Method and system for detecting thickness of biological slice sample based on image processing algorithm

Abstract

The invention discloses a method and a system for detecting the thickness of a biological slice sample based on an image processing algorithm, the method comprises a biological sample thickness data acquisition link and a data processing and calculating link, and the biological sample thickness data acquisition link obtains the thickness of the sample by calculating the focusing distance between the surface layer and the bottom layer of the sample; the data processing and calculation link is used for processing and calculating the obtained thickness data of the sample to obtain the final real physical thickness of the sample; the system comprises an upper computer, an objective table, a camera and a light source, the upper computer is provided with a control system, the control system is respectively connected with the objective table and the camera, the control system is provided with a biological sample thickness data acquisition module and a data processing and calculating module, and the biological sample thickness data acquisition module obtains Z-axis image definition data of a sample; and the data processing and calculation module obtains the final real physical thickness of the sample through calculation.

Description

Technical Field

[0001] This invention belongs to the field of thickness detection technology, and in particular refers to a method and system for detecting the thickness of biological slice samples based on image processing algorithms. Background Technology

[0002] Biological sample thickness measurement is a crucial step in industrial manufacturing and biomedical research, driven by the widespread need for precise characterization of material properties and biological structures. In industry, this technology is widely used for quality control and process monitoring of materials such as semiconductor thin films, metal sheets, and polymer coatings, as thickness parameters directly impact the electrical, mechanical, and optical properties of the products.

[0003] In the biological field, thickness measurement is used for tissue section thickness calibration, in vivo corneal thickness diagnosis, and cell layer growth monitoring, providing fundamental data for pathological analysis, disease diagnosis, and biological experiments. Both fields require detection methods to be high-precision, non-destructive, and adaptable. Industrial testing often employs physical techniques such as X-rays, spectral ellipsometrics, and laser interferometry, while biological testing focuses on non-destructive detection methods such as optical coherence tomography, confocal microscopy, and ultrasound imaging.

[0004] Currently, thickness detection technology is continuously developing towards micro-nano scale, real-time dynamics, and intelligent analysis.

[0005] The main methods for detecting biological sample thickness are as follows:

[0006] (1) Contact thickness detection:

[0007] Mechanical or electronic micrometers use precision threaded pairs to convert rotary motion into linear displacement or use high-precision encoders to move values, and measure directly by contacting both sides of the sample with the measuring rod.

[0008] Capacitive displacement sensor measurement involves installing a high-precision displacement sensor (probe) on each side of the sample. Each sensor measures the distance to the sample surface. Given the initial distance (L) between the two sensors, the two distance values ​​(d1 and d2) can be measured based on the capacitance change caused by probe movement using the capacitance calculation formula (1). Therefore, the sample thickness T = L - (d1 + d2). In Equation 1 is the dielectric constant, which is related to the material properties; A is the area of ​​the plates directly opposite each other; and d is the distance between the plates that produce the capacitance effect.

[0009] ;

[0010]

[0011] (2) Non-contact thickness detection

[0012] Ultrasonic thickness testing has several common modes and is widely used for online / offline testing and corrosion monitoring of materials such as metals, plastics, fiberglass, and rubber. Single-sided coupling method measures the flight time of the first or multiple echoes and calculates the thickness d according to the sound velocity using formula (2), which is the most common process. Multiple echo method generates multiple back-wall echoes when ultrasonic waves reflect back and forth inside the material. The instrument can measure the time interval Δt′ between two adjacent back-wall echoes and calculate the sample thickness d using formula (3). It can also penetrate coatings (paint, insulation, epoxy) to directly obtain the substrate thickness.

[0013] ;

[0014] ;

[0015] Laser profilometry is a technique based on the interaction between a laser beam and an object's surface, used for non-contact measurement of the object's geometry or thickness. It calculates the sample's thickness by analyzing the changes in the laser beam's reflection from the object's surface, creating a three-dimensional profile of the surface. A laser emitter emits a laser beam towards the surface being measured. Upon impact, the beam is reflected due to the surface's geometry. Different surface morphologies, unevenness, and roughness affect the direction of the reflected light. The reflected laser light is captured by a photodetector (e.g., a CCD or CMOS sensor), which detects changes in the intensity and angle of the reflected light. During the scanning process, the laser beam typically moves along the object's surface, capturing and recording the light reflected from different angles. This constructs a three-dimensional profile dataset of the object.

