blood cells or animal blood cells.
Referring to fig. 1, the method is shown to include step S100, step S200 and step S300.
S100, obtaining a blood cell image; the blood cell image is obtained by electron microscopy, and the blood cell image comprises at least one group of red blood cells and white blood cells.
It can be understood that, in this embodiment, a scene may be laid, for example, in the process of question interaction by an examinee, the examinee performs screenshot submission by continuously adjusting the standard focal spiral and the objective lens multiple until a clear object image meeting the required specified multiple is observed, so as to obtain a blood cell image.
And S200, judging whether the blood cell image is clear under the specified multiple.
It is understood that three steps S201, S202 and S203 are included in this step, which states:
s201, obtaining an image definition judgment standard, wherein the image definition judgment standard comprises at least two definition classifications.
In the invention, images can be classified through a mobilev2 model, and after the images belong to correct categories, subsequent processing operation is carried out; otherwise, judging that the image submitted by the examinee is not in accordance with the requirement, and not scoring.
The mobilev2 model is a convolutional neural network for image classification, and can extract features of a blood cell image by using a depth separable convolutional structure to obtain a first feature image, where the first feature image includes textures, colors, and shapes of at least one group of red blood cells and white blood cells in each region of the blood cell image;
and inputting the first characteristic image into a fully-connected neural network, and calculating to obtain an image definition classification result.
Specifically, 4 categories are established, namely "the object image is not in the center of the image", "the object image is blurred", "the object image is not under a specified multiple", "the object image is clear under a specified multiple", the mobilev2 network identifies the blood cell image and then outputs the blood cell image, the probabilities of the 4 categories (the sum of the probabilities is 1), and the category with the highest probability is used as the output category.
It should be understood that the mobilev2 model used in one embodiment of the present invention is one of the image classification models, and the reason for selecting mobilev2 is that the processing speed is better under the same precision, but the processing of this step is not limited to mobilev2, and similar other image classification networks, such as ResNet, densnet and GoogleNet, can achieve the corresponding effects, and the process is not described again.
S202, performing definition classification calculation on the blood cell image according to the image classification model to obtain a definition classification result.
Wherein, the definition classification calculation to obtain the definition classification result comprises:
extracting image features in the blood cell image using an image classification model;
inputting the image characteristics into a fully-connected neural network to obtain the probability of each definition classification of the blood cell image;
and comparing the probability value of each definition classification to obtain a definition classification result, wherein the definition classification result is the definition classification with the highest definition classification probability value.
S203, judging whether the definition classification result is a preset definition classification type or not; if so, the blood cell image is clear under the specified multiple; if not, the blood cell image is not clear at the indicated fold.
It should be noted that if the category is clear at a given multiple, the category is identified and output, if the category is not clear, the category with the highest probability is output.
S300, respectively identifying the white blood cells and the red blood cells in the clear blood cell pattern under the specified multiple to obtain identification results, wherein the identification results comprise identification results of the white blood cells and identification results of the red blood cells.
After obtaining the identification result, the method further comprises the following steps: acquiring cell marker information in a blood cell image, wherein the cell marker information comprises marker information of white blood cells and marker information of red blood cells; and comparing the cell marking information with the identification result to judge whether the cell marking information is correct.
It is understood that in this step, there are two steps, S301 and S302, respectively.
Wherein, S301 is:
adjusting the clear blood cell image under the specified multiple to a preset image size;
and inputting the adjusted blood cell image into a target detection network to obtain at least one frame-shaped area, wherein the frame-shaped area comprises at least one leukocyte.
In this embodiment, the features of the blood cell image are extracted according to a Yolov4 model to obtain a second feature image, where the second feature image includes selection information of a trunk network, a non-maximum suppression threshold and prior frame information;
it should be noted that, yolo (young Only Look one) is a neural network target detection algorithm based on anchor, and training is performed by using methods such as CutMix, Mosaic, Self-adaptive, and the like. Specifically, the image may be adjusted to a preset size, for example, 416 × 416, and then input into a well-trained Yolov4 target detection model, so as to obtain a rectangular frame with boundaries x1, y1, x2, and y2, and then the region surrounded by the rectangular frame is regarded as a leukocyte region.
