CN111595758B - Ellipse cell detection device and detection method based on lens-free imaging - Google Patents

Abstract

The invention discloses an ellipse cell detection device based on lensless imaging, which includes a light box and a circuit board protection box arranged up and down. The light box is provided with an LED light source, a convex lens, and a CMOS image sensor. The invention can collect cells for the lensless imaging system. In the case of low resolution and severe diffraction interference in the image, the elliptical cells can be detected accurately. The detection method of elliptic cells of the present invention, the method uses the elliptical cell detection device based on lensless imaging of the present invention to detect elliptical cells, including making and fixing the elliptical cell samples to be tested, turning on the LED light source, and collecting image data by a CMOS image sensor, processing The image data can judge whether it is a normal ellipse cell according to the difference between the long axis and the short axis of the ellipse cell. The invention improves the automation degree of ellipse cell measurement and reduces the labor cost of operation.

Description

Ellipse cell detection device and detection method based on lens-free imaging

technical field

The invention belongs to the technical field of medical image analysis, relates to an ellipse cell detection device based on lensless imaging, and also relates to a detection method of the ellipse cell.

Background technique

Blood cell morphological parameters are very valuable data for disease diagnosis, health monitoring, and new drug development. The usual way to obtain this data is to acquire images of cells for analysis. A common acquisition device is an optical microscope. In order to ensure the imaging quality, optical microscopes have to use complex multi-lens groups. The problem of miniaturization of optical microscopes has been limited to lens groups. In the face of emergency situations such as sudden outbreaks or the application scenarios of real-time detection such as collection and on-site detection, high professional requirements and manufacturing costs have become constraints, which are not conducive to rapid deployment, low-cost large-scale deployment and real-time detection.

In 2006, Yang Changhuei's research group at California Institute of Technology first proposed the concept of a lensless optofluidic microchip. The advantage of this device is that the use of image sensors not only reduces the cost of manufacturing and using the device, but also improves portability. It has more advantages in the application scenarios of rapid deployment, low-cost large-scale deployment and instant detection.

There are two issues to be solved for lensless devices: 1. The resolution of cell images collected by lensless imaging systems is low due to the fact that the pixel size of commercial image sensors is very close to that of blood cells; 2. Contacts with less diffraction interference Traditional imaging requires that the distance from the imaging sample to the image sensor must be reduced to a few wavelengths of visible light, which is difficult to achieve for commercial sensors, and diffraction imaging with interference has to be used. Therefore, how to obtain accurate cell morphological parameters when using commercial sensors has become a research difficulty.

Contents of the invention

The purpose of the present invention is to provide an elliptical cell detection device based on lensless imaging, which can accurately detect elliptical cells in view of the low resolution and serious diffraction interference when the lensless imaging system collects cell images.

Another object of the present invention is to provide a method for detecting oval cells, which can accurately detect oval cells.

The first technical solution adopted by the present invention is an ellipse cell detection device based on lensless imaging, including a light box and a circuit board protection box arranged up and down, and LEDs on the same axis are sequentially arranged in the light box from top to bottom. A light source, a convex lens, and a CMOS image sensor. The CMOS image sensor is located on a circuit board protection box, and a CMOS image sensor circuit connected to the CMOS image sensor is arranged in the circuit board protection box.

The feature of the first technical solution of the present invention is also that,

The light box is rotatably connected with the circuit board protection box, the light box can rotate on the circuit board protection box, the side wall of the light box is provided with an opening, and the circuit board protection box is fixed with a shading plate matching the opening of the light box.

A sample pad is provided on the periphery of the CMOS image sensor.

A sample card slot is arranged on the sample pad.

The detection method of elliptic cells of the present invention, the method uses the elliptical cell detection device based on lensless imaging of the present invention to detect elliptical cells, specifically implemented according to the following steps:

Step 1, making the elliptical cell sample to be tested;

Step 2. Place the elliptical cell sample to be tested on the CMOS image sensor and fix it in the sample card slot. Rotate the light box so that its opening and the light shield form a closed collection room;

Step 3. Turn on the LED light source, so that the light emitted by the LED light source is gathered by the convex lens and vertically shoots towards the elliptical cell sample to be tested;

Step 4, open the CMOS image sensor, and export the image data collected to the processing module of the CMOS image sensor circuit through the data interface;

Step 5. Use the Hough transform algorithm to detect the center of the image of the circular elliptical cell after diffraction interference on the exported image data, and establish a rectangular coordinate system with it as the center, and segment the image to detect the cell diffraction ring;

Step 6. On the basis of step 5, perform a sub-pixel interpolation operation to obtain a sub-pixel image, and take the circle center coordinates obtained in step 5 as the center, convert the sub-pixel image from a rectangular coordinate system to a polar coordinate system;

