CN112819057A - Automatic identification method of urinary sediment image - Google Patents
Automatic identification method of urinary sediment image Download PDFInfo
- Publication number
- CN112819057A CN112819057A CN202110100358.1A CN202110100358A CN112819057A CN 112819057 A CN112819057 A CN 112819057A CN 202110100358 A CN202110100358 A CN 202110100358A CN 112819057 A CN112819057 A CN 112819057A
- Authority
- CN
- China
- Prior art keywords
- urinary sediment
- neural network
- classification
- fine
- convolutional neural
- 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.)
- Pending
Links
- 239000013049 sediment Substances 0.000 title claims abstract description 39
- 230000002485 urinary effect Effects 0.000 title claims abstract description 38
- 238000000034 method Methods 0.000 title claims abstract description 18
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 16
- 238000013528 artificial neural network Methods 0.000 claims abstract description 8
- 210000002700 urine Anatomy 0.000 claims abstract description 8
- 238000002372 labelling Methods 0.000 claims description 5
- 230000003321 amplification Effects 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000003199 nucleic acid amplification method Methods 0.000 claims description 3
- 238000011176 pooling Methods 0.000 claims description 3
- 239000013598 vector Substances 0.000 claims description 3
- 230000015572 biosynthetic process Effects 0.000 claims 1
- 230000011218 segmentation Effects 0.000 abstract 1
- 210000003743 erythrocyte Anatomy 0.000 description 12
- 210000000265 leukocyte Anatomy 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 210000004027 cell Anatomy 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 210000004085 squamous epithelial cell Anatomy 0.000 description 2
- 238000005353 urine analysis Methods 0.000 description 2
- 241000894006 Bacteria Species 0.000 description 1
- 108010049047 Echinocandins Proteins 0.000 description 1
- 206010027540 Microcytosis Diseases 0.000 description 1
- 244000052616 bacterial pathogen Species 0.000 description 1
- 238000003759 clinical diagnosis Methods 0.000 description 1
- 210000002919 epithelial cell Anatomy 0.000 description 1
- 210000004276 hyalin Anatomy 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 210000003097 mucus Anatomy 0.000 description 1
- 230000001575 pathological effect Effects 0.000 description 1
- 208000014001 urinary system disease Diseases 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
Abstract
The invention relates to an automatic identification method of a urinary sediment image, belonging to the field of medical image processing. The method comprises the steps of obtaining an urine sample image by using a full-automatic urinary sediment analyzer, obtaining a formed partial image by means of segmentation, manually marking, forming a training set, constructing a convolutional neural network, inputting the training set into the convolutional neural network for training, obtaining a network model, carrying out classification and identification to obtain a coarse classification result, and carrying out fine classification on each coarse classification by using the trained neural network to obtain a fine classification result of the urinary sediment sample. The method has the advantages that the urinary sediment samples can be roughly classified, the characteristics of the same network are utilized, the SVM classifier is combined, the samples can be accurately classified finely, the classification result of the samples is improved, the classification complexity is reduced, the recognition speed is improved, and the requirement of the modern hospital for large sample quantity is met.
Description
Technical Field
The invention relates to the field of medical image processing, in particular to an automatic identification method of a urinary sediment image.
Background
The detection of urine visible components is one of routine examination items in hospitals, can help doctors to know the change of the urinary system of patients, and plays an important role in assisting the diagnosis and identification of urinary system diseases. There are many kinds of visible components (sediments) in urine, such as red blood cells, white blood cells, casts, epithelial cells, bacteria, etc., and there may be many other impurities, so the observation and manual identification under a conventional microscope have a large workload, complicated operation and easy error. The appearance of full-automatic urine analysis appearance for urine analysis realizes the automation, has alleviateed hospital's testing personnel's operating pressure greatly, has improved detection efficiency. Due to the complexity of the visible components in urine, it is a very urgent task to improve the recognition accuracy and the detection speed.
Disclosure of Invention
The invention provides an automatic identification method of a urinary sediment image, aiming at quickly and accurately classifying the urinary sediment image to contain fine categories so as to provide more diagnosis information for clinic and reduce the burden of a clinician.