It should be understood that Yolov4 used in one embodiment of the invention is only one of the target detection models, and the target detection algorithm can be classified into one-stage algorithm, such as Yolo algorithm, SSD algorithm, etc., and two-stage algorithm, such as R-CNN algorithm, etc. In multiple tests of an actual environment, Yolov4 is enough to cope with complex scene detection in a physical experiment, and specific parameters such as selection of a feature extraction backbone network, a threshold value of non-maximum suppression, a prior frame and the like in Yolov4 can be adjusted to an ideal value by an experimenter, and no specific requirement is made here.
And obtaining white blood cell areas through calculation according to the second characteristic image, and marking each white blood cell area.
Wherein, S302 is:
characterizing an initial clustering center of a red blood cell part in the blood cell image which is clear at a specified multiple by using a first vector, and characterizing an initial clustering center of a gap part in the blood cell image which is clear at the specified multiple by using a second vector, wherein the gap part is a part which does not comprise red blood cells and white blood cells in the blood cell image;
inputting the first vector and the second vector into a k-means clustering model for calculation to obtain a clustering result image, wherein the clustering result image comprises two regions with different colors; the two areas of different colors are the red blood cell area and the gap part area respectively.
In this embodiment, before clustering, some a priori statistics are calculated, and the calculation formula is as follows:
in the formula (I), the compound is shown in the specification,
representing an image to be processed;
respectively representing the width and height of the image; image of a person
Is a three-dimensional tensor in which
Representing an image having a width of
Has a height of
The channel is
Pixel value () of (d);
respectively calculating a maximum value function and a minimum value function;
respectively count the images
A maximum value of a channel pixel;
respectively count the images
A minimum value of a channel pixel; wherein the content of the first and second substances,
vectors formed for the previous statistics respectively characterize the initial cluster centers of the interstitial portions in the clear blood cell images at the specified multiples and the initial cluster centers of the red blood cell portions in the clear blood cell images at the specified multiples, i.e., the first vector and the second vector.
Bringing the first vector and the second vector into a K-Means algorithm to serve as initialization conditions in the algorithm, and then performing K-Means algorithm separation to obtain a clustering result image, wherein the clustering result image comprises two regions with different colors; the two areas of different colors are the red blood cell area and the gap part area respectively.
In the present embodiment, the red blood cell region and the gap portion region are distinguished by two different colors, wherein the white portion is the gap portion region and the black portion is the red blood cell region.
Finally, in a scene embodiment, the label of the examinee can be obtained, and whether the label type (red blood cell or white blood cell) is correct or not can be checked by comparing the label with the specified label position; if the result is correct, scoring and ending; if not, the score is not obtained, and the operation is finished.
Example 2:
as shown in fig. 2, the present embodiment provides an apparatus for detecting blood cells, and the apparatus shown in fig. 2 includes an obtaining module 701, a determining module 702, and an identifying module 703:
an acquisition module 701: for obtaining blood cell images; the blood cell image is obtained by electron microscopy, and the blood cell image comprises at least one group of red blood cells and white blood cells.
The mobilev2 model is a convolutional neural network for image classification, and can extract features of a blood cell image by using a depth separable convolutional structure to obtain a first feature image, where the first feature image includes textures, colors, and shapes of at least one group of red blood cells and white blood cells in each region of the blood cell image;
and inputting the first characteristic image into a fully-connected neural network, and calculating to obtain an image definition classification result.
Specifically, 4 categories are established, namely "the object image is not in the center of the image", "the object image is blurred", "the object image is not under a specified multiple", "the object image is clear under a specified multiple", the mobilev2 network identifies the blood cell image and then outputs the blood cell image, the probabilities of the 4 categories (the sum of the probabilities is 1), and the category with the highest probability is used as the output category.
