技术领域technical field
本发明属于医疗检测技术领域，具体涉及一种角膜内皮细胞分析仪及分析方法。The invention belongs to the technical field of medical detection, and in particular relates to a corneal endothelial cell analyzer and an analysis method.
背景技术Background technique
在各种临床和研究过程中，细胞计数在各种疾病或异常状况的诊断过程中以及在监控经历某种疾病或状况的治疗病人的过程中是很有价值的。传统的人工细胞计数是利用栅格对应计数，计数乏味，且使用者很容易出错，后来出现了数字成像系统，其使细胞计数过程的自动化变为可能。目前，由于光学部件的限制，自动化细胞计数器很容易受到光线等环境因素的影响，采集到的细胞图像质量有偏差，在进行图像预处理的过程中会受到较大影响，直接导致后续的自动计数不准确。In various clinical and research procedures, cell counts are valuable in the diagnosis of various diseases or abnormal conditions and in the monitoring of treated patients experiencing a disease or condition. The traditional manual cell counting uses grid correspondence counting, which is tedious and prone to errors by users. Later, digital imaging systems appeared, which made it possible to automate the cell counting process. At present, due to the limitations of optical components, automated cell counters are easily affected by environmental factors such as light, and the quality of the collected cell images is deviated, which will be greatly affected during the image preprocessing process, which directly leads to subsequent automatic counting. Inaccurate.
角膜内皮细胞镜检查的目的是对内皮细胞密度的中间和/或平均差和标准偏差、细胞平均大小、变异系数、形状小于六边、等于六边和大于六边的细胞百分数，以及形状因子进行评估，分析得出维持角膜活性(透明度和屈光力)的解剖学结构。其在眼库常规工作中是必不可少的，用来测定捐赠角膜的活性，以提高角膜移植的成功指数。The purpose of corneal endothelial cytoscopy is to evaluate the median and/or mean difference and standard deviation of endothelial cell density, mean cell size, coefficient of variation, percentage of cells with shape smaller than six sides, equal to six sides, and larger than six sides, and shape factor. Assess and analyze the anatomy that maintains corneal activity (transparency and refractive power). It is essential in the routine work of eye banks to measure the activity of donated corneas in order to improve the success index of corneal transplantation.
角膜内皮细胞镜根据是否与检查者角膜接触可分为：非接触式角膜内皮镜(这类设备所要分析的角膜与设备物镜光学部分不需要接触)和接触式角膜内皮镜(这类设备所要分析的角膜与物镜之间必须发生物理接触)。Corneal endothelial cytoscope can be divided into: non-contact corneal endothelioscope (the cornea to be analyzed by this type of equipment does not need to contact with the optical part of the equipment objective lens) and contact type corneal endotheloscope (to be analyzed by this type of equipment) according to whether it is in contact with the examiner's cornea. There must be physical contact between the cornea of the lens and the objective lens).
根据数据分析形式，角膜内皮细胞镜可分为以下几类：According to the form of data analysis, corneal endothelial cytoscopy can be divided into the following categories:
(1)非自动角膜内皮细胞镜：这类设备没有提供内皮评估来源，操作者自行负责内皮评估，即不能提供有内皮密度和细胞形态标准(多形性和polymegatism)估计值比较。(1) Non-automatic corneal endothelial cytoscope: This type of equipment does not provide a source of endothelial assessment, and the operator is responsible for the endothelial assessment, that is, it cannot provide a comparison of estimated values with endothelial density and cell morphology standards (pleomorphism and polymegatism).
(2)半自动角膜内皮细胞镜：这类设备提供了内皮分析来源，但是操作者、软件和获取的内皮图像之间必须发生交互，才能提供有内皮密度和细胞形态标准(多形性和polymegatism)的估计值。(2) Semi-automatic corneal endothelial cytoscope: This type of equipment provides a source of endothelial analysis, but there must be interaction between the operator, software, and acquired endothelial images to provide endothelial density and cell morphology criteria (pleomorphism and polymegatism) estimated value.
(3)自动角膜内皮细胞镜：这类设备提供了内皮分析来源，不需要与设备软件或与获取的内皮图像发生交互，即可提供有内皮密度和细胞形态标准(多形性和polymegatism)的估计值。(3) Automated corneal endothelial cytoscope: This type of device provides a source of endothelial analysis without interaction with device software or with acquired endothelial images, providing endothelial density and cell morphology criteria (pleomorphism and polymegatism) estimated value.
现有的一些设备室利用自动角膜内皮细胞镜摄取图像，然后通过这些图像来分析角膜内皮细胞的特征参数，其包括：步骤一、自动定位眼球位置坐标；步骤二、采集角膜内皮细胞图像；步骤三、对角膜内皮细胞图像进行处理分析，得出角膜内皮细胞特征参数。但是其细胞识别率不够，容易导致医生误判，对诊断眼疾具有很大影响。Some existing equipment rooms use automatic corneal endothelial cell mirrors to capture images, and then use these images to analyze the characteristic parameters of corneal endothelial cells, which includes: step 1, automatically positioning the eyeball position coordinates; step 2, collecting corneal endothelial cell images; 3. Process and analyze the corneal endothelial cell image to obtain the characteristic parameters of the corneal endothelial cell. However, its cell recognition rate is not enough, which can easily lead to misjudgment by doctors, which has a great impact on the diagnosis of eye diseases.
“角膜地形测量中眼球定位跟踪若干技术的研究”一文中，介绍了一些眼球定位跟踪方法，其研究了一种实时对眼球进行自动定位跟踪的技术，用正交的两个线阵CCD实时捕捉眼球的黑白图像，在输出信号中用硬件实现图像信号二值化并提取出瞳孔图像，通过单片机和CPLD对瞳孔图像进行处理，包括计算瞳孔中心与光路中心的偏差，控制步进电机调整光路中心位置使其与瞳孔中心重合等。其定位方法不够精细，判断出的瞳孔中心不一定准确，线阵CCD捕捉到的图像有可能出现偏差，导致后续分析出现误判。In the article "Research on Several Eyeball Positioning and Tracking Techniques in Corneal Topography", some eyeball positioning and tracking methods are introduced. It studies a real-time automatic positioning and tracking technology for eyeballs, which are captured in real time by two orthogonal linear array CCDs. The black-and-white image of the eyeball uses hardware to realize binarization of the image signal in the output signal and extracts the pupil image, and processes the pupil image through the single-chip microcomputer and CPLD, including calculating the deviation between the pupil center and the center of the optical path, and controlling the stepping motor to adjust the center of the optical path Position it so that it coincides with the center of the pupil, etc. The positioning method is not precise enough, the judged pupil center may not be accurate, and the image captured by the linear array CCD may be deviated, resulting in misjudgment in subsequent analysis.
公告号为CN100580702C、名称为“角膜内皮细胞图像处理方法”的专利公开了一种角膜内皮细胞图像处理方法，其能很好的恢复细胞图像边界，使图像边界更清晰，但是其在处理过程中需要预先设定经验阈值，如果设定的经验阈值不同，其得到的诊断结果很可能不同，通用性和适用性较差。The patent with the notification number CN100580702C and titled "Corneal Endothelial Cell Image Processing Method" discloses a corneal endothelial cell image processing method, which can well restore the cell image boundary and make the image boundary clearer, but it is not clear during the processing process. The empirical threshold needs to be set in advance. If the set empirical threshold is different, the diagnostic results obtained are likely to be different, and the versatility and applicability are poor.