The technical scheme adopted by the invention is as follows: comprises the following steps:
acquiring a urine sample image by using a full-automatic urinary sediment analyzer;
step two, segmenting a sample image shot by the full-automatic urinary sediment analyzer to obtain a formed partial image;
thirdly, carrying out amplification and reduction operations on the tangible component images to enable the images to be identical in size, and carrying out manual labeling to form a training set;
step four, constructing a convolutional neural network, inputting the training set into the convolutional neural network for training to obtain a network model;
step five, classifying and identifying the new visible component sample image by using a network model to obtain a coarse classification result of the urinary sediment sample;
and step six, using the feature maps of the last convolution layer of the trained neural network as feature vectors, using an SVM as a classifier, and performing fine classification on each coarse classification to obtain a fine classification result of the urinary sediment sample.
The full-automatic urinary sediment analyzer in the first step is a planar flow type urinary sediment analyzer;
the labeling method with formed partial images in the third step of the invention roughly classifies the visible components in the urinary sediment to form a large class, and the label is y ═ (alpha)1,α2,...,αn) Wherein n is the number of classes of the rough classification, the visible components can be further classified in a fine classification mode, the forms of the fine classifications are similar, in order to train in the same network, the fine classifications of the same large class are labeled, and the fine classification is labeled as alphaijWherein j is 1,2, and m is the number of categories of the fine category, and each alpha isijDifferent from each other, and each alpha is between 0 and 1ijThe values should differ from each other sufficiently to constitute a training set of the neural network;
the convolutional neural network of the fourth step of the invention is composed of a series of convolutional layers, the middle or sandwich pooling layer, the training set is used as input, and the error function adopts a cross entropy form:
the convolutional neural network adopts deep separable convolution to reduce the complexity of the network;
the convolutional neural network adopts a residual error structure and Batch Normalization to accelerate the training speed;
the method has the advantages that through the steps, the urinary sediment samples can be roughly classified, the samples can be accurately classified in a subdivided mode by utilizing the characteristics of the same network and combining an SVM classifier, the classification result of the samples is improved, the classification complexity is reduced, the identification speed is improved, and the requirement of a modern hospital for large sample quantity is met.
Drawings
FIG. 1 is a flow chart of the invention;
FIG. 2 is a picture of normal red blood cells;
FIG. 3 is a photograph of a erythrocyte;
FIG. 4 is a photograph of echinoid red blood cells;
FIG. 5 is a photograph of other heteromorphic red blood cells.
Detailed Description
Comprises the following steps:
acquiring a urine sample image by using a full-automatic urinary sediment analyzer, wherein the full-automatic urinary sediment analyzer is a planar flow urinary sediment analyzer;
step two, segmenting a sample image shot by the full-automatic urinary sediment analyzer to obtain a formed partial image;
thirdly, carrying out amplification and reduction operations on the tangible component images to enable the images to be identical in size, and carrying out manual labeling to form a training set; the method comprises the following steps:
roughly classifying the visible components in the urine sediment to form a large class labeled with y ═ alpha1,α2,...,αn) Wherein n is the number of classes of the rough classification, the visible components can be further classified in a fine classification mode, the forms of the fine classifications are similar, in order to train in the same network, the fine classifications of the same large class are labeled, and the fine classification is labeled as alphaijWherein j is 1,2, and m is the number of categories of the fine category, and each alpha isijDifferent from each other, and each alpha is between 0 and 1ijThe values should differ from each other sufficiently to constitute a training set of the neural network;
for example, the visible components in the urinary sediment are roughly classified into 10 major classes, i.e., erythrocytes, leukocytes, leukocyte aggregates, squamous epithelial cells, non-squamous epithelial cells, hyaline casts, pathological casts, bacilli, mucus silks, and sperms, and then the labels of erythrocytes are (1, 0, 0, 0, 0, 0, 0) and leukocytes are (0, 1, 0, 0, 0, 0, 0), and so on, and erythrocytes can be subdivided into normal erythrocytes, microcytosis cells, echinocandin cells, ghost erythrocytes, and other erythrocytes, and then the labels for subdividing erythrocytes can be: (0.2, 0, 0, 0, 0, 0), (0.4, 0, 0, 0, 0, 0, 0, 0, 0), (0.6, 0, 0, 0, 0, 0), (0.7, 0, 0, 0, 0, 0), (0.8, 0, 0, 0, 0, 0, 0, 0, 0) and a tube type fine classification label can be similarly arranged, and a label is made for a visible component according to a fine classification to obtain a training set;
step four, constructing a convolutional neural network, inputting the training set into the convolutional neural network for training to obtain a network model; the deep neural network is composed of a series of convolution layers, a pooling layer is arranged in the middle or in the middle, a training set is used as input, and an error function adopts a cross entropy form:
the convolutional neural network adopts deep separable convolution to reduce the complexity of the network;
the convolutional neural network adopts a residual error structure and Batch Normalization to accelerate the training speed;
step five, classifying and identifying the new visible component sample image by using a network model to obtain a coarse classification result of the urinary sediment sample;
and step six, using the feature maps of the last convolution layer of the trained neural network as feature vectors, using an SVM as a classifier, and performing fine classification on each coarse classification to obtain a fine classification result of the urinary sediment sample for reference of clinical diagnosis.
