WO2020211173A1 - 基于机器视觉的眼前节断层图像的图像特征提取方法 - Google Patents

基于机器视觉的眼前节断层图像的图像特征提取方法 Download PDF

Info

Publication number
WO2020211173A1
WO2020211173A1 PCT/CN2019/089776 CN2019089776W WO2020211173A1 WO 2020211173 A1 WO2020211173 A1 WO 2020211173A1 CN 2019089776 W CN2019089776 W CN 2019089776W WO 2020211173 A1 WO2020211173 A1 WO 2020211173A1
Authority
WO
WIPO (PCT)
Prior art keywords
anterior segment
images
tomographic images
iris
cornea
Prior art date
Application number
PCT/CN2019/089776
Other languages
English (en)
French (fr)
Inventor
黄锦海
于航
陈浩
陈世豪
梅晨阳
王俊杰
Original Assignee
温州医科大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 温州医科大学 filed Critical 温州医科大学
Publication of WO2020211173A1 publication Critical patent/WO2020211173A1/zh

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10072—Tomographic images
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30041—Eye; Retina; Ophthalmic

Definitions

  • the invention relates to an ophthalmic medical image processing method, in particular to an image feature extraction method of a tomographic image of the anterior segment based on machine vision.
  • the anterior segment is a part of the eyeball, including: the entire cornea, iris, ciliary body, anterior chamber, posterior chamber, lens suspensory ligament, angle of the chamber, part of the lens, peripheral vitreous, retina and extraocular muscle attachment points and conjunctiva.
  • the purpose of the present invention is to overcome the shortcomings and deficiencies of the prior art, and provide a method for extracting image features of tomographic images of the anterior segment based on machine vision.
  • the technical solution adopted by the present invention is as follows:
  • the image feature extraction method of the tomographic image of the anterior segment based on machine vision includes the following steps:
  • the present invention first performs gray-scale histogram statistics on the tomographic images of the anterior segment collected by the camera, and eliminates images that cannot be the anterior segment, and then performs gray-scale normalization according to the histogram to reduce environmental light
  • the influence of imaging quality is then roughly segmented by K-mean clustering algorithm.
  • the cornea, iris, and lens regions can be segmented, and then blob analysis can be performed.
  • the non-anterior segment image is screened out, and then Based on the rough boundary of the region, the fine boundary tracking in a fixed direction is performed to obtain the precise contours of the cornea, iris, and lens, which provides reliable basic data for the subsequent determination of clinical parameters of the anterior segment.
  • the invention can be applied to the processing of images collected by equipment with similar principles of slit lamp imaging.
  • Figure 1 is a schematic flow diagram of the present invention.
  • a method for extracting image features of tomographic images of the anterior segment based on machine vision includes the following steps:
  • Gray-level histogram statistics the maximum, minimum, average, and mean square error of the image can be obtained. If the average value of the degree is too small, too large, or the gray-scale variance is too small and other related parameters, the image of the non-anterior segment is roughly screened to reduce unnecessary processing;
  • K-mean is the simplest clustering method, which is to find K clustering centers through iteration and allocate all data To the nearest cluster center, the sum of squares of the distance between each point and its corresponding cluster center is the smallest.
  • K-mean algorithm is simple, it can usually obtain better clustering results. Its disadvantage is that the selection of the initial clustering center may lead to different clustering results, but because we have prior knowledge of the image of the anterior segment, and the image is also The gray level is normalized, so the initial cluster center value can be set based on prior knowledge, so that classification errors can be avoided, and the segmentation effect and efficiency can be improved.
  • the cornea, iris, and lens regions can be roughly segmented. Roughly because the overall image information is used for segmentation, the boundary information at certain gray value critical points will not be particularly accurate;
  • Blob analysis According to the positional relationship and shape information of each area, screen out the non-anterior segment image. Blob analysis can obtain the position information (such as center, center of gravity, etc.) and shape information (eccentricity, circularity, compactness, etc.) of the target area Since the prior information of the anterior segment is known, the position information and shape information can be used to exclude non-anterior segment images to avoid further redundant processing;
  • a person of ordinary skill in the art can understand that all or part of the steps in the method of the foregoing embodiments can be implemented by a program instructing related hardware.
  • the program can be stored in a computer readable storage medium.
  • Media such as ROM/RAM, magnetic disk, optical disk, etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Probability & Statistics with Applications (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Eye Examination Apparatus (AREA)