发明内容Contents of the invention
本发明的目的是提供一种角膜内皮细胞分析仪及分析方法，对目标个体的眼球的角膜内皮细胞进行图像采集、图像处理分析，以提高其细胞识别率、通用性和适用性，使分析结果更加准确。The purpose of the present invention is to provide a corneal endothelial cell analyzer and analysis method, which can perform image acquisition, image processing and analysis on the corneal endothelial cells of the eyeball of the target individual, so as to improve its cell recognition rate, versatility and applicability, and make the analysis results more precise.
本发明所述的角膜内皮细胞分析仪，包括定位采集摄像头、红外距离传感器、第一、第二、第三步进电机、分析仪主机、打印机、触摸式液晶显示器、键盘和鼠标；所述键盘、鼠标都与分析仪主机连接，进行基本信息输入和手动操作；所述定位采集摄像头与分析仪主机连接，接受分析仪主机的控制并将采集到的眼球定位图像和角膜内皮细胞图像送入分析仪主机内处理；所述红外距离传感器与分析仪主机连接，将检测的眼球到定位采集摄像头的距离送入分析仪主机内处理；所述分析仪主机通过驱动电路与第一、第二、第三步进电机连接，根据处理后得到的眼球位置坐标控制第一、第二、第三步进电机转动，带动定位采集摄像头对准眼球并抓拍角膜内皮细胞图像；所述触摸式液晶显示器与分析仪主机连接，用于进行手动操作和显示处理分析后的角膜内皮细胞特征参数；所述打印机与分析仪主机连接，打印处理分析后的角膜内皮细胞特征参数。The corneal endothelial cell analyzer of the present invention includes a positioning acquisition camera, an infrared distance sensor, the first, second, and third stepping motors, an analyzer host, a printer, a touch-type liquid crystal display, a keyboard and a mouse; the keyboard , the mouse are all connected with the analyzer mainframe for basic information input and manual operation; the positioning acquisition camera is connected with the analyzer host, accepts the control of the analyzer host and sends the collected eyeball positioning images and corneal endothelial cell images for analysis Processing in the instrument host; the infrared distance sensor is connected with the analyzer host, and the distance from the detected eyeball to the positioning acquisition camera is sent to the analyzer host for processing; the analyzer host communicates with the first, second, and second through the drive circuit The three stepper motors are connected, and the first, second and third stepper motors are controlled to rotate according to the eyeball position coordinates obtained after processing, driving the positioning acquisition camera to align with the eyeball and capture images of corneal endothelial cells; the touch-type liquid crystal display and analysis The printer is connected to the host computer for manual operation and displaying the characteristic parameters of the processed and analyzed corneal endothelial cells; the printer is connected to the analyzer host computer for printing the processed and analyzed characteristic parameters of the corneal endothelial cells.
采用上述角膜内皮细胞分析仪对角膜内皮细胞进行分析的方法，包括：The method for analyzing corneal endothelial cells by using the above-mentioned corneal endothelial cell analyzer includes:
步骤一、自动定位眼球位置坐标；步骤二、定位采集摄像头采集角膜内皮细胞图像，并将其送入分析仪主机；步骤三、分析仪主机对角膜内皮细胞图像进行处理分析；步骤四、触摸式液晶显示器显示处理分析后的角膜内皮细胞特征参数，打印机打印处理分析后的角膜内皮细胞特征参数。Step 1: Automatic positioning of the coordinates of the eyeball position; Step 2: Positioning and acquisition The camera collects images of corneal endothelial cells and sends them to the host of the analyzer; Step 3: The host of the analyzer processes and analyzes the images of corneal endothelial cells; Step 4: Touch-type The liquid crystal display displays the characteristic parameters of the corneal endothelial cells after processing and analysis, and the printer prints the characteristic parameters of the corneal endothelial cells after processing and analysis.
其中，自动定位眼球位置坐标的方法为：Among them, the method of automatically locating the coordinates of the eyeball position is:
1)红外距离传感器检测眼球到定位采集摄像头的距离，并将其送入分析仪主机，延时t秒，等待图像光照平衡后，分析仪主机控制定位采集摄像头采集眼球定位图像，并获取该眼球定位图像，将空间按X、Y、Z三轴分为三个方向，以此时的定位采集摄像头的位置为原点(红外距离传感器与定位采集摄像头安装在同一位置，且能同时被移动，因此，该原点处也为红外距离传感器的位置)；1) The infrared distance sensor detects the distance from the eyeball to the positioning acquisition camera, and sends it to the main unit of the analyzer. After a delay of t seconds, after waiting for the image light balance, the main unit of the analyzer controls the positioning acquisition camera to collect the eyeball positioning image, and obtains the eyeball positioning image. Position the image, divide the space into three directions according to the X, Y, and Z axes, and take the position of the positioning acquisition camera at this time as the origin (the infrared distance sensor and the positioning acquisition camera are installed at the same position and can be moved at the same time, so , the origin is also the position of the infrared distance sensor);
2)对眼球定位图像进行变阈值二值化处理：先将眼球定位图像白平衡，再灰度化，再根据外界光照动态选取二值化阈值(即将一张图像分成若干个不同的区域，取出任意一块区域的每一个像素的像素值，并对选中区域的所有像素值进行加权平均，所得到的值即为选中区域的亮度值，最后根据亮度值的不同来选取二值化的阈值，阈值取亮度值的一半)进行二值化，再去噪，得到理想的二值化图像；2) Perform variable threshold binarization processing on the eyeball positioning image: first white balance the eyeball positioning image, then grayscale, and then dynamically select the binarization threshold according to the external light (that is, divide an image into several different regions, take out The pixel value of each pixel in any area, and the weighted average of all pixel values in the selected area, the obtained value is the brightness value of the selected area, and finally select the threshold value of binarization according to the difference in brightness value, the threshold value Take half of the brightness value) to perform binarization, and then denoise to obtain an ideal binarized image;
3)将该二值化图像分割成像素点后，存入一个二维数组，数组号对应每个像素点的x、y坐标；3) After the binarized image is divided into pixels, it is stored in a two-dimensional array, and the array number corresponds to the x and y coordinates of each pixel;
4)对二维数组内的像素点进行逐个扫描，当任意数组元素数据变化时，判定此点为眼球边缘，正向扫描得到眼球的上边缘(x1,y1)，反向扫描得到眼球的下边缘(x2,y2)，计算得到眼球在Y轴上的对应位置；4) Scan the pixels in the two-dimensional array one by one. When the data of any array element changes, it is judged that this point is the edge of the eyeball. The upper edge (x1, y1) of the eyeball is obtained by scanning forward, and the lower edge of the eyeball is obtained by scanning backward. edge(x2,y2), compute Get the corresponding position of the eyeball on the Y axis;
5)从二维数组的第Y行开始，对得到的Y轴对应的像素点再进行扫描，正向扫描得到眼球的左边缘(x3,y3)，反向扫描得到眼球的右边缘(x4,y4)，计算得到眼球在X轴上的对应位置；5) Starting from row Y of the two-dimensional array, scan the pixels corresponding to the obtained Y axis again, scan forward to get the left edge of the eyeball (x3, y3), and scan backward to get the right edge of the eyeball (x4, y4), calculate Get the corresponding position of the eyeball on the X-axis;
6)对红外距离传感器采集到的数字量(即反应眼球到定位采集摄像头距离的输出电压)进行加权平均，除去随机噪声，根据该数字量与距离成线性关系的线性表达式6) Carry out weighted average of the digital quantity collected by the infrared distance sensor (that is, the output voltage reflecting the distance from the eyeball to the positioning acquisition camera), remove random noise, and according to the linear expression that the digital quantity has a linear relationship with the distance
u表示红外距离传感器输出的电压，z表示眼球到定位采集摄像头的距离，计算眼球在Z轴上的对应位置；u represents the voltage output by the infrared distance sensor, z represents the distance from the eyeball to the positioning acquisition camera, and calculates the corresponding position of the eyeball on the Z axis;
7)分析仪主机根据眼球在X、Y、Z轴上的对应位置，控制第一、第二、第三步进电机转动，带动定位采集摄像头对准眼球(即瞳孔中心)，完成自动定位。7) The host of the analyzer controls the rotation of the first, second, and third stepping motors according to the corresponding positions of the eyeballs on the X, Y, and Z axes, and drives the positioning acquisition camera to align with the eyeballs (that is, the center of the pupil) to complete automatic positioning.