It should be understood that the mobilev2 model used in one embodiment of the present invention is one of the image classification models, and the reason for selecting mobilev2 is that the processing speed is better under the same precision, but the processing of this step is not limited to mobilev2, and similar other image classification networks, such as ResNet, densnet and GoogleNet, can achieve the corresponding effects, and the process is not described again.
A judging module 702: used for judging whether the blood cell image is clear under the specified multiple.
Preferably, the determining module 702 further includes an obtaining unit 7021, a first calculating unit 7022, and a determining unit 7023;
wherein, obtaining unit 7021: the image definition judging standard comprises at least two definition classifications;
first calculation unit 7022: the system is used for carrying out definition classification calculation on the blood cell image according to the image classification model to obtain a definition classification result;
the method comprises the steps of extracting image features in blood cell images by using an image classification model; inputting the image characteristics into a fully-connected neural network to obtain the probability of each definition classification of the blood cell image; and comparing the probability value of each definition classification to obtain a definition classification result, wherein the definition classification result is the definition classification with the highest definition classification probability value.
Wherein, the definition classification calculation to obtain the definition classification result comprises: extracting image features in the blood cell image using an image classification model; inputting the image characteristics into a fully-connected neural network to obtain the probability of each definition classification of the blood cell image; and comparing the probability value of each definition classification to obtain a definition classification result, wherein the definition classification result is the definition classification with the highest definition classification probability value.
Determination unit 7023: the definition classification result is used for judging whether the definition classification result is a preset definition classification type or not; if so, the blood cell image is clear under the specified multiple; if not, the blood cell image is not clear at the indicated fold.
The identification module 703: the method is used for respectively identifying the white blood cells and the red blood cells in the blood cell pattern which is clear under the specified times to obtain identification results, wherein the identification results comprise the identification results of the white blood cells and the identification results of the red blood cells.
Preferably, the recognition module 703 further includes an adjusting unit 7031, a detecting unit 7032, a characterizing unit 7033, and a second calculating unit 7034:
wherein, adjusting unit 7031: the blood cell image processing device is used for adjusting the clear blood cell image under the specified multiple to a preset image size;
in this embodiment, the features of the blood cell image are extracted according to a Yolov4 model to obtain a second feature image, where the second feature image includes selection information of a trunk network, a non-maximum suppression threshold and prior frame information;
detection unit 7032: the system is used for inputting the adjusted blood cell image into a target detection network to obtain at least one frame-shaped area, and the frame-shaped area comprises at least one leukocyte;
characterization unit 7033: the method comprises the steps of characterizing an initial clustering center of a red blood cell part in a blood cell image which is clear at a specified multiple by using a first vector, and characterizing an initial clustering center of a gap part in the blood cell image which is clear at the specified multiple by using a second vector, wherein the gap part is a part which does not comprise red blood cells and white blood cells in the blood cell image;
second calculation unit 7034: the system comprises a k-means clustering model, a first vector, a second vector, a first color image and a second color image, wherein the first vector and the second vector are input into the k-means clustering model to be calculated to obtain a clustering result image; the two areas of different colors are the red blood cell area and the gap part area respectively.
In this embodiment, prior to clustering, a few a priori statistics are calculated, where V _1 and V _2 are vectors formed by the previous statistics, and respectively represent the initial cluster center of the gap portion in the clear blood cell image at a given multiple and the initial cluster center of the red blood cell portion in the clear blood cell image at a given multiple, that is, the first vector and the second vector.
Bringing the first vector and the second vector into a K-Means algorithm to serve as initialization conditions in the algorithm, and then performing K-Means algorithm separation to obtain a clustering result image, wherein the clustering result image comprises two regions with different colors; the two areas of different colors are the red blood cell area and the gap part area respectively.
In the present embodiment, the red blood cell region and the gap portion region are distinguished by two different colors, wherein the white portion is the gap portion region and the black portion is the red blood cell region.
It should be noted that, regarding the apparatus in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated herein.
Example 3:
in accordance with the above method embodiment, a blood cell detecting apparatus is also provided in this embodiment, and a blood cell detecting apparatus described below and a blood cell detecting method described above may be referred to in correspondence with each other.