[0016] Existing contact thickness measurement technologies require direct contact between the probe or sensor and the sample, inevitably causing some physical damage and contamination. This limits their applicability to applications where sample purity requirements are not high, or to materials that are relatively robust, such as metals or inorganic substances. These methods typically rely on physical contact to obtain thickness data, which can lead to surface scratches and contamination, affecting the physical or chemical properties of the sample and restricting their application in delicate fields.

[0017] Non-contact thickness measurement technologies, such as ultrasound and laser scanning, have demonstrated significant advantages in many industrial and laboratory applications. However, in the biological field, especially in the measurement of living cell samples, these technologies also have certain limitations. Both ultrasound and lasers are high-energy detection methods. Although they can provide high-precision thickness measurements, they can damage cell structures when in contact with living cells. The mechanical vibrations of ultrasound and the photon irradiation of lasers can induce physical changes within cells, even leading to cell membrane rupture and photobleaching. These changes significantly affect cell biological activity and experimental results. Using these technologies to measure biological samples may cause unnecessary damage, thus interfering with subsequent biological experiments and research. Summary of the Invention

[0018] One of the objectives of this invention is to provide a method for detecting the thickness of biological slice samples based on image processing algorithms. This method enables non-contact detection of biological sample thickness without causing any damage to live cell biological samples, thus avoiding potential damage and contamination to the samples.

[0019] This objective of the present invention is achieved through the following technical solution: a method for detecting the thickness of biological slice samples based on image processing algorithms, characterized in that: the method includes a biological sample thickness data acquisition stage and a data processing and calculation stage, wherein the biological sample image data acquisition stage scans the entire sample's F-axis using a Z-axis stacked imaging method. GSL Data; the F-axis of the data processing and computation steps on the obtained samples GSL The data is processed and calculated to obtain the final true physical thickness of the sample.

[0020] To enable the detection of biological slice sample thickness, the method of this invention employs a specific search strategy. The biological sample image data acquisition process is divided into a search calibration stage and a stepwise scanning stage. In the search calibration stage, the piezoelectric motor stage is controlled to rapidly search with large step sizes throughout its entire stroke, and the threshold change trend is recorded to determine the search direction and F. GSL Threshold range.

[0021] This invention obtains the sample thickness by calculating the focusing distance between the sample surface and the bottom layer. The first step is to find a clear image of the sample surface. For a given image, more high-frequency signals indicate more image details, resulting in greater clarity and a higher gradient. However, the effectiveness of using gradients calculated solely by differential operators to represent sharpness is not high.

[0022] In the biological sample thickness data acquisition stage, it is necessary to calculate the image sharpness metric F. GSL First, let's introduce the clear metric F. GSL The specific calculation and processing procedures.

[0023] In the detection method of this invention, the image sharpness metric F is calculated by convolving the image data with a first-order difference operator and a Laplacian operator. GSL Clarity metric F GSL It has a more monotonous response to changes in sharpness and a sharper peak, making it more suitable for sharpness calculation under small displacements. The image sharpness evaluation function mainly applies a specified convolution template to the image grayscale to characterize the image sharpness.

[0024] This method assumes the input grayscale image is... The size is M×N. After Gaussian smoothing, the first-order gradient energy and the second-order Laplacian energy are calculated and weighted summed to obtain the sharpness metric F. GSL .

[0025] In this article, the same symbols represent the same meaning.

[0026] Sharpness metric F GSL The equation representing a continuous function is:

[0027] ;

[0028] in:

[0029] ;

[0030] For F GSL The gradient energy term;

[0031] ;

[0032] For F GSL Laplace energy term,

[0033] Sharpness measurement middle and As a weighting factor, This indicates that a convolution calculation is being performed. It is a two-dimensional Gaussian kernel. The discretization formula is:

[0034] ;

[0035] Where i and j are the offsets of other pixels in the convolution kernel relative to the center pixel, and I is the image data. For difference operators, The operation matrix is ​​represented as:

[0036] ;

[0037] ;

[0038] in, and These represent the operation matrices in the vertical and horizontal directions, respectively. The gray-level gradient of the image data is calculated by convolving them with the Gaussian-smoothed image data.

[0039] For the Laplace operator When the image signal remains unchanged, the calculated value will be very small. However, once edge signals appear within the image data, the value will increase rapidly using the Laplacian operator. This method calculates the intensity of edge signals within an image signal. The Laplacian operator in this method... Specifically, the four-neighbor difference operator is selected. .