Abstract

一种基于机器视觉的眼前节断层图像的图像特征提取方法，首先对由相机采集的眼前节断层图像进行灰度直方图统计，剔除不可能是眼前节的图像，接着根据直方图进行灰度归一化，减少环境光对成像质量的影响，然后通过K-mean聚类算法粗略分割，可分割出角膜、虹膜、晶状体区域，再进行blob分析，根据各区域的位置关系及形状信息，筛除非眼前节图像，再在各区域的粗略边界基础上进行固定方向的精细边界跟踪，从而获得角膜、虹膜、晶状体的精确轮廓，为后续求得眼前节临床参数提供了可靠的基础数据。

Description

基于机器视觉的眼前节断层图像的图像特征提取方法 技术领域
本发明涉及眼科医学图像处理方法，具体涉及一种基于机器视觉的眼前节断层图像的图像特征提取方法。
 背景技术
眼前节是眼球的一部分，具体包括：全部角膜、虹膜、睫状体、前房、后 房、晶状体悬韧带、房角、部分晶状体、周边玻璃体、视网膜及眼外肌附着点 部和结膜等。
近年来眼科医学影像技术发展迅速，使眼科医生对眼睛的观察更直接、更 清晰，确诊率也更高。计算机辅助诊断技术主要研究如何通过图像处理技术对 这些眼科医学影像信息进行有效的处理，辅助眼科医生的诊断甚至进行手术规 划，具有重大的社会效益和广泛的应用前景。
医学图像处理技术作为计算机辅助诊断的关键不断发展，各学科的交叉已 是必然的趋势。特别是随着眼科医疗的蓬勃发展，对眼科医学图像处理与分析 提出的要求也越来越高，所以进一步研究研究医学图像处理与分析具有十分重 要的意义。
因此，本领域的技术人员致力于开发一种眼前节断层图像特征提取方法。
 技术问题
本发明的目的是为了克服现有技术存在的缺点和不足，而提供一种基于机器视觉的眼前节断层图像的图像特征提取方法。
 
 技术解决方案
本发明所采取的技术方案如下：基于机器视觉的眼前节断层图像的图像特征提取方法，包括以下步骤：
（1）     输入采集到的眼前节断层图像；
（2）     对眼前节断层图像进行灰度直方图统计，剔除不可能是眼前节的图像；
（3）     根据直方图进行灰度归一化；
（4）     通过K-mean聚类算法粗略分割，分割出角膜、虹膜、晶状体区域；
（5）     进行blob分析，根据各区域的位置关系及形状信息，筛除非眼前节图像；
（6）  在各区域的粗略边界基础上进行固定方向的精细边界跟踪，获得角膜、虹膜、晶状体的精确轮廓。
 有益效果
本发明的有益效果如下：本发明首先对由相机采集的眼前节断层图像进行灰度直方图统计，剔除不可能是眼前节的图像，接着根据直方图进行灰度归一化，减少环境光对成像质量的影响，然后通过K-mean聚类算法粗略分割，可分割出角膜、虹膜、晶状体区域，再进行blob分析，根据各区域的位置关系及形状信息，筛除非眼前节图像，再在各区域的粗略边界基础上进行固定方向的精细边界跟踪，从而获得角膜、虹膜、晶状体的精确轮廓，为后续求得眼前节临床参数提供了可靠的基础数据。本发明可应用在裂隙灯成像类似原理设备所采集图像的处理上。
 附图说明
图1为本发明的流程示意图。
 本发明的最佳实施方式
在此处键入本发明的最佳实施方式描述段落。
 本发明的实施方式
如图1所示，一种基于机器视觉的眼前节断层图像的图像特征提取方法，包括以下步骤：
（1）     输入由裂隙灯成像类似原理设备所采集到的眼前节断层图像；
（2）     对眼前节断层图像进行灰度直方图统计，剔除不可能是眼前节的图像，灰度直方图统计：可获得图像灰度最大、最小、平均值、均方差等参数，可以通过灰度平均值过小、过大，或灰度方差过小等相关参数，粗略筛除非眼前节的图像，减少不必要的处理；
（3）     根据直方图进行灰度归一化，拍摄时的环境光的变化会引起图像上灰度值的变化，该变化可能会对后续的图像分割造成很大影响，因此需要通过灰度归一化来剔除环境光的影响，例如通过设置固定的均值、设置固定的最大最小值等方法；
（4）     通过K-mean聚类算法粗略分割，分割出角膜、虹膜、晶状体区域，K-mean是最简单的一种聚类方法，就是通过迭代寻找K个聚类中心，将所有的数据分配到距离最近的聚类中心，使得每个点与其相应的聚类中心距离的平方和最小。K-mean算法虽然简单，但通常能获得较好的聚类结果，其缺点是初始聚类中心的选择可能导致不同的聚类结果，但因为我们对眼前节图像有先验知识，并且图像也进行了灰度归一化，所以可以根据先验知识设定初始聚类中心值，从而可以避免分类错误，提高分割效果及效率。可以粗略分割出角膜、虹膜、晶状体区域，粗略是由于用的是整体图像信息进行分割，所以某些灰度值临界处的边界信息不会特别准确；
（5）     进行blob分析，根据各区域的位置关系及形状信息，筛除非眼前节图像，Blob分析可以获得目标区域的位置信息（例如中心、重心等）和形状信息（偏心度、圆心度、紧凑度、各阶矩等）等，由于已知眼前节的先验信息，所以可以通过位置信息和形状信息排除非眼前节图像，避免进一步冗余处理；
（6）     在各区域的粗略边界基础上进行固定方向的精细边界跟踪，获得角膜、虹膜、晶状体的精确轮廓，以K-mean算法获得的每一个粗略边界点为中心，在指定的方向或某个区域内进行精细边界搜寻，可通过搜寻灰度跳变最大值，或小范围内用K-mean算法进行边界定位；
（7）最后输出眼前节各组织结构边界信息。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分步骤是可以通过程序来指令相关的硬件来完成，所述的程序可以存储于一计算机可读取存储介质中，所述的存储介质，如ROM/RAM、磁盘、光盘等。
以上所揭露的仅为本发明较佳实施例而已，当然不能以此来限定本发明之权利范围，因此依本发明权利要求所作的等同变化，仍属本发明所涵盖的范围。