角膜内皮细胞的图像处理分析主要有三种方式：There are three main methods for image processing and analysis of corneal endothelial cells:
第一种，全手动分析：即操作者可以先通过图像预处理得到细胞格图像，再由操作者自己重新绘制所有的细胞格线，将完成后的纯手工画出的细胞格图像通过设备再进行后续的处理和分析得出结果。The first one is full manual analysis: that is, the operator can first obtain the cell grid image through image preprocessing, and then redraw all the cell grid lines by the operator himself. Subsequent processing and analysis are carried out to obtain the results.
第二种，半自动分析：即操作者可以先通过图像预处理得到细胞格图像，再由操作者对部分细胞格图像进行手动干预处理，如画线增加漏掉的细胞格、擦除细胞线除去多识别的细胞格、重新选择需要分析图像范围等，最后再将手动干预后的细胞格图像通过设备进行后续的处理和分析得出结果。The second type is semi-automatic analysis: that is, the operator can first obtain cell grid images through image preprocessing, and then manually intervene on some cell grid images by the operator, such as drawing lines to add missing cell grids, erasing cell lines to remove Multi-recognized cell grids, reselecting the range of images that need to be analyzed, etc., and finally the cell grid images after manual intervention are processed and analyzed by the device to obtain the results.
第三种，全自动分析：即操作者将角膜内皮细胞图像通过算法直接进行处理分析得出结果。The third type is fully automatic analysis: that is, the operator directly processes and analyzes the corneal endothelial cell image through an algorithm to obtain the result.
本发明中分析仪主机对角膜内皮细胞图像进行处理分析的方法(属于全自动分析)为：The method (belonging to full-automatic analysis) that analyzer host computer carries out processing analysis to corneal endothelial cell image among the present invention is:
1)选取一幅RGB格式的角膜内皮细胞图像，将其转换成灰度图像；1) Select a corneal endothelial cell image in RGB format and convert it into a grayscale image;
2)通过直方图确认该灰度图像的灰度级的主要分布，得到灰度级分布曲线，然后对灰度图像做均衡化处理，均衡化处理之后再与原直方图进行匹配，得到与原直方图的灰度级分布曲线大致相同且在过渡段均匀分布的新直方图(即得到一幅亮度分布均匀的图像)；2) Confirm the main distribution of the gray level of the gray level image through the histogram, obtain the gray level distribution curve, and then perform equalization processing on the gray level image, and then match it with the original histogram after equalization processing to obtain the original The gray level distribution curve of the histogram is approximately the same and the new histogram is uniformly distributed in the transition section (that is, an image with uniform brightness distribution is obtained);
3)采用高斯滤波法去除新直方图中的高斯噪声；高斯滤波的基本思路是对高斯函数进行离散化，以离散点上的高斯函数值为权值，对灰度图像的每个像素点做一定范围邻域内的加权平均，即可有效消除高斯噪声，高斯滤波后的图像灰度值较之前变化很小；3) Gaussian filtering is used to remove Gaussian noise in the new histogram; the basic idea of Gaussian filtering is to Gaussian function Carry out discretization, take the Gaussian function on the discrete point as the weight, and do a weighted average within a certain range of neighborhoods for each pixel of the grayscale image, which can effectively eliminate Gaussian noise, and the grayscale value of the image after Gaussian filtering is relatively low Little change before;
4)采用均值滤波法去除新直方图中的椒盐噪声，均值滤波函数为，4) The salt-and-pepper noise in the new histogram is removed by the mean filtering method, and the mean filtering function is,
Ω＝{(x+1,y),(x+1,y+1),(x+1,y-1),(x-1,y),(x-1,y+1),(x-1,y-1),(x,y+1),(x,y-1)}，Ω={(x+1,y),(x+1,y+1),(x+1,y-1),(x-1,y),(x-1,y+1),( x-1,y-1),(x,y+1),(x,y-1)},
其基本思路是将一个像素点周围的8个点的灰度值的平均值作为该点滤波后的灰度值，也即：→滤波后x的灰度值变换为5，均值滤波后的图像边缘灰度值较之前变大，图像的中心区域灰度值较之前变小；The basic idea is to use the average of the gray values of the 8 points around a pixel as the filtered gray value of the point, that is: → After filtering, the gray value of x is converted to 5, the gray value of the edge of the image after mean filtering is larger than before, and the gray value of the central area of the image is smaller than before;
5)对新直方图进行二值化处理，创建一个与均值滤波后的图像大小相同的零矩阵，然后对高斯滤波后的图像与均值滤波后的图像的对应像素点的灰度值做比较，用Gi表示高斯滤波后的图像各点的灰度值，用Mj表示均值滤波后的图像各点的灰度值，当Gi＞Mj时，将零矩阵中对应的像素点赋值为1，当Gi≤Mj时，零矩阵中对应的像素点的值不变，得到一幅二进制图像；5) Binarize the new histogram, create a zero matrix with the same size as the mean-filtered image, then compare the gray values of the corresponding pixels between the Gaussian-filtered image and the mean-filtered image, Use Gi to represent the gray value of each point in the image after Gaussian filtering, use Mj to represent the gray value of each point in the image after mean filtering, when Gi >Mj , assign the corresponding pixel in the zero matrix to 1. When Gi ≤ Mj , the value of the corresponding pixel in the zero matrix remains unchanged, and a binary image is obtained;
6)细胞孔洞填充，填充区域是相对一个封闭的区域而言，选定1为前景色，0为背景色，在封闭区域中将背景色填充为前景色；6) Cell hole filling, the filling area is relative to a closed area, select 1 as the foreground color, 0 as the background color, and fill the background color as the foreground color in the closed area;
7)对图像做形态学处理，建立形态学结构元素：然后再对二值化矩阵做开运算、腐蚀，得到边缘光滑且没有毛刺的图像；7) Perform morphological processing on the image to establish morphological structural elements: Then open and corrode the binary matrix to obtain an image with smooth edges and no burrs;
8)对经形态学处理后的图像做细化处理，分别轮换使用上、下、左、右、左上、右上、左下、右下，这八个方向的模板削去一层像素，直到图像不再变化为止，得到清楚的图像轮廓；8) Thinning the morphologically processed image, respectively using up, down, left, right, upper left, upper right, lower left, and lower right in rotation. The templates in these eight directions cut off a layer of pixels until the image is no longer Until it changes again, a clear image outline is obtained;
9)将细化后的图像中值等于1的点的位置映射到原RGB格式的角膜内皮细胞图像中的R通道，将这些位置上的值都赋为255，将G、B两个通道的值与细化后的图像值对应进行点乘，得到边缘着色的图像；9) Map the position of the point whose value is equal to 1 in the thinned image to the R channel in the corneal endothelial cell image in the original RGB format, assign the values at these positions to 255, and set the G and B channels value and the thinned image value correspond to dot multiplication to obtain an edge-colored image;
10)先统计细化后且边缘着色的图像的细胞面积与细胞边数，然后对不同面积范围与不同多边形进行上色，对于区别不同多边形，采用交叉点的个数进行判断，统计一个3×3邻域内0,1交替出现的个数p，当p等于阈值threshold时，以该邻域的中心像素点为交叉点(这里的threshold使用的是动态阈值，根据每个不同连通域的大小来设定的)。10) First count the cell area and the number of cell edges of the thinned and edge-colored image, and then color different area ranges and different polygons. For different polygons, use the number of intersections to judge, and count a 3× 3 The number p of 0 and 1 alternately appearing in the neighborhood. When p is equal to the threshold threshold, the center pixel point of the neighborhood is used as the intersection point (the threshold here uses a dynamic threshold, which is determined according to the size of each different connected domain. set).