[0040] ;

[0041] First, the image data is smoothed by Gaussian, and then its gradient energy term and Laplacian energy term are calculated, resulting in better robustness.

[0042] Measure the sharpness After discretization, we get:

[0043] ;

[0044] in:

[0045] ;

[0046] ;

[0047] ;

[0048] ;

[0049] This method calculates local image feature information, including local gradients and local Laplacian energy, by convolving a gradient and Laplacian matrix operator of a certain size with image data. Subsequently, the operator slides across the entire image range by changing the position of the center pixel of the convolution kernel. As shown, the squares of the local response values ​​are summed to comprehensively reflect the magnitude of the gradient change. This process not only considers the bidirectional changes in gradient ascent and descent, making the metric more sensitive to local grayscale changes, but also avoids information loss caused by sign cancellation.

[0050] Based on this, through and The weighting function performs a weighted summation of the local energies. The weighted gradient energy and the Laplacian energy result in the sharpness metric F. GSL The perception index has been transformed from a simple global average to a target-oriented and adaptive one, thus exhibiting higher robustness in noisy environments and better highlighting the structural information of the image.

[0051] The operators used in this method fully consider the anisotropic characteristics of local image edges, improving computational stability and reducing the impact of noise on the results. The computation process achieves a discrete approximation of the differential through convolution operations, transforming the differential operation into a local weighted summation, significantly reducing computational complexity and improving the algorithm's computational efficiency and real-time response capability. This allows for the rapid acquisition of the image sharpness metric F. GSL .

[0052] By using this method of first calculating the Gaussian gradient and then weighting it with edge signal intensity, the sharpness metric F of an image can be calculated quickly. GSL By comparing and clearly measuring F GSL The value is used to determine whether the sample is within the focal plane of the camera.

[0053] In this invention, the biological sample thickness data acquisition process includes a search and calibration stage and a stepwise scanning stage, specifically including the following steps:

[0054] S101: Initialize the control system, read the camera control mode, the relative position of the stage, the light source output mode, and the physical travel of the entire stage, and determine the physical scanning start and end points;

[0055] S102: Reads the set depth of field and camera exposure parameters, as well as the lens depth of field DOF, and sets the light source output time;

[0056] S103: Control the stage to move to the preset scanning start point;

[0057] S104: Read the search stage flag bit and determine the Z-axis stage displacement step size through the flag bit. In the search calibration stage, the step size is 2 times DOF.

[0058] S105: Move from the starting point to the ending point in fixed steps; acquire an image at each location using formula (4):

[0059] ,

[0060] Calculate the corresponding sharpness metric F GSL Record (Z, F) GSL (Data, until the scanning endpoint is reached;)

[0061] S106: Traverse F throughout the entire journey GSL Data, calculate control parameters, first find the position Z0 where the sharpness gradually increases, and record the stage position 10 times the DOF distance before Z0 as Z1. Continue to traverse the sharpness data downward from Z0, find the turning point Z2 where the rate of change of sharpness changes from a rapid decrease to a gradual flattening, and record the stage position 10 times the DOF distance after Z2 as Z3; S107: The stage returns to Z1, and the step-by-step scanning phase begins;

[0062] S108: Use 0.5x DOF to search downwards, acquire images at each location and calculate F. GSL Record high density (Z, F) GSL The data was gradually searched up to Z3.

[0063] S109: Combine the search calibration phase with all (Z, F) steps of the scan. GSL The data is summarized, sorted, and deduplicated to form a complete data sequence;

[0064] S110: Outputs overall biological sample collection data.

[0065] After sample data collection, the next step is data processing, specifically processing the (Z, F) data recorded during the search calibration phase and the stepwise scanning phase. GSL The data is augmented to generate (Z, F) GSL A smooth curve, from which the curve passes through the full width at half maximum (F) and (Z, F) GSL The rate of change of the curve is used to analyze the thickness of biological samples.

[0066] First, we will introduce the calculation process of the formulas involved in the data processing stage.

[0067] The point of maximum positive slope within the rising edge region of the analysis curve, at which point F... GSL A rapid ascent is considered an indication that the sample has entered the focal plane of the lens. Therefore, a target scoring function J(z) is constructed, and the optimal boundary is determined by finding the maximum value of this function. This function contains two weighted terms: a height term and a slope term. The following equation represents the search for the point of maximum slope in the curved region.