Fig. 3 is a block diagram illustrating an apparatus 800 for detecting blood cells according to an exemplary embodiment. As shown in fig. 3, the apparatus for detecting blood cells 800 may include: a processor 801, a memory 802. The apparatus for detecting blood cells 800 may further comprise one or more of a multimedia component 803, an I/O interface 804, and a communication component 805.
The processor 801 is configured to control the overall operation of the blood cell detecting apparatus 800, so as to complete all or part of the steps of the blood cell detecting method. The memory 802 is used to store various types of data to support the operation of the blood cell detection device 800, such data may include, for example, instructions for any application or method operating on the blood cell detection device 800, as well as application-related data, such as contact data, transmitted and received messages, pictures, audio, video, and so forth. The Memory 802 may be implemented by any type of volatile or non-volatile Memory device or combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic Memory, flash Memory, magnetic disk or optical disk. The multimedia components 803 may include screen and audio components. Wherein the screen may be, for example, a touch screen and the audio component is used for outputting and/or inputting audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may further be stored in the memory 802 or transmitted through the communication component 805. The audio assembly also includes at least one speaker for outputting audio signals. The I/O interface 804 provides an interface between the processor 801 and other interface modules, such as a keyboard, mouse, buttons, etc. These buttons may be virtual buttons or physical buttons. The communication module 805 is used for wired or wireless communication between the blood cell detecting apparatus 800 and other apparatuses. Wireless communication, such as Wi-Fi, bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination of one or more of them, so that the corresponding communication component 805 may include: Wi-Fi module, bluetooth module, NFC module.
In an exemplary embodiment, the blood cell detecting apparatus 800 may be implemented by one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components for performing the above blood cell detecting method.
In another exemplary embodiment, a computer readable storage medium comprising program instructions which, when executed by a processor, carry out the steps of the method of detecting blood cells described above is also provided. For example, the computer readable storage medium may be the memory 802 described above comprising program instructions executable by the processor 801 of the blood cell detection apparatus 800 to perform the method for detecting blood cells described above.
Example 4:
in accordance with the above method embodiment, a readable storage medium is also provided, and a readable storage medium described below and a method for detecting blood cells described above are referred to with respect to each other.
A readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method of detecting blood cells of the above-mentioned method embodiment.
The readable storage medium may be a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and various other readable storage media capable of storing program codes.
In summary, the invention firstly uses the lightweight neural network with stronger feature extraction capability to classify the intercepted images, and the related categories include that the observed object is not in the center of the visual field, the microscope multiple is wrong, the object image is fuzzy and the object image is clear. By the method, three problems of processing speed, feature extraction capability and fuzzy detection robustness can be considered; the invention firstly counts the color distribution condition in the image, then uses the prior distribution as an initialization condition, and then uses Kmeans clustering to segment the red blood cells from the intercellular spaces. And finally, combining the clustering result and the target detection result together to form a final cell segmentation result.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (8)
1. A method of detecting blood cells, comprising:
obtaining a blood cell image; the blood cell image is obtained through an electron microscope, and the blood cell image comprises at least one group of red blood cells and white blood cells;
judging whether the blood cell image is clear under a specified multiple;
respectively identifying white blood cells and red blood cells in the blood cell pattern which is clear under the specified multiple to obtain identification results, wherein the identification results comprise identification results of the white blood cells and identification results of the red blood cells;
judging whether the blood cell image is clear under a specified multiple, comprising: acquiring an image definition judgment standard, wherein the image definition judgment standard comprises at least two definition classifications;
performing definition classification calculation on the blood cell image according to an image classification model to obtain a definition classification result;
judging whether the definition classification result is a preset definition classification type or not; if yes, the blood cell image is clear under a specified multiple; if not, the blood cell image is not clear under the specified multiple;
classifying the images through a mobilev2 model, and performing subsequent processing operation after the images belong to correct categories; otherwise, judging that the image does not meet the requirements; the mobilev2 model is a convolutional neural network for image classification, and features of the blood cell image are extracted by using a depth separable convolutional structure to obtain a first feature image, wherein the first feature image comprises textures, colors and shapes of at least one group of red blood cells and white blood cells in each region of the blood cell image;
inputting the first characteristic image into a fully-connected neural network, and calculating to obtain an image definition classification result;
4 categories are established, namely 'the object image is not in the center of the image', 'the object image is fuzzy', 'the object image is not under a specified multiple', 'the object image is clear under a specified multiple', the probability of 4 categories is output after the blood cell image is identified by the mobilev2 network, and the category with the highest probability is used as the output category.