本发明具有如下效果：角膜内皮细胞分析仪采用一体机设计，操作便捷，数据保存安全；自动定位眼球的方法，以机器视觉的方式，通过第一、第二、第三步进电机以及红外距离传感器、定位采集摄像头的配合实现了眼球的准确定位，为定位采集摄像头采集精确的角膜内皮细胞图像奠定了基础；分析仪主机对角膜内皮细胞图像进行处理分析中不需要预先设定经验阈值，其通用性和适用性较高，并且细胞识别率和计数准确率较高，医生根据其分析结果能准确的做出判断。The present invention has the following effects: the corneal endothelial cell analyzer adopts an all-in-one design, which is convenient to operate and safe in data storage; the method for automatically locating the eyeball, in the form of machine vision, through the first, second and third stepping motors and infrared distance The cooperation of the sensor and the positioning acquisition camera realizes the accurate positioning of the eyeball, and lays the foundation for the positioning acquisition camera to collect accurate corneal endothelial cell images; the host computer of the analyzer does not need to pre-set the empirical threshold when processing and analyzing the corneal endothelial cell images. The versatility and applicability are high, and the cell recognition rate and counting accuracy are high, and doctors can make accurate judgments based on the analysis results.
附图说明Description of drawings
图1为本发明的系统原理框图。Fig. 1 is the functional block diagram of the system of the present invention.
图2为本发明在自动定位眼球位置坐标过程中确定眼球在X、Y轴上对应位置的流程图。Fig. 2 is a flow chart of determining the corresponding position of the eyeball on the X and Y axes in the process of automatically locating the coordinates of the eyeball position according to the present invention.
图3为本发明中自动采集眼球角膜内皮细胞图像的流程图。Fig. 3 is a flow chart of automatically collecting images of eyeball corneal endothelial cells in the present invention.
图4为本发明中角膜内皮细胞图像处理流程框图。Fig. 4 is a flow chart of corneal endothelial cell image processing in the present invention.
图5为采集到的角膜内皮细胞图像。Figure 5 is the collected images of corneal endothelial cells.
图6为角膜内皮细胞图像经形态学处理(即图像增强)后的效果图。Fig. 6 is an effect diagram of corneal endothelial cell images after morphological processing (ie, image enhancement).
图7为角膜内皮细胞图像经细化处理后的效果图。Fig. 7 is an effect diagram of the corneal endothelial cell image after thinning processing.
图8为角膜内皮细胞图像的边缘着色图。Fig. 8 is an edge coloring map of an image of corneal endothelial cells.
图9为按角膜内皮细胞的边数着色图。Figure 9 is a graph colored by the number of sides of corneal endothelial cells.
图10为按角膜内皮细胞面积范围着色图。Figure 10 is a coloring map according to the area range of corneal endothelial cells.
具体实施方式detailed description
下面结合附图对本发明作详细说明。The present invention will be described in detail below in conjunction with the accompanying drawings.
如图1至图4所示的角膜内皮细胞分析仪，包括定位采集摄像头1、红外距离传感器2、第一步进电机3、第二步进电机4、第三步进电机5、分析仪主机6、打印机7、触摸式液晶显示器8、键盘9和鼠标10，第一步进电机3、第二步进电机4、第三步进电机5都与定位采集摄像头1机械连接，带动定位采集摄像头1移动，红外距离传感器2与定位采集摄像头1安装在同一位置，能与定位采集摄像头1一起被带动移动。键盘9、鼠标10都与分析仪主机6连接，进行基本信息输入和手动操作；定位采集摄像头1与分析仪主机6连接，接受分析仪主机6的控制并将采集到的眼球定位图像和角膜内皮细胞图像送入分析仪主机6内处理；红外距离传感器2与分析仪主机6连接，将检测的眼球到定位采集摄像头的距离送入分析仪主机6内处理；分析仪主机6通过驱动电路11与第一步进电机3、第二步进电机4、第三步进电机5连接，根据处理后得到的眼球位置坐标控制第一步进电机3、第二步进电机4、第三步进电机5转动，带动定位采集摄像头1对准眼球并抓拍角膜内皮细胞图像；触摸式液晶显示器8与分析仪主机6连接，用于进行手动操作和显示处理分析后的角膜内皮细胞特征参数；打印机7与分析仪主机6连接，打印处理分析后的角膜内皮细胞特征参数。The corneal endothelial cell analyzer shown in Figures 1 to 4 includes a positioning acquisition camera 1, an infrared distance sensor 2, a first stepping motor 3, a second stepping motor 4, a third stepping motor 5, and an analyzer host 6, printer 7, touch-type liquid crystal display 8, keyboard 9 and mouse 10, the first stepper motor 3, the second stepper motor 4, the third stepper motor 5 are all mechanically connected with the positioning acquisition camera 1, drive the positioning acquisition camera 1 moves, the infrared distance sensor 2 is installed at the same position as the positioning acquisition camera 1, and can be driven to move together with the positioning acquisition camera 1. The keyboard 9 and the mouse 10 are all connected to the analyzer host 6 for basic information input and manual operation; the positioning acquisition camera 1 is connected to the analyzer host 6 to accept the control of the analyzer host 6 and collect the eye positioning images and corneal endothelial images. The cell image is sent to the analyzer host 6 for processing; the infrared distance sensor 2 is connected to the analyzer host 6, and the distance from the detected eyeball to the positioning acquisition camera is sent to the analyzer host 6 for processing; the analyzer host 6 communicates with the analyzer host 6 through the drive circuit 11. The first stepping motor 3, the second stepping motor 4, and the third stepping motor 5 are connected, and the first stepping motor 3, the second stepping motor 4, and the third stepping motor are controlled according to the eyeball position coordinates obtained after processing 5 to rotate, drive the positioning acquisition camera 1 to align with the eyeball and capture corneal endothelial cell images; the touch-type liquid crystal display 8 is connected to the analyzer host 6 for manual operation and display of the corneal endothelial cell characteristic parameters after processing and analysis; the printer 7 and The analyzer host 6 is connected to print and process the analyzed characteristic parameters of the corneal endothelial cells.