Claims (6)
1. An automatic identification method of urinary sediment images is characterized by comprising the following steps:
acquiring a urine sample image by using a full-automatic urinary sediment analyzer;
step two, segmenting a sample image shot by the full-automatic urinary sediment analyzer to obtain a formed partial image;
thirdly, carrying out amplification and reduction operations on the tangible component images to enable the images to be identical in size, and carrying out manual labeling to form a training set;
step four, constructing a convolutional neural network, inputting the training set into the convolutional neural network for training to obtain a network model;
step five, classifying and identifying the new visible component sample image by using a network model to obtain a coarse classification result of the urinary sediment sample;
and step six, using the feature maps of the last convolution layer of the trained neural network as feature vectors, using an SVM as a classifier, and performing fine classification on each coarse classification to obtain a fine classification result of the urinary sediment sample.
2. The method for automatically recognizing urinary sediment images according to claim 1, wherein the method comprises the following steps: the full-automatic urinary sediment analyzer in the first step is a planar flow type urinary sediment analyzer.
3. The method for automatically recognizing urinary sediment images according to claim 1, wherein the method comprises the following steps: the labeling method with partial image formation in the third step roughly classifies the visible components in the urinary sediment to form a large class, and the label is y ═ alpha1,α2,...,αn) Wherein n is the number of classes of the rough classification, the visible components can be further classified in a fine classification mode, the forms of the fine classifications are similar, in order to train in the same network, the fine classifications of the same large class are labeled, and the fine classification is labeled as alphaijWherein j is 1,2, and m is the number of categories of the fine category, and each alpha isijDifferent from each other, and each alpha is between 0 and 1ijThe values should differ from each other sufficiently to constitute a training set of the neural network.
4. The method for automatically recognizing urinary sediment images according to claim 1, wherein the method comprises the following steps: the convolutional neural network of the fourth step is composed of a series of convolutional layers, a pooling layer is arranged in the middle or in the middle, a training set is used as input, and an error function adopts a cross entropy form:
5. The method for automatically recognizing an urinary sediment image according to claim 1 or 4, wherein: the convolutional neural network adopts deep separable convolution to reduce the complexity of the network.
