CN101750272A - Blood cell image recognition counting method - Google Patents

Abstract

The invention relates to a blood cell image recognition and counting method realized by computer recognition and statistics. The method comprises the following steps: 1) collecting blood samples, quantifying, diluting, and staining; Zoom in, focus, and image; 3) Image acquisition, smoothing, denoising, enhancement, edge sharpening, and image element segmentation processing; 4) Pattern recognition for each element image after segmentation, judging its nature, and performing classification statistics , calculate the relevant data of various blood components; display, print, store and transmit the test results. The advantages are: high work efficiency, accurate judgment and statistics, and can exclude the influence of human factors.

Description

Blood cell image recognition and counting method

technical field

The invention relates to a blood cell image recognition and counting method realized by computer recognition and statistics.

Background technique

In the field of modern clinical laboratory medicine, blood routine examination is one of the most basic and important inspection items. The purpose of this examination is to understand the number, shape, ratio and other data of various cells and important components in the blood, as an important basis for clinicians to make a diagnosis.

Existing blood routine examination mainly contains two kinds of methods---manual method and machine method.

1. Manual method, which is manually operated by inspectors to quantify, dilute, and stain the tested blood samples, and then observe, judge, and count with the naked eye under a microscope. The advantage of this method is that the counting accuracy is high, and it is suitable for clinical diagnosis with high requirements; the disadvantage is that the speed is slow, the efficiency is low, the blood consumption is large, and it is easily disturbed by human factors. Now this method is rarely used, and it is only used in two situations. One is to use it as a re-examination method when there is doubt about the test results of the machine method; the other is to diagnose special diseases such as tumors that require high accuracy of test results. used when.

2. The machine method is to use a special machine to automatically or semi-automatically quantify, dilute, and stain the blood sample, and then use the Coulter principle (electrical impedance method), photoelectric colorimetry principle, laser diffraction principle, sheath flow Technology and other technical means to count and calculate, and output the detection results. The advantages of this method are fast speed, high efficiency, less blood consumption, and can avoid interference from human factors; the disadvantage is that the detection accuracy is low, and it is only suitable for general screening.

Contents of the invention

The purpose of the present invention is to provide a blood cell image recognition and counting method with high work efficiency, accurate judgment and statistics, and can exclude the influence of human factors.

The present invention is realized through the following technical solutions:

A blood cell image recognition and counting method, the method comprising the steps of:

1) Collect blood samples, quantify, dilute, and stain;

2) Enlarge, focus, and image the blood sample to be tested according to the predetermined magnification through the optical instrument;

3) Image acquisition, smoothing, denoising, enhancement, edge sharpening, and image element segmentation processing;

4) Carry out pattern recognition to each element image after segmentation, judge its nature, and carry out classification statistics, calculate the relevant data of various blood components;

5) Display, print, store and transmit the test results.

The blood sample collection, quantification, dilution, and staining are all automatically completed by the sampling component.

The image acquisition, smoothing, denoising, enhancement, edge sharpening, and image element segmentation processing are completed through pre-programmed computer software, and image processing is performed on each acquisition target to meet the feature requirements for recognition and judgment.

The pattern recognition of each element image after segmentation is carried out, its nature is judged, and classification and statistics are carried out, and the relevant data of various blood components are calculated through comparison, judgment, and statistics. By establishing a database, storing each Collect the features of the image, and then compare each image element after collection and processing with the image features in the database to achieve statistical results.

The target image recognition feature is distinguished by two elements: the color difference and the different shape of the cells after dyeing, and the color after dyeing and the shape of the cells are used as the recognition basis for distinguishing different types of cells.

The staining agent is Wright's stain.

Compared with the prior art, the advantages of the present invention are: this method not only has the advantages of fast detection by traditional machine methods, high efficiency, less blood consumption, and can avoid interference from human factors, but also can achieve the accuracy of detection by manual methods. , to provide accurate and rapid blood routine examination results for clinical diagnosis.