[0068] ;

[0069] in:

[0070] To impose a high constraint on the weights, the penalty coefficient for the magnitude term in the objective function is increased; it strengthens the constraint on the location of the slope point, forcing the detected slope point to be located in a low-signal region far from the peak (i.e., close to the baseline); increasing... It can effectively suppress erroneous edge detection caused by local perturbations near the peak value;

[0071] Gradient-dominated weights are the reward coefficients for the slope term in the objective function; they emphasize that the target point must possess significant signal change characteristics; increasing... This makes the model more inclined to locate the positions where signal changes are most abrupt; the gradient-dominant weights and height weights satisfy the normalization constraint: + = 1.

[0072] K is the amplitude sensitivity index, a hyperparameter greater than or equal to 1, used to adjust the nonlinearity of the height penalty term.

[0073] Since the sample thickness analyzed from the data differs from the actual thickness due to factors such as the depth of field of the objective lens and the refractive index of the immersion medium, a correction function is needed to correct the calculated sample thickness.

[0074] The correction function used in this invention is:

[0075] ;

[0076] in:

[0077] The thickness of the sample obtained after correction;

[0078] : Represents the distance between feature points calculated in the sharpness curve, representing the stage displacement difference;

[0079] This represents a combined parameter value of depth of field, incident light ray, and the angle between the objective lens and the incident light ray.

[0080] : The refractive index of the sample and its mounting medium (plant tissue / glycerol / resin is typically between 1.45 and 1.55).

[0081] : The refractive index of the medium into which the objective lens is immersed (air = 1.0, water = 1.33, oil = 1.51).

[0082] In this invention, the data processing and calculation stage includes the following steps:

[0083] S201: Data Preprocessing and Smoothing: Read all collected (Z, F) data. GSL The original data points are smoothed using a one-dimensional Gaussian filter to remove high-frequency random noise during the acquisition process and preserve the overall trend of the curve.

[0084] S202: Height-restricted region screening: In order to avoid the "double peak" interference that may be caused by the internal structure of the sample, a height threshold is set; the part of the dataset that is higher than this threshold is temporarily blocked, and only the data of the low region is retained for subsequent detection;

[0085] S203: Gradient Calculation and Slope Extreme Value Location: Within the low-order region selected in step S202, calculate the first derivative of the curve; using formula (16):

[0086] ,

[0087] Find the point with the maximum positive slope in the rising edge region and the point with the minimum negative slope in the falling edge region respectively;

[0088] S204: Construct the tangent equation: Using the maximum / minimum slope points found in step S203 and their corresponding coordinates (Z, F) GSL Construct the tangent equations for the rising edge and the falling edge, respectively;

[0089] S205: Calculate the boundary intersection: Combine the equations of the ascending / descending tangents with the baseline values ​​to calculate the coordinates of the intersection points of the tangents and the baseline; mark the positions Z corresponding to these two intersection points as the starting boundary Z. up and the falling boundary Z down ;

[0090] S206: Calculate the optical travel thickness: Calculate the uncorrected optical travel thickness Z based on the difference between the two boundary positions. raw = |Z up -Z down |;

[0091] S207: Based on F GSLMAX At half-height position, the full width at half-peak (FWHM) is calculated to evaluate the depth-of-field performance and data reliability of the current optical system.

[0092] S208: Physical Refractive Index Correction: Reads the preset sample refractive index. and the refractive index of the objective lens medium Apply formula (17):

[0093] The thickness of the biological sample is then corrected and calculated to obtain the final true physical thickness.

[0094] The second objective of this invention is to provide a system for detecting the thickness of biological slice samples based on image processing algorithms, which enables non-contact detection of biological sample thickness.

[0095] This objective of the present invention is achieved through the following technical solution: a system for detecting the thickness of biological slide samples based on image processing algorithms, characterized in that: the system includes a host computer, a stage, a camera, and a light source; the stage is used to carry biological slide samples; the camera is used to acquire image data of the biological slide samples; the host computer has a control system, which is connected to the stage and the camera respectively, and is used to receive image data acquired by the camera and control the movement of the stage; the control system has a biological sample thickness data acquisition module and a data processing and calculation module; the biological sample thickness data acquisition module obtains the sample thickness by calculating the focusing distance between the sample surface and the bottom layer; the data processing and calculation module processes and calculates the obtained sample thickness data to obtain the final true physical thickness of the sample.