Step 7, determine the curve of the distance from the strongest light intensity point of the diffraction bright ring to the center of the cell diffraction ring through the super-resolution algorithm;

Step 8, according to the distance curve in step 6, calculate the light intensity maximum point of the diffraction bright ring, and characterize the position of the diffraction bright ring;

Step 9. Increase the parameter θ of the polar coordinate system from 0 to 360 degrees, increase according to the set step size, repeat steps 7 to 8 in sequence until the radius data of 360 degrees is obtained, and form a radius sequence { x′ _pk1 (θ ₁ ), x′ _pk2 (θ ₂ ), …, x′ _pkn (θ _n )}, x′ _pkn (θ _n ) represents the radius value corresponding to the polar coordinate system parameter θ, and n is not 0 natural numbers of

Step 10. Characterize the diameter of the diffraction ring by the sum of the two radii, that is, match the radius to obtain the diameter length of the diffraction ring at angles α ₁ , α ₂ ,...,α _i , where i is a natural number other than 0:

In formula (10), D ₁ (α ₁ ), D ₂ (α ₂ ),…,D _i (α _i ) are the diameter lengths of the diffraction ring at angles α ₁ , α ₂ ,…,α _i ;

Define a collection A to represent these data:

A={D ₁ (α ₁ ), D ₂ (α ₂ ),..., D _i (α _i )} (11);

Step 11, carry out the matching recognition of the ellipse cells, the shadow length of the long axis position of the ellipse is expressed as:

D _maj (α _maj )=min(A) (12)

Assuming π/4<α _maj <3π/4, the definition set B is expressed as

B＝{D ₁ (β ₁ ), D ₂ (β ₂ ),..., D _i (β _i )} (13)

The length of the shadow at the position of the minor axis of the ellipse can be expressed as:

D _min (α _min ) = min(B)

Define the parameter M to represent the difference between the major and minor axes of the elliptical cells:

M＝|D _maj (α _maj )-D _min (α _min )|

Define the threshold M _T , if M>M _T , then it is judged to be a normal elliptical cell, otherwise it is other cells.

The feature of the second technical solution of the present invention is also that,

Step 7 is specifically, according to the light intensity information carried by the image data itself, the mapping relationship between the light intensity and the distance from the point where the diffraction occurs is obtained. The point where the diffraction occurs is the edge of the cell. This relationship means that there will be multiple Peak, each peak represents a diffraction bright ring, and the position of the diffraction bright ring is represented by the maximum point of the peak;

According to the radial super-resolution and circumferential super-resolution, the curve of the distance from the strongest light intensity point of the diffraction bright ring to the center of the cell diffraction ring is obtained: the radial super-resolution is to find the fastest rising point and the fastest decline of the peak corresponding to the diffraction bright ring point, use the intersection of the slopes of these two points to correct the position of the peak maximum point; circumferential super-resolution, denoise through Gaussian smoothing, and then obtain the corrected representative position of the diffraction bright ring; calculate the representative position of the diffraction bright ring Calculate the distance from the diffraction bright ring to the cell diffraction ring, and obtain the curve of the distance from the strongest light intensity point of the diffraction bright ring to the center of the cell diffraction ring.

The beneficial effects of the present invention are:

The elliptical cell detection device based on lensless imaging of the present invention has a stable structure and simple device; it is oriented to application scenarios such as rapid deployment, low-cost large-scale deployment and instant detection; it has the advantages of low cost and high portability; it can be used for lensless imaging When the system collects cell images, there are low resolution and serious diffraction interference, and the elliptical cells are accurately detected, which realizes the miniaturization of the elliptical cell detection device.

The detection method of the ellipse cell of the present invention is simple and clear, has strong operability, improves the automation degree of the ellipse cell measurement, and reduces the labor cost of operation.

Description of drawings

FIG. 1 is a schematic structural diagram of an ellipse cell detection device based on lensless imaging according to the present invention.

In the figure, 1. Light box, 2. LED light source, 3. Convex lens, 4. Sample card slot, 5. Shading plate, 6. Sample pad, 7. CMOS image sensor, 8. Circuit board protection box, 9. Sample pad .

Detailed ways

The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments.

The present invention is based on the elliptical cell detection device without lens imaging, as shown in Figure 1, including a light box 1 and a circuit board protection box 8 arranged up and down, the light box 1 and the circuit board protection box 8 are rotatably connected, and the light box 1 can be connected to the circuit board The board protection box 8 is rotated, and the side wall of the light box 1 is provided with an opening. The LED light source 2, the convex lens 3, and the CMOS image sensor 7 on the top, a sample pad 9 is arranged on the periphery of the CMOS image sensor 7, and a sample card slot 4 is arranged on the sample pad 9, and the CMOS image sensor 7 is located on the circuit board protection box 8, and the circuit board A CMOS image sensor circuit connected to the CMOS image sensor 7 is provided in the protective box 8 .