6. The method for automatically recognizing an urinary sediment image according to claim 1 or 4, wherein: the convolutional neural network adopts a residual error structure and Batch Normalization to accelerate the training speed.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110100358.1A CN112819057A (en) | 2021-01-25 | 2021-01-25 | Automatic identification method of urinary sediment image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110100358.1A CN112819057A (en) | 2021-01-25 | 2021-01-25 | Automatic identification method of urinary sediment image |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112819057A true CN112819057A (en) | 2021-05-18 |
Family
ID=75859211
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110100358.1A Pending CN112819057A (en) | 2021-01-25 | 2021-01-25 | Automatic identification method of urinary sediment image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112819057A (en) |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH10302067A (en) * | 1997-04-23 | 1998-11-13 | Hitachi Ltd | Pattern recognition device |
CN107330449A (en) * | 2017-06-13 | 2017-11-07 | 瑞达昇科技(大连)有限公司 | A kind of BDR sign detection method and device |
CN109447119A (en) * | 2018-09-26 | 2019-03-08 | 电子科技大学 | Cast recognition methods in the arena with SVM is cut in a kind of combining form credit |
CN110188592A (en) * | 2019-04-10 | 2019-08-30 | 西安电子科技大学 | A kind of urinary formed element cell image disaggregated model construction method and classification method |
CN110473167A (en) * | 2019-07-09 | 2019-11-19 | 哈尔滨工程大学 | A kind of urine sediment image identifying system and method based on deep learning |
CN111126455A (en) * | 2019-12-05 | 2020-05-08 | 北京科技大学 | Abrasive particle two-stage identification method based on Lightweight CNN and SVM |
CN111428807A (en) * | 2020-04-03 | 2020-07-17 | 桂林电子科技大学 | Image processing method and computer-readable storage medium |
-
2021
- 2021-01-25 CN CN202110100358.1A patent/CN112819057A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH10302067A (en) * | 1997-04-23 | 1998-11-13 | Hitachi Ltd | Pattern recognition device |
CN107330449A (en) * | 2017-06-13 | 2017-11-07 | 瑞达昇科技(大连)有限公司 | A kind of BDR sign detection method and device |
CN109447119A (en) * | 2018-09-26 | 2019-03-08 | 电子科技大学 | Cast recognition methods in the arena with SVM is cut in a kind of combining form credit |
CN110188592A (en) * | 2019-04-10 | 2019-08-30 | 西安电子科技大学 | A kind of urinary formed element cell image disaggregated model construction method and classification method |
CN110473167A (en) * | 2019-07-09 | 2019-11-19 | 哈尔滨工程大学 | A kind of urine sediment image identifying system and method based on deep learning |
CN111126455A (en) * | 2019-12-05 | 2020-05-08 | 北京科技大学 | Abrasive particle two-stage identification method based on Lightweight CNN and SVM |
CN111428807A (en) * | 2020-04-03 | 2020-07-17 | 桂林电子科技大学 | Image processing method and computer-readable storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Loddo et al. | Recent advances of malaria parasites detection systems based on mathematical morphology | |
CN113256637B (en) | Urine visible component detection method based on deep learning and context correlation | |
CN110473167B (en) | Deep learning-based urinary sediment image recognition system and method | |
Wu et al. | A hematologist-level deep learning algorithm (BMSNet) for assessing the morphologies of single nuclear balls in bone marrow smears: algorithm development | |
CN111079620B (en) | White blood cell image detection and identification model construction method and application based on transfer learning | |
JP2021517255A (en) | How to identify biological substances under a microscope | |
El-Melegy et al. | Identification of tuberculosis bacilli in ZN-stained sputum smear images: A deep learning approach | |
CN114945941A (en) | Non-tumor segmentation for supporting tumor detection and analysis | |
CN114494197A (en) | Cerebrospinal fluid cell identification and classification method for small-complexity sample | |
CN112348839A (en) | Image segmentation method and system based on deep learning | |
CN112784767A (en) | Cell example segmentation algorithm based on leukocyte microscopic image | |
Pellegrino et al. | Automated RBC morphology counting and grading using image processing and support vector machine | |
RU2308745C1 (en) | Method for microscopic examination of a sample which contains micro-objects with heterogeneous zones | |
CN112001315A (en) | Bone marrow cell classification and identification method based on transfer learning and image texture features | |
Evangeline et al. | Computer aided system for human blood cell identification, classification and counting | |
CN113902669A (en) | Method and system for reading urine exfoliative cell fluid-based smear | |
Mustafa et al. | Malaria parasite diagnosis using computational techniques: a comprehensive review | |
CN112036334A (en) | Method, system and terminal for classifying visible components in sample to be detected | |
Magpantay et al. | A transfer learning-based deep CNN approach for classification and diagnosis of acute lymphocytic leukemia cells | |
Khalid et al. | Urine sediment analysis by using convolution neural network | |
CN112819057A (en) | Automatic identification method of urinary sediment image | |
Ur Rahman et al. | Efficient segmentation of lymphoblast in acute lymphocytic leukemia | |
Kurniawan et al. | Segmentation of Tuberculosis Bacilli Using Watershed Transformation and Fuzzy C-Means | |
Vo et al. | A deep learning approach in detection of malaria and acute lymphoblastic leukemia diseases utilising blood smear microscopic images | |
CN112508909A (en) | Disease association method of peripheral blood cell morphology automatic detection system |
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 |