[0096] In this invention, the control system is connected to the stage via serial communication to control the movement of the stage, and the control system is connected to the camera via CameraLink communication.

[0097] In this invention, the light source is cool white light with a color temperature of 4800K generated by an LED, which greatly reduces the damage to live cell samples caused by lighting during image acquisition.

[0098] Compared with the prior art, the present invention has the following significant effects:

[0099] This application innovatively uses a sharpness evaluation algorithm deployed on a computer to scan the sample layer by layer with a high-precision camera, and then calculates the sample thickness by analyzing the sharpness data curve of the layer-by-layer microscopic imaging images along the Z-axis direction.

[0100] This method does not require the use of additional probes or sensors and does not directly contact the sample, thus avoiding contamination and damage to the sample caused by contact. The measured signal is the microscopic imaging data of the sample, and the illumination light is 4800K cool white light generated by LED, which will not cause damage to the live cell sample due to the transmission of the acquired signal.

[0101] Furthermore, to further reduce potential damage to samples during image acquisition, this method can be equipped with a high-speed synchronous bright-field control system. This system can strictly synchronize the timing of light source illumination and camera sampling, avoiding additional illumination time caused by system control delays. For some photosensitive biological samples, this can significantly reduce light-induced damage.

[0102] This application avoids the damage to biological samples caused by existing thickness measurement technologies when measuring the thickness of biological samples, and realizes non-destructive thickness measurement of live cell biological samples, which facilitates subsequent biological experiments and provides preliminary sample data for researchers. Attached Figure Description

[0103] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0104] Figure 1 This is a schematic diagram of the detection system of the present invention;

[0105] Figure 2 This is a block diagram of the biological sample thickness data acquisition process of the present invention;

[0106] Figure 3 This is a block diagram of the data processing and calculation steps of the present invention;

[0107] Figure 4 The sharpness measure F obtained in the embodiments of the present invention GSL Line graph. Detailed Implementation

[0108] like Figure 1 The system shown is for detecting the thickness of biological slide samples based on image processing algorithms. The system includes a host computer, a stage, a camera, and a light source. The stage is a high-precision Z-axis stage used to support the biological slide samples. The Z-axis stage has stable nanometer-level displacement control capabilities, meeting the precise positioning requirements for continuous Z-axis imaging of the slide samples. The camera is a Hamamatsu C13440 camera used to acquire image data of the biological slide samples. The light source is 4800K cool white light generated by LEDs. The accompanying bright-field illumination system can output a uniform and adjustable illumination field, ensuring the contrast and detail integrity of images of different stained slide samples.

[0109] The host computer has a control system, which is connected to the stage via serial communication to control the movement of the stage. The control system is also connected to the camera via CameraLink to receive image data acquired by the camera. The control system has a biological sample thickness data acquisition module and a data processing and calculation module. The biological sample thickness data acquisition module obtains the sample thickness by calculating the focusing distance between the sample surface and the bottom layer. The data processing and calculation module processes and calculates the obtained sample thickness data to obtain the final true physical thickness of the sample.

[0110] After the system starts measuring, the camera acquires image data at a rate of 100 frames per second and transmits it to the host computer via CameraLink. The control system then moves the stage to allow the camera to scan the entire physical path.

[0111] The detection method of the present invention will be described and verified below through specific wide-field microscope verification experiments.

[0112] Experimental Objective

[0113] 1. To verify the feasibility of this method in the field of thickness detection of biological samples under microscopic imaging, and to design a systematic test scheme for subsequent performance verification of thickness detection.

[0114] 2. Calculate the sharpness metric F for each image in the Z-axis stack. GSL This verifies the applicability of the function to sliced ​​samples.

[0115] 3. Analyze the sharpness distribution pattern of the tissue section samples along the Z-axis to determine the Z-axis position where the image is sharpest, providing image data for tissue section thickness analysis.

[0116] 4. By photographing biological standard samples of different thicknesses, the calculated sample thickness is compared with the actual standard sample thickness.