2. The method for detecting blood cells according to claim 1, wherein obtaining the recognition result of the white blood cells includes:
adjusting the blood cell image which is clear under the specified multiple to a preset image size;
and inputting the adjusted blood cell image into a target detection network to obtain at least one frame-shaped area, wherein the frame-shaped area comprises at least one leukocyte.
3. The method for detecting blood cells according to claim 1, wherein obtaining the result of identifying the red blood cells comprises:
characterizing initial clustering centers of red blood cell parts in the blood cell images which are clear at the specified times by using a first vector, and characterizing initial clustering centers of gap parts in the blood cell images which are clear at the specified times by using a second vector, wherein the gap parts are parts which do not comprise red blood cells and white blood cells in the blood cell images, and the first vector and the second vector are both a priori statistics with image statistical characteristics;
inputting the first vector and the second vector into a k-means clustering model for calculation to obtain a clustering result image, wherein the clustering result image comprises two regions with different colors; the two areas with different colors are a red blood cell area and a gap part area respectively.
4. A device for detecting blood cells, comprising:
an acquisition module: for obtaining blood cell images; the blood cell image is obtained through an electron microscope, and the blood cell image comprises at least one group of red blood cells and white blood cells;
a judging module: used for judging whether the blood cell image is clear under a specified multiple;
the judging module further comprises:
an acquisition unit: the image definition judging standard comprises at least two definition classifications;
the first calculation unit: the blood cell image classification module is used for carrying out definition classification calculation on the blood cell image according to an image classification model to obtain a definition classification result;
a judging unit: the definition classification result is used for judging whether the definition classification result is a preset definition classification type or not; if yes, the blood cell image is clear under a specified multiple; if not, the blood cell image is not clear under the specified multiple;
an identification module: the recognition device is used for respectively recognizing the white blood cells and the red blood cells in the blood cell pattern which is clear under the specified multiple to obtain recognition results, wherein the recognition results comprise recognition results of the white blood cells and recognition results of the red blood cells;
the first computing unit is also used for classifying the images through a mobilev2 model, and performing subsequent processing operation after the images belong to correct categories; otherwise, judging that the image does not meet the requirements; the mobilev2 model is a convolutional neural network for image classification, and features of the blood cell image are extracted by using a depth separable convolutional structure to obtain a first feature image, wherein the first feature image comprises textures, colors and shapes of at least one group of red blood cells and white blood cells in each region of the blood cell image;
inputting the first characteristic image into a fully-connected neural network, and calculating to obtain an image definition classification result;
4 categories are established, namely 'the object image is not in the center of the image', 'the object image is fuzzy', 'the object image is not under a specified multiple', 'the object image is clear under a specified multiple', the probability of 4 categories is output after the blood cell image is identified by the mobilev2 network, and the category with the highest probability is used as the output category.
5. The apparatus for detecting blood cells according to claim 4, wherein the identification module further comprises:
an adjusting unit: the blood cell image which is clear under the specified multiple is adjusted to a preset image size;
a detection unit: and the system is used for inputting the adjusted blood cell image into a target detection network to obtain at least one frame-shaped area, and the frame-shaped area comprises at least one leukocyte.