The detection method of elliptic cells of the present invention, the method is applied and the elliptical cell detection device based on lensless imaging of the present invention is used for detection of elliptical cells, specifically implemented according to the following steps:

Step 1, making the elliptical cell sample 6 to be tested.

Step 2. Place the elliptical cell sample 6 to be tested on the CMOS image sensor 7 and fix it in the sample card slot 4. Rotate the light box 1 so that its opening and the light shield form a closed collection chamber.

Step 3, turn on the LED light source 2, so that the light emitted by the LED light source 2 is collected by the convex lens 3 and vertically directed toward the elliptical cell sample 6 to be tested.

Step 4, turn on the CMOS image sensor 7, and export the collected image data to the processing module of the CMOS image sensor circuit through the data interface.

Step 5. Use the Hough transform algorithm to detect the center of the blood cell circle on the exported image data, and segment the image according to the center of the blood cell circle to detect the blood cell diffraction ring.

Step 6. On the basis of step 5, perform sub-pixel interpolation operation to obtain a sub-pixel image, and center the circle center coordinate obtained in step 5, convert the sub-pixel image from a rectangular coordinate system to a polar coordinate system.

Step 7, determine the curve of the distance from the strongest light intensity point of the diffraction bright ring to the center of the cell diffraction ring through the super-resolution algorithm;

Step 7 is specifically, according to the light intensity information carried by the image data itself, the mapping relationship between the light intensity and the distance from the point where the diffraction occurs is obtained. The point where the diffraction occurs is the edge of the cell. This relationship means that there will be multiple Peak, each peak represents a diffraction bright ring, and the position of the diffraction bright ring is represented by the maximum point of the peak;

According to the radial super-resolution and circumferential super-resolution, the curve of the distance from the strongest light intensity point of the diffraction bright ring to the center of the cell diffraction ring is obtained: the radial super-resolution is to find the fastest rising point and the fastest decline of the peak corresponding to the diffraction bright ring point, use the intersection of the slopes of these two points to correct the position of the peak maximum point; circumferential super-resolution, denoise through Gaussian smoothing, and then obtain the corrected representative position of the diffraction bright ring; calculate the representative position of the diffraction bright ring Calculate the distance from the diffraction bright ring to the cell diffraction ring, and obtain the curve of the distance from the strongest light intensity point of the diffraction bright ring to the center of the cell diffraction ring.

Step 8. According to the distance curve in step 6, calculate the maximum light intensity point of the diffraction bright ring, which represents the position of the diffraction bright ring.

Step 9. Use the polar coordinate system parameter θ to increase from 0 degrees to 360 degrees, and increase according to the set step size. For example, increase by 1 degree each time, repeat steps 7 to 8 in sequence until the radius data of 360 degrees is obtained, and the 360 degrees Radius data constitute a radius sequence {x′ _pk1 (θ ₁ ), x′ _pk2 (θ ₂ ), …, x′ _pkn (θ _n )}, x′ _pkn (θ _n ) represents the radius corresponding to the polar coordinate system parameter θ value, n is a natural number not equal to 0.

Step 10. Characterize the diameter of the diffraction ring by the sum of the two radii, and match the radii to obtain the diameter length of the diffraction ring at angles α ₁ , α ₂ ,..., α _i , where i is a natural number other than 0:

In formula (10), D ₁ (α ₁ ), D ₂ (α ₂ ),…,D _i (α _i ) are the diameter lengths of the diffraction ring at angles α ₁ , α ₂ ,…,α _i ;

Define a collection A to represent these data:

A={D ₁ (α ₁ ), D ₂ (α ₂ ), . . . , D _i (α _i )} (11).

Step 11, carry out the matching recognition of the ellipse cells, the shadow length of the long axis position of the ellipse is expressed as:

D _maj (α _maj )=min(A) (12)

Assuming π/4<α _maj <3π/4, the definition set B is expressed as

B＝{D ₁ (β ₁ ), D ₂ (β ₂ ),..., D _i (β _i )} (13)

The length of the shadow at the position of the minor axis of the ellipse can be expressed as:

D _min (α _min ) = min(B)

Define the parameter M to represent the difference between the major and minor axes of the elliptical cells:

M＝|D _maj (α _maj )-D _min (α _min )|

Define the threshold M _T , if M>M _T , then it is judged to be a normal elliptical cell, otherwise it is other cells.