角膜内皮细胞分析仪还具有如下功能：在机身上还设手动调节按钮，通过手动调节来移动定位采集摄像头；其可以手动标记当前采集的是左眼或右眼；可以选择手动控制拍照或者自动控制拍照；可以设有三种级别的闪光灯，自动调节、高光、低光；设瞳孔窗口用来参照是否对准了中心位置；设清除键，用来清除已拍摄的所有图像；支持手动选择分析的图像，可以选择单张左眼，单张右眼，一张左眼和一张右眼两张；支持图像质量推荐图像功能，即是可以自动选择质量最好的图像；处理双眼图时，设有左/右眼切换功能，以通过该键切换待分析的左右眼图像；设有图像选择功能，用于重新为待分析的左右眼选择图像；可支持全自动图像处理和半自动图像处理；设有分析区域选择的功能，可以重新划定分析区域；设有橡皮擦功能，可以对分析区域的结果进行修正，擦去多余的细胞格；设有画笔功能，可以对分析区域的结果进行修正，划分为更多的细胞格；多模式显示分析结果，多模式显示分析图像，多方式显示分析数据。The corneal endothelial cell analyzer also has the following functions: there is also a manual adjustment button on the body, and the positioning and acquisition camera can be moved and positioned by manual adjustment; it can manually mark whether the current acquisition is the left eye or the right eye; you can choose to manually control the camera or automatically Control photography; can set three levels of flash, automatic adjustment, high light, low light; set the pupil window to refer to whether it is aligned with the center position; set a clear key to clear all the images that have been taken; support manual selection of analysis For images, you can choose a single left-eye image, a single right-eye image, one left-eye image and one right-eye image; it supports image quality recommended image function, that is, the image with the best quality can be automatically selected; when processing a binocular image, set There is a left/right eye switching function, so that the left and right eye images to be analyzed can be switched through this key; there is an image selection function, which is used to reselect images for the left and right eyes to be analyzed; it can support fully automatic image processing and semi-automatic image processing; With the function of selecting the analysis area, you can redefine the analysis area; with an eraser function, you can correct the results of the analysis area and erase redundant cells; with a brush function, you can correct the results of the analysis area. Divide into more cells; display analysis results in multiple modes, display analysis images in multiple modes, and display analysis data in multiple modes.
采用上述角膜内皮细胞分析仪对角膜内皮细胞进行分析的方法为：先通过键盘9输入基本信息(包括病人ID、性别、出生日期等)，并输入目标个体(可利用定位采集摄像头1拍摄一张目标个体的头像，也可采用静态的匿名头像)，然后进入拍摄界面，再次确认目标个体眼睛(对应于瞳孔中心)的位置，由分析仪主机6内的内皮分析单元发送拍照指令给其内的主控制单元，即调用主控制单元提供的接口函数，最后由该主控制单元去完成整个采集、分析过程。The method for analyzing corneal endothelial cells by using the above-mentioned corneal endothelial cell analyzer is: first input basic information (including patient ID, gender, date of birth, etc.) The avatar of the target individual can also be a static anonymous avatar), and then enter the shooting interface to confirm the position of the target individual's eyes (corresponding to the center of the pupil) again, and the endothelial analysis unit in the analyzer host 6 sends a photographing instruction to the The main control unit calls the interface function provided by the main control unit, and finally the main control unit completes the entire collection and analysis process.
具体包括如下步骤：Specifically include the following steps:
步骤一、自动定位眼球位置坐标。主要是调用自动定位算法，如图2、图3所示，其自动定位方法(算法)为：Step 1. Automatically locate the eyeball position coordinates. It mainly calls the automatic positioning algorithm, as shown in Figure 2 and Figure 3, the automatic positioning method (algorithm) is:
1)红外距离传感器2检测眼球到定位采集摄像头1的距离，并将其送入分析仪主机6，延时3秒，等待图像光照平衡后，分析仪主机6控制定位采集摄像头1采集眼球定位图像，并获取该眼球定位图像，将空间按X、Y、Z三轴分为三个方向，以此时的定位采集摄像头1和红外距离传感器2的位置为原点。1) The infrared distance sensor 2 detects the distance from the eyeball to the positioning acquisition camera 1, and sends it to the analyzer host 6, and delays for 3 seconds. After waiting for the image light balance, the analyzer host 6 controls the positioning acquisition camera 1 to collect eyeball positioning images , and obtain the eyeball positioning image, divide the space into three directions according to the X, Y, and Z axes, and take the position of the positioning acquisition camera 1 and the infrared distance sensor 2 as the origin at this time.
2)对眼球定位图像进行变阈值二值化处理：先将眼球定位图像白平衡，再灰度化，再根据外界光照动态选取二值化阈值(即将一张图像分成若干个不同的区域，取出任意一块区域的每一个像素的像素值，并对选中区域的所有像素值进行加权平均，所得到的值即为选中区域的亮度值，最后根据亮度值的不同来选取二值化的阈值，阈值取亮度值的一半)进行二值化，再去噪，得到理想的640*480的二值化图像，有眼球的部分为黑色，背景为白色，采用变阈值二值化的方法可以消除环境光照对二值化的影响。2) Perform variable threshold binarization processing on the eyeball positioning image: first white balance the eyeball positioning image, then grayscale, and then dynamically select the binarization threshold according to the external light (that is, divide an image into several different regions, take out The pixel value of each pixel in any area, and the weighted average of all pixel values in the selected area, the obtained value is the brightness value of the selected area, and finally select the threshold value of binarization according to the difference in brightness value, the threshold value Take half of the luminance value) for binarization, and then denoise to obtain an ideal 640*480 binarized image. The part with the eyeball is black and the background is white. The method of variable threshold binarization can eliminate ambient light Effect on binarization.