6. The apparatus for detecting blood cells according to claim 4, wherein the identification module further comprises:
a characterization unit: the method comprises the steps of characterizing an initial clustering center of a red blood cell part in a blood cell image which is clear at a specified multiple by a first vector, and characterizing an initial clustering center of a gap part in the blood cell image which is clear at the specified multiple by a second vector, wherein the gap part is a part which does not comprise red blood cells and white blood cells in the blood cell image, and the first vector and the second vector are both a priori statistics with image statistical characteristics;
a second calculation unit: the first vector and the second vector are input into a k-means clustering model to be calculated to obtain a clustering result image, and the clustering result image comprises two regions with different colors; the two areas with different colors are a red blood cell area and a gap part area respectively.
7. An apparatus for detecting blood cells, comprising:
a memory for storing a computer program;
a processor for implementing the steps of the method for detecting blood cells according to any one of claims 1 to 3 when executing said computer program.
8. A readable storage medium, characterized by: the readable storage medium has stored thereon a computer program which, when being executed by a processor, carries out the steps of the method for detecting blood cells according to any one of claims 1 to 3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111358569.1A CN113793336B (en) | 2021-11-17 | 2021-11-17 | Method, device and equipment for detecting blood cells and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111358569.1A CN113793336B (en) | 2021-11-17 | 2021-11-17 | Method, device and equipment for detecting blood cells and readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113793336A CN113793336A (en) | 2021-12-14 |
| CN113793336B true CN113793336B (en) | 2022-03-11 |
Family
ID=78877359
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111358569.1A Active CN113793336B (en) | 2021-11-17 | 2021-11-17 | Method, device and equipment for detecting blood cells and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113793336B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114881541B (en) * | 2022-07-06 | 2022-11-11 | 成都西交智汇大数据科技有限公司 | Evaluation method, device and equipment for examinee score and readable storage medium |
| CN115019305B (en) * | 2022-08-08 | 2022-11-11 | 成都西交智汇大数据科技有限公司 | Method, device and equipment for identifying root tip cells and readable storage medium |
| CN115393846B (en) * | 2022-10-28 | 2023-03-03 | 成都西交智汇大数据科技有限公司 | Blood cell identification method, device, equipment and readable storage medium |
| CN116630294B (en) * | 2023-06-08 | 2023-12-05 | 南方医科大学南方医院 | Whole blood sample detection method and device based on deep learning and storage medium |
| CN116503859A (en) * | 2023-06-27 | 2023-07-28 | 成都云芯医联科技有限公司 | Data enhancement three-classification leukemia algorithm based on deep learning |
| CN117523649B (en) * | 2024-01-04 | 2024-03-15 | 成都科瑞特电气自动化有限公司 | Mining iris safety recognition detection method, system, medium and terminal |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE102007013971A1 (en) * | 2007-03-23 | 2008-09-25 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Method and device for determining a cell contour of a cell |
| CN110647875A (en) * | 2019-11-28 | 2020-01-03 | 北京小蝇科技有限责任公司 | Method for segmenting and identifying model structure of blood cells and blood cell identification method |
| CN111656247A (en) * | 2018-07-13 | 2020-09-11 | 深圳迈瑞生物医疗电子股份有限公司 | Cell image processing system, cell image processing method, automatic film reading device and storage medium |
| CN111882521A (en) * | 2020-06-21 | 2020-11-03 | 浙江大学山东工业技术研究院 | Image processing method of cell smear |
| CN112419295A (en) * | 2020-12-03 | 2021-02-26 | 腾讯科技（深圳）有限公司 | Medical image processing method, apparatus, computer device and storage medium |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2425241A4 (en) * | 2009-04-27 | 2015-05-13 | Abbott Lab | Method for discriminating red blood cells from white blood cells by using forward scattering from a laser in an automated hematology analyzer |
| CN110223290A (en) * | 2019-06-19 | 2019-09-10 | 深圳市亚辉龙生物科技股份有限公司 | Film appraisal procedure, device, computer equipment and storage medium |