Detailed ways

A blood cell image recognition and counting method, the method comprising the steps of:

1) Collect blood samples, quantify, dilute, and stain—use constant negative pressure to draw a small amount of blood samples from blood collection test tubes, use bubble quantitative method to accurately extract quantitative blood samples, inject quantitative diluent and Wright's staining solution for sample dilution Mix well in the cup, inject the diluted, stained and mixed blood sample into the blood counting plate under the optical assembly (microscope);

2) Enlarge, focus, and image the blood sample to be tested according to the predetermined magnification through the optical instrument;

3) Image collection and processing—use CMOS camera components to collect the optically amplified image of the blood cell sample, and transmit it to the computer in BMP format to form an image file, and use a 5×5 window to process the image pixels in the image file The two-dimensional data matrices of the three color components of R, G, and B are subjected to median filtering to eliminate noise interference, and the image enhancement and edge sharpening of the three color component data matrices are carried out according to the set threshold and gradient. Perform background deletion on the image pixel data with a predetermined threshold to increase the gradient of the pixel data, and perform edge detection and image segmentation on the image pixels according to the set numerical gradient to extract blood cell image elements;

4) Blood cell identification and calculation - compare the segmented image data of each element with various blood cell model data in the blood cell model database to determine the type of cell it belongs to, and perform classification statistics to calculate the correlation of various blood components data;

6) Display, print, store and transmit the test results.

The blood sample collection, quantification, dilution, and staining are all automatically completed by the sampling component.

The image acquisition, smoothing, denoising, enhancement, edge sharpening, and image element segmentation processing are completed through pre-programmed computer software, and image processing is performed on each acquisition target to meet the feature requirements for recognition and judgment.

The pattern recognition of each element image after segmentation is carried out, its nature is judged, and classification and statistics are carried out, and the relevant data of various blood components are calculated through comparison, judgment, and statistics. By establishing a database, storing each Collect the features of the image, and then compare each image element after collection and processing with the image features in the database to achieve statistical results.

The target image recognition feature is distinguished by two elements: the color difference and the different shape of the cells after dyeing, and the color after dyeing and the shape of the cells are used as the recognition basis for distinguishing different types of cells.

The staining agent is Wright's stain, a commercially available product.

Claims (6)

1. blood cell image recognition counting method, it is characterized in that, the method comprises the steps: 1) Collect blood samples, quantify, dilute, and stain; 2) Enlarge, focus, and image the blood sample to be tested according to the predetermined magnification through the optical instrument; 3) Image acquisition, smoothing, denoising, enhancement, edge sharpening, and image element segmentation processing; 4) Carry out pattern recognition to each element image after segmentation, judge its nature, and carry out classification statistics, calculate the relevant data of various blood components; 5) Display, print, store and transmit the test results. 2. The blood cell image recognition and counting method according to claim 1, characterized in that, the blood sample collection, quantification, dilution, and staining are completed automatically by the sampling component. 3. The blood cell image recognition and counting method according to claim 1, wherein the image acquisition, smoothing, denoising, enhancement, edge sharpening, and image element segmentation processing are completed by pre-programmed computer software , through image processing for each acquisition target, to meet the characteristic requirements of recognition and judgment. 4. The blood cell image recognition and counting method according to claim 1, wherein the pattern recognition is carried out for each element image after the segmentation, its properties are judged, and classification and statistics are carried out to calculate the correlation of various blood components. Data is realized through comparison, judgment, and statistics. By establishing a database, storing the characteristics of each collected image, and then comparing each image element after collection and processing with the image characteristics in the database, the statistical results can be realized. . 5. The blood cell image recognition and counting method according to claim 1, characterized in that, the target image recognition feature is distinguished by two elements, the color difference and the different shape of the cells after staining, the color after staining and the cell The shape serves as the identification basis for distinguishing different types of cells. 6. The blood cell image recognition and counting method according to claim 1, wherein the staining agent is Wright's stain.