[0117] Experimental environment

[0118] Wide-field microscope: Equipped with a high-resolution 16-bit grayscale camera, high-magnification low-numerical-aperture oil-medium objectives, and a fully electric closed-loop control Z-axis drive stage (maximum control accuracy 50nm); the accompanying software supports stacked acquisition and imaging, and it is equipped with a bright-field illumination system, and is compatible with slides for imaging tissue samples and culture dish stages.

[0119] Biological sample: Spinach root slice (10 μm thick, stained with hematoxylin and eosin (HE) for plant tissue imaging).

[0120] Software tools: The microscope's host computer is developed based on the QT framework. The thickness detection algorithm and stage control logic are deployed in C++, and the OpenCV library is used to implement image processing such as contrast enhancement and noise reduction.

[0121] Experimental Design

[0122] like Figure 2 , Figure 3 As shown, the thickness of biological slide samples of different thicknesses was measured according to the method for detecting the thickness of biological slide samples described in this invention, and the F-values ​​of all captured images were recorded. GSL Data is collected for subsequent analysis and statistical analysis of performance parameters, including maximum resolution (F). GSLMAX Measurement time T, thickness error Thickness correction .

[0123] Preliminary experimental stage: Verify the accuracy of the microscope stage and whether the camera and bright-field illumination meet the method requirements. After verification, fix the camera exposure, bright-field intensity and other experimental parameters, and conduct thickness measurement experiments.

[0124] Formal experiment: Thickness calculations were performed multiple times for three fields of view of the same sample, and corrected thickness was calculated for data from a single field of view. This ensured that consistent exposure parameters and brightfield intensity were maintained for each image capture. To minimize errors, the microscope was placed on a pneumatically balanced optical platform to avoid interference from external vibrations.

[0125] Data processing: Organize and statistically analyze the experimental measurement data, focusing on the following indicators.

[0126] Including maximum resolution F GSLMAX The maximum sharpness measure within a field of view of a sample;

[0127] Measurement time T: The time taken for one measurement, in seconds;

[0128] Thickness error The error between the corrected thickness and the standard sample thickness is calculated.

[0129] Thickness Correction Physically corrected thickness within a single field of view of a single sample. Unit: μm;

[0130] The experimental data are shown in the table below:

[0131]

[0132] The experimental results show that this method has extremely high accuracy and takes relatively little time. The sharpness metric F obtained in this embodiment... GSL Curve graph as Figure 4 As shown.

Claims

1. A method for detecting the thickness of biological slide samples based on image processing algorithms, characterized in that: The method includes a biological sample thickness data acquisition stage and a data processing and calculation stage. The biological sample image data acquisition stage scans the entire sample's F-axis using a Z-axis stacked imaging method. GSL Data; the F-axis of the data processing and computation steps on the obtained samples GSL The data is processed and calculated to obtain the final true physical thickness of the sample.

2. The method for detecting the thickness of biological slice samples based on image processing algorithms according to claim 1, characterized in that: The biological sample thickness data acquisition process includes a search and calibration phase and a stepwise scanning phase, specifically including the following steps: S101: Initialize the control system, read the camera control mode, the relative position of the stage, the light source output mode, and the physical travel of the entire stage, and determine the physical scanning start and end points; S102: Reads the set depth of field and camera exposure parameters, as well as the lens depth of field DOF, and sets the light source output time; S103: Control the stage to move to the preset scanning start point; S104: Read the search stage flag bit and determine the Z-axis stage displacement step size through the flag bit. In the search calibration stage, the step size is 2 times DOF. S105: Move from the starting point to the ending point in fixed steps; acquire an image at each location and calculate the corresponding sharpness metric F. GSL Record (Z, F) GSL (Data, until the scanning endpoint is reached;) S106: Traverse F throughout the entire journey GSL Data, calculate control parameters, first find the position Z0 where the sharpness gradually increases, and record the stage position 10 times the DOF distance before Z0 as Z1. Continue to traverse the sharpness data downward from Z0 to find the turning point Z2 where the rate of change of sharpness changes from a rapid decrease to a gradual flattening, and record the stage position 10 times the DOF distance after Z2 as Z3. S107: The stage retracts to Z1, and the step-by-step scanning phase begins; S108: Use 0.5x DOF to search downwards, acquire images at each location and calculate F. GSL Record high density (Z, F) GSL The data was gradually searched up to Z3. S109: Combine the search calibration phase with all (Z, F) steps of the scan. GSL The data is summarized, sorted, and deduplicated to form a complete data sequence; S110: Outputs overall biological sample collection data.