3)将640*480的二值化图像分割成4800个8*8像素点后，存入一个二维数组，数组号对应每个像素点的x、y坐标；3) After dividing the binary image of 640*480 into 4800 8*8 pixel points, store it in a two-dimensional array, and the array number corresponds to the x and y coordinates of each pixel;
4)对二维数组内的像素点进行逐个扫描，当任意数组元素数据变化时，判定此点为眼球边缘(即判断相邻坐标点是否有像素值，如果没有，则此点为噪声，返回继续扫描，如果有则判断此点为眼球边缘)，正向扫描得到眼球的上边缘(x1,y1)，反向扫描得到眼球的下边缘(x2,y2)，计算得到眼球在Y轴上的对应位置；4) Scan the pixels in the two-dimensional array one by one. When the data of any array element changes, determine this point as the edge of the eyeball (that is, judge whether there is a pixel value at the adjacent coordinate point, if not, then this point is noise, return Continue to scan, if there is, judge that this point is the edge of the eyeball), scan forward to get the upper edge of the eyeball (x1, y1), scan backward to get the lower edge of the eyeball (x2, y2), calculate Get the corresponding position of the eyeball on the Y axis;
5)从二维数组的第Y行开始，对得到的Y轴对应的像素点再进行扫描，当任意数组元素数据变化时，判定此点为眼球边缘(即判断相邻坐标点是否有像素值，如果没有，则此点为噪声，返回继续扫描，如果有则判断此点为眼球边缘)，正向扫描得到眼球的左边缘(x3,y3)，反向扫描得到眼球的右边缘(x4,y4)，计算得到眼球在X轴上的对应位置；5) Starting from the Yth row of the two-dimensional array, scan the pixel point corresponding to the obtained Y axis. When the data of any array element changes, determine this point as the edge of the eyeball (that is, determine whether the adjacent coordinate point has a pixel value , if not, then this point is noise, return to continue scanning, if there is, this point is judged to be the edge of the eyeball), scan forward to get the left edge of the eyeball (x3, y3), and scan backward to get the right edge of the eyeball (x4, y4), calculate Get the corresponding position of the eyeball on the X-axis;
6)对红外距离传感器2采集到的数字量(即反应眼球到定位采集摄像头距离的输出电压)进行加权平均，除去随机噪声，根据该数字量与距离成线性关系的线性表达式6) Carry out weighted average of the digital quantity collected by the infrared distance sensor 2 (that is, the output voltage reflecting the distance from the eyeball to the positioning acquisition camera), remove random noise, and according to the linear expression of the linear relationship between the digital quantity and the distance
u表示红外距离传感器输出的电压，z表示眼球到定位采集摄像头的距离，计算眼球在Z轴上的对应位置；u represents the voltage output by the infrared distance sensor, z represents the distance from the eyeball to the positioning acquisition camera, and calculates the corresponding position of the eyeball on the Z axis;
7)分析仪主机6根据眼球在X、Y轴上的对应位置，控制第一步进电机3、第二步进电机4带动定位采集摄像头1和红外距离传感器2移动到对应的X、Y轴位置坐标，分析仪主机6根据眼球在Z轴上的对应位置，判断该对应位置到定位采集摄像头1的距离是否等于定位采集摄像头1的最佳焦距(为事先设定的定位采集摄像头1的最佳阈值)，如果不等于，则分析仪主机6控制第三步进电机5带定位采集摄像头1和红外距离传感器2在Z轴方向移动(相当于调整Z轴的坐标)，红外距离传感器2同时检测定位采集摄像头1到眼球(即瞳孔中心)的距离，如果该距离等于定位采集摄像头1的最佳焦距(相当于Z轴坐标达到最佳焦距的坐标)，则对焦准确，完成自动定位(即对准眼球)。7) Analyzer host 6 controls the first stepper motor 3 and the second stepper motor 4 to drive the positioning and acquisition camera 1 and the infrared distance sensor 2 to move to the corresponding X and Y axes according to the corresponding positions of the eyeballs on the X and Y axes Position coordinates, the analyzer host 6 judges whether the distance from the corresponding position to the positioning acquisition camera 1 is equal to the optimal focal length of the positioning acquisition camera 1 according to the corresponding position of the eyeball on the Z axis (the maximum focal length of the positioning acquisition camera 1 set in advance) optimal threshold), if not equal, the analyzer host 6 controls the third stepping motor 5 with positioning acquisition camera 1 and infrared distance sensor 2 to move in the Z-axis direction (equivalent to adjusting the coordinates of the Z-axis), and the infrared distance sensor 2 simultaneously Detect the distance from the positioning acquisition camera 1 to the eyeball (i.e. the center of the pupil), if the distance is equal to the best focal length of the positioning acquisition camera 1 (equivalent to the coordinates at which the Z-axis coordinates reach the best focal length), then the focus is accurate and automatic positioning is completed (i.e. aimed at the eye).
步骤二、抓拍眼球图像。完成眼球自动定位以后，分析仪主机6再判定是否具有拍摄条件(即定位采集摄像头1是否对准瞳孔中心，且定位采集摄像头1到眼球的距离正好等于定位采集摄像头1的最佳焦距)，若具有拍摄条件，则发送指令采集图像，由定位采集摄像头完成眼球角膜内皮细胞图像的抓拍；若不具有拍摄条件，则重新自动定位眼球，再判定是否具有拍摄条件，直至最后拍摄完成。拍摄过程中采集到的角膜内皮细胞图像会存储在FLASH中，操作者可以选择采集一张角膜内皮细胞图像，也可以选择采集多张角膜内皮细胞图像，采集到的角膜内皮细胞图像被送入分析仪主机6内，分析仪主机6对该图像文件统一命名并保存(可通过触摸式液晶显示器显示)。Step 2, capturing eyeball images. After completing the automatic positioning of the eyeball, the analyzer host 6 determines whether there is a shooting condition (that is, whether the positioning acquisition camera 1 is aimed at the center of the pupil, and the distance from the positioning acquisition camera 1 to the eyeball is just equal to the best focal length of the positioning acquisition camera 1), if If the shooting conditions are met, an instruction will be sent to collect images, and the positioning and acquisition camera will complete the capture of the corneal endothelial cell image of the eyeball; if the shooting conditions are not met, the eyeball will be re-positioned automatically, and then it will be determined whether the shooting conditions are met until the final shooting is completed. The corneal endothelial cell images collected during the shooting process will be stored in FLASH. The operator can choose to collect one corneal endothelial cell image, or choose to collect multiple corneal endothelial cell images, and the collected corneal endothelial cell images will be sent for analysis. In the instrument host 6, the analyzer host 6 uniformly names and saves the image file (can be displayed by a touch-type liquid crystal display).
步骤三、分析仪主机6对角膜内皮细胞图像进行处理分析。具体包括三个部分：Step 3, the analyzer host 6 processes and analyzes the corneal endothelial cell image. Specifically, it includes three parts:
第一部分：选择分析的图像文件。根据拍摄的情况图像文件只有左右眼两大类，若是选择同类照片分析，每次只能选择一张进行分析；若是选择两类照片分析，每次可以选择左右眼各一张进行分析，分析系统提供了两种图像文件的选择方式，一种是由操作者手动选择，另一种是系统推荐。Part 1: Select the image file for analysis. According to the shooting situation, there are only two types of image files, left and right eyes. If you choose the same type of photo analysis, you can only select one for analysis each time; if you choose two types of photos for analysis, you can choose one for left and right eyes for analysis each time. Two ways of selecting image files are provided, one is manually selected by the operator, and the other is recommended by the system.