Claims (1)

  1. 基于机器视觉的眼前节断层图像的图像特征提取方法，其特征在于包括以下步骤：
    输入采集到的眼前节断层图像；
    对眼前节断层图像进行灰度直方图统计，剔除不可能是眼前节的图像；
    根据直方图进行灰度归一化；
    通过K-mean聚类算法粗略分割，分割出角膜、虹膜、晶状体区域；
    进行blob分析，根据各区域的位置关系及形状信息，筛除非眼前节图像；
    在各区域的粗略边界基础上进行固定方向的精细边界跟踪，获得角膜、虹膜、晶状体的精确轮廓。
PCT/CN2019/089776 2019-04-18 2019-06-03 基于机器视觉的眼前节断层图像的图像特征提取方法 WO2020211173A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910313547.XA CN110110761A (zh) 2019-04-18 2019-04-18 基于机器视觉的眼前节断层图像的图像特征提取方法
CN201910313547.X 2019-04-18

Publications (1)

Publication Number Publication Date
WO2020211173A1 true WO2020211173A1 (zh) 2020-10-22

Family

ID=67485840

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/089776 WO2020211173A1 (zh) 2019-04-18 2019-06-03 基于机器视觉的眼前节断层图像的图像特征提取方法

Country Status (2)

Country Link
CN (1) CN110110761A (zh)
WO (1) WO2020211173A1 (zh)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111652841A (zh) * 2020-04-24 2020-09-11 温州医科大学附属眼视光医院 一种基于图像处理的角膜损伤检测方法
CN111861977A (zh) * 2020-05-27 2020-10-30 温州医科大学附属眼视光医院 一种基于机器视觉的眼前节断层图像的特征提取方法
CN116309594B (zh) * 2023-05-23 2023-08-15 广东麦特维逊医学研究发展有限公司 眼前节oct图像处理方法
CN116342636B (zh) * 2023-05-23 2023-12-29 广东麦特维逊医学研究发展有限公司 眼前节oct图像轮廓拟合方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140055791A1 (en) * 2012-08-27 2014-02-27 Canon Kabushiki Kaisha Image processing apparatus and image processing method
CN104013384A (zh) * 2014-06-11 2014-09-03 温州眼视光发展有限公司 眼前节断层图像特征提取方法
CN107230211A (zh) * 2017-05-05 2017-10-03 上海联影医疗科技有限公司 一种图像分割方法及系统
CN108471995A (zh) * 2015-09-30 2018-08-31 上海联影医疗科技有限公司 确定医学图像中乳房区域的系统和方法
CN108470348A (zh) * 2018-02-13 2018-08-31 温州眼视光发展有限公司 裂隙灯眼前节断层图像特征提取方法