| CN110647874B (en) * | 2019-11-28 | 2020-08-28 | 北京小蝇科技有限责任公司 | End-to-end blood cell identification model construction method and application |
| CN111259957A (en) * | 2020-01-15 | 2020-06-09 | 上海眼控科技股份有限公司 | Visibility monitoring and model training method, device, terminal and medium based on deep learning |
| CN113241154A (en) * | 2020-12-28 | 2021-08-10 | 中国人民解放军陆军军医大学第二附属医院 | Artificial intelligent blood smear cell labeling system and method |
-
2021
- 2021-11-17 CN CN202111358569.1A patent/CN113793336B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE102007013971A1 (en) * | 2007-03-23 | 2008-09-25 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Method and device for determining a cell contour of a cell |
| CN111656247A (en) * | 2018-07-13 | 2020-09-11 | 深圳迈瑞生物医疗电子股份有限公司 | Cell image processing system, cell image processing method, automatic film reading device and storage medium |
| CN110647875A (en) * | 2019-11-28 | 2020-01-03 | 北京小蝇科技有限责任公司 | Method for segmenting and identifying model structure of blood cells and blood cell identification method |
| CN111882521A (en) * | 2020-06-21 | 2020-11-03 | 浙江大学山东工业技术研究院 | Image processing method of cell smear |
| CN112419295A (en) * | 2020-12-03 | 2021-02-26 | 腾讯科技（深圳）有限公司 | Medical image processing method, apparatus, computer device and storage medium |
Non-Patent Citations (3)
| Title |
|---|
| A fast and yet efficient YOLOv3 for blood cell detection;Ashkan Shakarami等;《Biomedical Signal Processing and Control》;20210216;第66卷;1-11页 * |
| Colour Based Segmentation of Red Blood Cells using K-means and Image Morphological Operations;Faten F Abushmmala等;《International Journal of Advanced and Innovative Research》;20131130;第2卷(第11期);第344-350页 * |
| 前列腺病理切片图像中对象的分割及其特征提取;李敏等;《中国优秀博硕士学位论文全文数据库(硕士) 信息科技辑》;20130215(第02期);正文第7-8页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113793336A (en) | 2021-12-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113793336B (en) | Method, device and equipment for detecting blood cells and readable storage medium | |
| US11681418B2 (en) | Multi-sample whole slide image processing in digital pathology via multi-resolution registration and machine learning | |
| CN109697416B (en) | Video data processing method and related device | |
| CN109389129B (en) | Image processing method, electronic device and storage medium | |
| CN110889312B (en) | Living body detection method and apparatus, electronic device, computer-readable storage medium | |
| CN112052186B (en) | Target detection method, device, equipment and storage medium | |
| KR101896357B1 (en) | Method, device and program for detecting an object | |
| CN112329619B (en) | Face recognition method and device, electronic equipment and readable storage medium | |
| US10007834B2 (en) | Detection control device, detection system, non-transitory storage medium, and detection control method | |
| JP5413501B1 (en) | Image processing apparatus, image processing system, and program | |
| CN114241548A (en) | Small target detection algorithm based on improved YOLOv5 | |
| CN112381104A (en) | Image identification method and device, computer equipment and storage medium | |
| CN110443184A (en) | ID card information extracting method, device and computer storage medium | |
| CN111382791B (en) | Deep learning task processing method, image recognition task processing method and device | |
| CN113947613B (en) | Target area detection method, device, equipment and storage medium | |
| KR101672814B1 (en) | Method for recognizing gender using random forest | |
| KR20130128097A (en) | Object recognition method and apparatus using depth information | |
| CN114037868B (en) | Image recognition model generation method and device | |
| US10146042B2 (en) | Image processing apparatus, storage medium, and image processing method | |
| CN115019305A (en) | Method, device and equipment for identifying root tip cells and readable storage medium | |
| CN114882511A (en) | Handwritten Chinese character recognition method, system, equipment and storage medium based on AMNN and Chinese character structure dictionary | |
| CN114758384A (en) | Face detection method, device, equipment and storage medium | |
| CN114565918A (en) | Face silence living body detection method and system based on multi-feature extraction module | |
| CN107992853B (en) | Human eye detection method and device, computer equipment and storage medium | |
| CN113837236A (en) | Method and device for identifying target object in image, terminal equipment and storage medium |
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 |