第二部分：系统中保存的图像超过上限时，操作者仍可以继续获取图像文件，系统将自动覆写掉序列中最旧时间的图像，这一步可以由操作者决定要不要进行。The second part: When the images stored in the system exceed the upper limit, the operator can still continue to acquire image files, and the system will automatically overwrite the oldest image in the sequence. This step can be decided by the operator or not.
第三部分：处理分析所选的角膜内皮细胞图像文件。在需要分析的角膜内皮细胞图像被选定以后，直接进入到分析工具，分析仪提供两种处理模式，一种是全自动处理，一种是半自动处理，具体处理方式由操作者选择，对角膜内皮细胞图像进行第一次分析以后，可以根据分析得到的细胞格图判定是否进行人为干预(人为干预实际上是分析仪提供的可以对待分析的图像文件进行分析范围的勾选、细胞线的擦除、细胞线的添加的功能)。经过人工干预后的图像将继续通过图像处理算法进行处理分析，如图4所示，其自动处理方法(算法)为：The third part: process and analyze the selected corneal endothelial cell image files. After the corneal endothelial cell image to be analyzed is selected, it directly enters the analysis tool. The analyzer provides two processing modes, one is fully automatic processing and the other is semi-automatic processing. The specific processing method is selected by the operator. After the first analysis of the endothelial cell image, it is possible to determine whether to perform human intervention based on the analyzed cell grid (human intervention is actually the ticking of the analysis range provided by the analyzer, the wiping of the cell line, etc. function of addition and addition of cell lines). The image after manual intervention will continue to be processed and analyzed through the image processing algorithm, as shown in Figure 4, the automatic processing method (algorithm) is:
1)选取(输入)一幅RGB格式的角膜内皮细胞图像(如图5所示)，将其转换成灰度图像；1) select (input) a corneal endothelial cell image in RGB format (as shown in Figure 5), and convert it into a grayscale image;
2)通过直方图确认该灰度图像的灰度级的主要分布，得到灰度级分布曲线，然后对灰度图像做均衡化处理，均衡化处理之后再与原直方图进行匹配，得到与原直方图的灰度级分布曲线大致相同且在过渡段均匀分布的新直方图，即是得到一幅亮度分布均匀的图像；2) Confirm the main distribution of the gray level of the gray level image through the histogram, obtain the gray level distribution curve, and then perform equalization processing on the gray level image, and then match it with the original histogram after equalization processing to obtain the original The gray level distribution curve of the histogram is approximately the same and the new histogram is uniformly distributed in the transition section, that is, an image with uniform brightness distribution is obtained;
3)采用高斯滤波法去除新直方图中的高斯噪声；高斯滤波的基本思路是对高斯函数进行离散化，以离散点上的高斯函数值为权值，对灰度图像的每个像素点做一定范围邻域内的加权平均，即可有效消除高斯噪声，高斯滤波后的图像灰度值较之前变化很小；3) Gaussian filtering is used to remove Gaussian noise in the new histogram; the basic idea of Gaussian filtering is to Gaussian function Carry out discretization, take the Gaussian function on the discrete point as the weight, and do a weighted average within a certain range of neighborhoods for each pixel of the grayscale image, which can effectively eliminate Gaussian noise, and the grayscale value of the image after Gaussian filtering is relatively low Little change before;
4)采用均值滤波法去除新直方图中的椒盐噪声；均值滤波函数为，4) The salt-and-pepper noise in the new histogram is removed by the mean filtering method; the mean filtering function is,
Ω＝{(x+1,y),(x+1,y+1),(x+1,y-1),(x-1,y),(x-1,y+1),(x-1,y-1),(x,y+1),(x,y-1)}，Ω={(x+1,y),(x+1,y+1),(x+1,y-1),(x-1,y),(x-1,y+1),( x-1,y-1),(x,y+1),(x,y-1)},
将一个像素点周围的8个点的灰度值的平均值作为该点滤波后的灰度值，也即：The average value of the gray values of the 8 points around a pixel point is used as the filtered gray value of the point, that is:
→滤波后x的灰度值变换为5，均值滤波后的图像边缘灰度值较之前变大，图像的中心区域灰度值较之前变小； → After filtering, the gray value of x is converted to 5, the gray value of the edge of the image after mean filtering is larger than before, and the gray value of the central area of the image is smaller than before;
5)对新直方图进行二值化处理，创建一个与滤波图像大小相同的零矩阵(即全黑的图像)，然后对高斯滤波后的图像与均值滤波后的图像的对应像素点的灰度值做比较，用Gi表示高斯滤波后的图像各点的灰度值，用Mj表示均值滤波后的图像各点的灰度值，当Gi＞Mj时，将零矩阵中对应的像素点赋值为1，当Gi≤Mj时，零矩阵中对应的像素点的值不变，得到一幅二进制图像；5) Perform binarization on the new histogram, create a zero matrix with the same size as the filtered image (that is, a completely black image), and then compare the grayscale of the corresponding pixels of the Gaussian filtered image and the mean filtered image Values for comparison, use Gi to represent the gray value of each point in the image after Gaussian filtering, use Mj to represent the gray value of each point in the image after mean filtering, when Gi >Mj , the corresponding zero matrix The pixel is assigned a value of 1. When Gi ≤ Mj , the value of the corresponding pixel in the zero matrix remains unchanged, and a binary image is obtained;
6)细胞孔洞填充，填充区域(细胞)是相对一个封闭的区域而言，选定1为前景色，0为背景色，在封闭区域中将背景色填充为前景色；6) Cell hole filling, the filling area (cell) is relative to a closed area, select 1 as the foreground color, 0 as the background color, and fill the background color as the foreground color in the closed area;
7)对图像做形态学处理，建立形态学结构元素：然后再对二值化矩阵做开运算、腐蚀，得到边缘光滑且没有毛刺的图像(即增强后的图像)；7) Perform morphological processing on the image to establish morphological structural elements: Then open and corrode the binary matrix to obtain an image with smooth edges and no burrs (that is, the enhanced image);
8)对经形态学处理后的图像做细化处理，分别轮换使用上、下、左、右、左上、右上、左下、右下，这八个方向的模板削去一层像素，直到图像不再变化为止，得到清楚的图像轮廓(即边界提取图)；8) Thinning the morphologically processed image, respectively using up, down, left, right, upper left, upper right, lower left, and lower right in rotation. The templates in these eight directions cut off a layer of pixels until the image is no longer Until it changes again, a clear image outline (that is, the boundary extraction map) is obtained;
9)将细化后的图像中值等于1的点的位置映射到原RGB格式的角膜内皮细胞图像中的R通道，将这些位置上的值都赋为255，将G、B两个通道的值与细化后的图像值对应进行点乘，得到边缘着色的图像；9) Map the position of the point whose value is equal to 1 in the thinned image to the R channel in the corneal endothelial cell image in the original RGB format, assign the values at these positions to 255, and set the G and B channels value and the thinned image value correspond to dot multiplication to obtain an edge-colored image;
10)先统计细化后且边缘着色的图像的细胞面积与细胞边数，然后对不同面积范围与不同多边形进行上色，对于区别不同多边形，采用交叉点的个数进行判断，统计一个3×3邻域内0,1交替出现的个数p，当p等于阈值threshold时，以该邻域的中心像素点为交叉点(这里的threshold使用的是动态阈值，根据每个不同连通域的大小来设定的)。10) First count the cell area and the number of cell edges of the thinned and edge-colored image, and then color different area ranges and different polygons. For different polygons, use the number of intersections to judge, and count a 3× 3 The number p of 0 and 1 alternately appearing in the neighborhood. When p is equal to the threshold threshold, the center pixel point of the neighborhood is used as the intersection point (the threshold here uses a dynamic threshold, which is determined according to the size of each different connected domain. set).