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101539991B (zh) * 2008-03-20 2011-07-06 中国科学院自动化研究所 用于虹膜识别的有效图像区域检测和分割方法
JP6304928B2 (ja) * 2012-03-30 2018-04-04 キヤノン株式会社 前眼部断層像解析方法及び前眼部断層像解析装置
CN105894521A (zh) * 2016-04-25 2016-08-24 中国电子科技集团公司第二十八研究所 基于高斯拟合的亚像素边缘检测方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140055791A1 (en) * 2012-08-27 2014-02-27 Canon Kabushiki Kaisha Image processing apparatus and image processing method
CN104013384A (zh) * 2014-06-11 2014-09-03 温州眼视光发展有限公司 眼前节断层图像特征提取方法
CN108471995A (zh) * 2015-09-30 2018-08-31 上海联影医疗科技有限公司 确定医学图像中乳房区域的系统和方法
CN107230211A (zh) * 2017-05-05 2017-10-03 上海联影医疗科技有限公司 一种图像分割方法及系统
CN108470348A (zh) * 2018-02-13 2018-08-31 温州眼视光发展有限公司 裂隙灯眼前节断层图像特征提取方法

Also Published As

Publication number Publication date
CN110110761A (zh) 2019-08-09

Similar Documents

Publication Publication Date Title
WO2020211173A1 (zh) 基于机器视觉的眼前节断层图像的图像特征提取方法
Fu et al. Age challenge: angle closure glaucoma evaluation in anterior segment optical coherence tomography
US8705826B2 (en) Automatic cup-to-disc ratio measurement system
Kumar et al. Deep transfer learning approaches to predict glaucoma, cataract, choroidal neovascularization, diabetic macular edema, drusen and healthy eyes: an experimental review
JP2021531098A (ja) Ａｉを利用した眼の状態判定のためのシステムおよび方法
Kumar et al. Detection of Glaucoma using image processing techniques: A review
Al-Bander et al. A novel choroid segmentation method for retinal diagnosis using deep learning
Praveena et al. Determination of cup to disc ratio using unsupervised machine learning techniques for glaucoma detection
CN114445666A (zh) 基于深度学习眼底图像左右眼及视野位置分类方法及系统
Aruchamy et al. Automated glaucoma screening in retinal fundus images
Priya et al. Detection and grading of diabetic retinopathy in retinal images using deep intelligent systems: a comprehensive review
WO2020029064A1 (zh) 光学相干断层图像处理方法
Pathan et al. Automated detection of pathological and non-pathological myopia using retinal features and dynamic ensemble of classifiers
Krishna et al. Retinal vessel segmentation techniques
Shashank et al. Con-Ker: a convolutional neural network based approach for keratoconus detection and classification
Khan et al. Diabetic retinopathy detection by image processing algorithms and machine learning technique
Yudono et al. Comparison of Cataract Classification System Based on Retinal Blood Vessels Objects and Retinal Optic Disc Using Backpropagation Neural Network
Kuppusamy An artificial intelligence formulation and the investigation of glaucoma in color fundus images by using BAT algorithm
Panda et al. Cataract Detection Using Deep Learning
Devi et al. A SURVEY ON IDENTIFICATION OF DIABETIC RETINOPATHY FOR MEDICAL DIAGNOSIS
Su Comparative Analysis of Accuracy and Loss during Cataract Detection using Deep Learning
US20230346216A1 (en) Methods and systems for quantifying retinal vascular patterns and treatment of disease
Prashanth et al. Early and efficient detection of glaucoma using image processing and deep learning
Selvakumar et al. An Efficient Investigation on Age-Related Macular Degeneration Using Deep Learning with Cloud-Based Teleophthalmology Architecture
Tahoun Automatic Discrimination of Color Retinal Images using the Bag of Words Approach

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19925503

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19925503

Country of ref document: EP

Kind code of ref document: A1