3. The method for detecting the thickness of biological slice samples based on image processing algorithms according to claim 2, characterized in that: In step S105, formula (4) is used: ， Calculate the corresponding sharpness metric F GSL ; In formula (4): ； For F GSL The gradient energy term; ； For F GSL Laplace energy term, Sharpness measurement middle and As a weighting factor, This indicates that a convolution calculation is being performed. It is a two-dimensional Gaussian kernel. The discretization formula is: ； Where i and j are the offsets of other pixels in the convolution kernel relative to the center pixel, and I is the image data. For difference operators, The operation matrix is ​​represented as: ； ； in, and These represent the operation matrices in the vertical and horizontal directions, respectively. For the Laplace operator The four-neighbor difference operator is selected. ; ； Measure the sharpness After discretization, we get: ； in: ； ； ； 。 4. The method for detecting the thickness of biological slice samples based on image processing algorithms according to claim 1, 2, or 3, characterized in that: The data processing and calculation process includes the following steps: S201: Data Preprocessing and Smoothing: Read all collected (Z, F) data. GSL The original data points are smoothed using a one-dimensional Gaussian filter to remove high-frequency random noise during the acquisition process and preserve the overall trend of the curve. S202: Height-restricted region screening: In order to avoid the "double peak" interference that may be caused by the internal structure of the sample, a height threshold is set; the part of the dataset that is higher than this threshold is temporarily blocked, and only the data of the low region is retained for subsequent detection; S203: Gradient Calculation and Slope Extreme Value Location: Within the low-order region selected in step S202, calculate the first derivative of the curve; using formula (16): ， Find the point with the maximum positive slope in the rising edge region and the point with the minimum negative slope in the falling edge region, respectively. in: The penalty coefficient for the magnitude term in the objective function is used to constrain the weights. Gradient-dominated weights are the reward coefficients for the slope term in the objective function. The gradient-dominant weights and height weights satisfy the normalization constraint: + = 1; K is the amplitude sensitivity index, a hyperparameter greater than or equal to 1; S204: Construct the tangent equation: Using the maximum / minimum slope points found in step S203 and their corresponding coordinates (Z, F) GSL Construct the tangent equations for the rising edge and the falling edge respectively; S205: Calculate the boundary intersection: Combine the equations of the ascending / descending tangents with the baseline values ​​to calculate the coordinates of the intersection points of the tangents and the baseline; mark the positions Z corresponding to these two intersection points as the starting boundary Z. up and the falling boundary Z down ; S206: Calculate the optical travel thickness: Calculate the uncorrected optical travel thickness Z based on the difference between the two boundary positions. raw = |Z up -Z down |; S207: Based on F GSLMAX At half-height position, the full width at half-peak (FWHM) is calculated to evaluate the depth-of-field performance and data reliability of the current optical system. S208: Physical Refractive Index Correction: Reads the preset sample refractive index. and the refractive index of the objective lens medium Apply formula (17): ， The corrections are then made to calculate the final true physical thickness of the biological sample; in: The thickness of the sample obtained after correction; : Represents the distance between feature points calculated in the sharpness curve, representing the stage displacement difference; This represents a combined parameter value of depth of field, incident light ray, and the angle between the objective lens and the incident light ray. : The refractive index of the sample and its sealing medium; : The refractive index of the medium into which the objective lens is immersed.

5. A detection system for implementing the method for detecting the thickness of biological slide samples based on image processing algorithms as described in any one of claims 1 to 4, characterized in that: The system includes a host computer, a stage, a camera, and a light source. The stage is used to hold biological slide samples, and the camera is used to acquire image data of the biological slide samples. The host computer has a control system, which is connected to both the stage and the camera. The control system receives image data acquired by the camera and controls the movement of the stage. The control system includes a biological sample thickness data acquisition module and a data processing and calculation module. The biological sample thickness data acquisition module obtains the sample thickness by calculating the focusing distance between the sample surface and the bottom layer. The data processing and calculation module processes and calculates the obtained sample thickness data to obtain the final true physical thickness of the sample.

6. The detection system according to claim 5, characterized in that: The control system is connected to the stage via serial communication to control the movement of the stage, and the control system is connected to the camera via CameraLink communication.

7. The detection system according to claim 5, characterized in that: The light source is cool white light with a color temperature of 4800K generated by an LED.

Images