Claims (2)

1. The detection method of elliptic cells is characterized in that the method uses an elliptical cell detection device based on lens-free imaging for elliptical cell detection, and the elliptical cell detection device based on lens-free imaging includes a light box (1) and a circuit set up and down A board protection box (8), the light box (1) is sequentially provided with an LED light source (2), a convex lens (3), and a CMOS image sensor (7) on the same axis from top to bottom, and the CMOS image sensor (7) on the circuit board protection box (8), the circuit board protection box (8) is provided with a CMOS image sensor circuit connected to the CMOS image sensor (7), and the periphery of the CMOS image sensor (7) is provided with A sample pad (9), the sample pad (9) is provided with a sample card slot (4), the light box (1) is rotatably connected to the circuit board protection box (8), and the light box (1) can be placed on the circuit board The protection box (8) is rotated, the side wall of the light box (1) is provided with an opening, and the circuit board protection box (8) is fixed with a light shield (5) matching the opening of the light box (1); Specifically follow the steps below: Step 1, making the elliptical cell sample to be tested (6); Step 2. Place the elliptical cell sample (6) to be tested on the CMOS image sensor (7), and fix it in the sample card slot (4), rotate the light box (1) so that its opening and the shading plate form a closed collection room; Step 3, turning on the LED light source (2), so that the light emitted by the LED light source (2) is gathered by the convex lens (3) and vertically directed to the elliptical cell sample (6) to be tested; Step 4, open described CMOS image sensor (7), the image data collected is exported on the processing module of CMOS image sensor circuit through data interface; Step 5. Use the Hough transform algorithm to detect the center of the blood cell circle on the exported image data, and segment the image according to the center of the blood cell circle to detect the blood cell diffraction ring; Step 6. On the basis of step 5, perform a sub-pixel interpolation operation to obtain a sub-pixel image, and take the circle center coordinates obtained in step 5 as the center, convert the sub-pixel image from a rectangular coordinate system to a polar coordinate system; Step 7, determine the curve of the distance from the strongest light intensity point of the diffraction bright ring to the center of the cell diffraction ring through the super-resolution algorithm; Step 8, according to the distance curve in step 6, calculate the light intensity maximum point of the diffraction bright ring, which represents the position of the first diffraction bright ring; Step 9. Increase the parameter θ of the polar coordinate system from 0 to 360 degrees, increase according to the set step size, repeat steps 7 to 8 in sequence until the radius data of 360 degrees is obtained, and form a radius sequence { x' _pk1 (θ ₁ ), x' _pk2 (θ ₂ ), ..., x' _pkn (θ _n )}, x' _pkn (θ _n ) represents the radius value corresponding to the polar coordinate system parameter θ, and n is not 0 the natural number of Step 10. Characterize the diameter of the diffraction ring by the sum of the two radii, and match the radii to obtain the diameter length of the diffraction ring at angles α ₁ , α ₂ ,..., α _i , where i is a natural number other than 0:

In formula (10), D ₁ (α ₁ ), D ₂ (α ₂ ),…,D _i (α _i ) are the diameter lengths of the diffraction ring at angles α ₁ , α ₂ ,…,α _i ; Define a collection A to represent these data: A={D ₁ (α ₁ ), D ₂ (α ₂ ),..., D _i (α _i )} (11); Step 11, carry out the matching recognition of the ellipse cells, the shadow length of the long axis position of the ellipse is expressed as: D _maj (α _maj )=min(A) (12) Assuming π/4<α _maj <3π/4, the definition set B is expressed as B＝{D ₁ (β ₁ ), D ₂ (β ₂ ),..., D _i (β _i )} (13)

The length of the shadow at the position of the minor axis of the ellipse can be expressed as: D _min (α _min ) = min(B) Define the parameter M to represent the difference between the major and minor axes of the elliptical cells: M＝|D _maj (α _maj )-D _min (α _min )| Define the threshold M _T , if M>M _T , then it is judged to be a normal elliptical cell, otherwise it is other cells. 2. The detection method of elliptic cells according to claim 1, wherein said step 7 is specifically, according to the light intensity information carried by the image data itself, the mapping relationship between the light intensity and the length of the point where the diffraction occurs is obtained, The point where the diffraction occurs is the edge of the cell. This relationship means that there will be multiple peaks in the two-dimensional coordinate system. Each peak represents a diffraction bright ring, and the maximum point of the peak represents the position of the diffraction bright ring; The first diffraction bright ring is the first diffraction bright ring: Radial super-resolution is to find the fastest rising point and fastest falling point of the peak corresponding to the first diffraction bright ring, and correct it by the intersection of the slopes of these two points The position of the peak maximum point; circumferential super-resolution, noise reduction through Gaussian smoothing, and then obtain the corrected representation position of the first diffraction bright ring; calculate the first diffraction bright ring to cell diffraction through the representation position of the first diffraction bright ring The distance of the ring, the curve of the distance from the strongest light intensity point of the first diffraction bright ring to the center of the cell diffraction ring is obtained.

Images