步骤四、触摸式液晶显示器8显示处理分析后的角膜内皮细胞特征参数，图6、图7、图8、图9、图10是经处理的角膜内皮细胞图，等到所有的图像都处理分析完毕之后，将结果显示出来，具体包括统计到的细胞特征参数，如细胞的总数、细胞多晶形态(顶点数)的细胞个数、细胞多形变(面积)的细胞个数、细胞面积的标准差、细胞变异系数、细胞中细胞的最大面积、细胞中细胞的最小面积、属于正规六边形的细胞数、分析的细胞个数、所有识别出来的细胞的密度等。其中，对于多形变参数和多晶态参数，是以直方图的方式来显示，通过使用不同颜色以示区分。Step 4, the touch-type liquid crystal display 8 displays the corneal endothelial cell characteristic parameters after processing and analysis, Fig. 6, Fig. 7, Fig. 8, Fig. 9, Fig. 10 are the corneal endothelial cell diagrams processed, wait until all images are processed and analyzed After that, the results are displayed, including the statistical cell characteristic parameters, such as the total number of cells, the number of cells with polymorphic morphology (number of vertices), the number of cells with polymorphic shape (area), and the standard deviation of cell area , cell variation coefficient, maximum area of cells in cells, minimum area of cells in cells, number of cells belonging to regular hexagons, number of analyzed cells, density of all identified cells, etc. Among them, the multi-deformation parameters and polycrystalline state parameters are displayed in the form of histograms, and different colors are used to distinguish them.
步骤五、打印机7打印处理分析后的角膜内皮细胞特征参数。通过分析仪主机的界面中的<打印>功能按钮将数据输出到其内置的打印机(为热敏打印机)，进行打印。Step 5, the printer 7 prints and processes the analyzed characteristic parameters of the corneal endothelial cells. Output the data to its built-in printer (a thermal printer) through the <Print> function button in the interface of the analyzer host for printing.
完成上述的步骤之后，既可对下一个目标个体进行相同的操作流程，也可以对同一个目标个体进行多次操作，不同的分析结果会给操作者提供一定的参考依据。After completing the above steps, the same operation process can be performed on the next target individual, or multiple operations can be performed on the same target individual. Different analysis results will provide certain references for the operator.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410215647.6ACN103955941B (en) | 2014-05-21 | 2014-05-21 | A kind of specularmicroscope and the method for analysis |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410215647.6ACN103955941B (en) | 2014-05-21 | 2014-05-21 | A kind of specularmicroscope and the method for analysis |
| Publication Number | Publication Date |
|---|---|
| CN103955941A CN103955941A (en) | 2014-07-30 |
| CN103955941Btrue CN103955941B (en) | 2016-08-17 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410215647.6AActiveCN103955941B (en) | 2014-05-21 | 2014-05-21 | A kind of specularmicroscope and the method for analysis |
| Country | Link |
|---|---|
| CN (1) | CN103955941B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112330610B (en)* | 2020-10-21 | 2024-03-29 | 郑州诚优成电子科技有限公司 | Accurate positioning method based on microvascular position cornea endothelial cell counting acquisition |
| CN112200873B (en)* | 2020-10-21 | 2023-12-22 | 郑州诚优成电子科技有限公司 | Cornea endothelial cell counter positioning photographing method based on iris characteristics |
| CN114693646B (en)* | 2022-03-31 | 2023-04-11 | 中山大学中山眼科中心 | Corneal endothelial cell active factor analysis method based on deep learning |
| CN117611585B (en)* | 2024-01-22 | 2024-06-18 | 天津医科大学眼科医院 | Eyeball wall deformation analysis method and system based on image processing |
| CN119006374A (en)* | 2024-07-25 | 2024-11-22 | 陕西省第二人民医院(陕西省老年病医院) | Brain disease category analysis method and system based on magnetic resonance |
| Publication number | Publication date |
|---|---|
| CN103955941A (en) | 2014-07-30 |
| Publication | Publication Date | Title |
|---|---|---|
| US11250580B2 (en) | Method, system and computer readable storage media for registering intraoral measurements | |
| CN109859203B (en) | Defect tooth image identification method based on deep learning | |
| CN110400289B (en) | Fundus image recognition method, fundus image recognition device, fundus image recognition apparatus, and fundus image recognition storage medium | |
| CN106774863B (en) | Method for realizing sight tracking based on pupil characteristics | |
| US9149179B2 (en) | System and method for identifying eye conditions | |
| CN103955941B (en) | A kind of specularmicroscope and the method for analysis | |
| CN113159227A (en) | Acne image recognition method, system and device based on neural network | |
| CN112967285A (en) | Chloasma image recognition method, system and device based on deep learning | |
| CN109360196A (en) | Method and device based on deep learning processing oral cavity radiation image | |
| CN109697719A (en) | A kind of image quality measure method, apparatus and computer readable storage medium | |
| CN110766656B (en) | Method, device, equipment and storage medium for screening fundus macular region abnormality | |
| CN104021382A (en) | Eye image collection method and system | |
| EP4381468A1 (en) | Intuitive intraoral scanning | |
| JP2023503781A (en) | Ocular fundus image recognition method, device and equipment | |
| JP2020140630A (en) | Gaze point estimation system, gaze point estimation method, gaze point estimation program, and information recording medium on which this is recorded. | |
| CN117475270B (en) | Lesion identification method and system for diabetic retinopathy image | |
| CN116974369A (en) | Method, system, equipment and storage medium for operating medical image in operation | |
| CN116725478A (en) | Analysis method of fundus refractive power and topography generation method | |
| CN113237881A (en) | Method and device for detecting specific cells and pathological section detection system | |
| JP2007252707A (en) | Image analysis apparatus and image analysis program | |
| CN110472546B (en) | A non-contact eye movement feature extraction device and method for infants and young children | |
| CN112329537A (en) | A detection method of feces formed components based on yolov3 algorithm | |
| WO2023014995A1 (en) | Intuitive intraoral scanning | |
| CN118196218B (en) | Fundus image processing method, device and equipment | |
| CN104966282A (en) | Image acquiring method and system for detecting single erythrocyte |
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract | Application publication date:20140730 Assignee:Chongqing Magic Cube Zhilian Technology Co.,Ltd. Assignor:Chongqing University of Technology Contract record no.:X2024980028102 Denomination of invention:A corneal endothelial cell analyzer and analysis method Granted publication date:20160817 License type:Common License Record date:20241202 | |
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract | Application publication date:20140730 Assignee:Chongqing Qing'an Energy Co.,Ltd. Assignor:Chongqing University of Technology Contract record no.:X2024980036672 Denomination of invention:A corneal endothelial cell analyzer and analysis method Granted publication date:20160817 License type:Common License Record date:20241230 |