干眼检测图像处理技术研究与实现

发布时间：2018-05-29 05:07

本文选题：蒸发性干眼 + 睑板腺　；参考：《南京理工大学》2017年硕士论文

【摘要】：干眼症,目前成为眼科疾病中最普遍以及最易容易被忽视的疾病之一。干眼的患病原因有很多,其中最普遍的为蒸发性干眼,该类干眼是由睑板腺分泌油脂偏少,导致泪液蒸发过快。因此睑板腺分布情况与泪液的蒸发速率成为了干眼检查目前临床诊断的重要依据。根据目前干眼检测的普及方式,还是集中于侵入式的检查方式,这种检查手段不仅给检查者带来一定的眼部损伤,同时复杂的检查流程也被很多患者与医生所诟病。本文研究的主要工作是通过将传统的诊断方法转变为目前先进的图像处理技术,通过计算机图像分析给医生带来更加方便,快速与准确的诊断依据。同时也希望通过非侵入时的检查方法消除对检查者的眼部损伤,检查过程也更加简单高效。本文主要解决的两大问题,第一是如何通过图像处理技术解决睑板腺目前较难观察的问题;同时还有如何利用图像的技术判别泪液量。围绕第一个问题,本文详细阐述了基于图像增强技术提高睑板腺的可视程度。在特定拍摄光源环境的条件下,我们对采集的图像进行适当的细节增强处理。在文章中我们结合传统的图像增强方法,提出更加适合睑板腺图像增强方法,使得睑板腺的细节位置更加突出明显,方便医生观察和诊断。第二个问题关于泪液分泌的情况。泪液通过瞬目在眼角膜处形成泪膜。本文通过分析泪膜的破裂情况来侧面反映泪液的分泌和蒸发。在文中,我们借用角膜地形图中角膜曲率变化情况映射于泪膜破裂的检测中。本文的方法借用角膜地形图Placido环光源与圆环成像原理,通过识别圆环的破裂与变化情况来标定泪膜破裂的位置。根据这两大研究问题,结合图像处理技术,本文提出了更加适合干眼检测的新的图像处理算法。包括图像增强的算法、图像配准的算法以及边缘轮廓追踪的算法。基于上述的理论分析,对这些算法进行了实验模拟。实验结果表明,本文提出的适合干眼检测的算法能够达到比较满意的效果。与此同时为了适应不同型号的后期产品设备,本文根据传统的Placido平面光源和相关几何光学的原理设计出新的不同类型的Placido环形光源结构,使检测设备更加贴合患者眼部,成像范围扩大。最终,文章还根据检测算法模块和病人信息管理设计了上位机设备的应用软件初期模型,建立了病人的数据模型,迎合了目前云端大数据的潮流。这也为患者更为全面医疗备案和更为准确的医学诊断提供了保障。
[Abstract]:Xerophthalmia is one of the most common and easily neglected ophthalmic diseases. There are many causes of dry eye, the most common of which is evaporative dry eye, which is secreted by tarsal gland, which results in excessive evaporation of tears. Therefore, the distribution of tarsal glands and the evaporation rate of tears have become the important basis for clinical diagnosis of dry eye examination. According to the present popularization of dry eye examination, it is still focused on the invasive examination method, which not only brings some ocular injuries to examiners, but also the complex examination process is criticized by many patients and doctors. The main work of this paper is to change the traditional diagnosis method into the current advanced image processing technology, and to bring more convenient, fast and accurate diagnostic basis to doctors through computer image analysis. At the same time, we hope to eliminate the eye injury by non-invasive examination method, and the examination process is simpler and more efficient. The first is how to solve the problem of meibomian gland which is difficult to observe by image processing technology, and how to use the image technology to judge the amount of tear. Focusing on the first problem, this paper describes in detail how to improve the visibility of tarsal glands based on image enhancement. Under the condition of special light source environment, we processed the collected image with appropriate detail enhancement. In this paper, combining with the traditional image enhancement method, we propose a more suitable method for tarsal gland image enhancement, which makes the details of the tarsal gland more prominent and more convenient for doctors to observe and diagnose. The second question is about the secretion of tears. Tears form in the cornea through the blink of the eye. The tear film rupture was analyzed to reflect the secretion and evaporation of tear. In this paper, we use the corneal topography to map the change of corneal curvature to the detection of tear film rupture. In this paper, the location of tear film rupture is calibrated by identifying the rupture and variation of the circular ring by using the Placido ring light source and the imaging principle of the ring of corneal topography. According to these two research problems and combined with image processing technology, this paper proposes a new image processing algorithm which is more suitable for dry eye detection. It includes image enhancement algorithm, image registration algorithm and edge contour tracking algorithm. Based on the above theoretical analysis, these algorithms are simulated experimentally. The experimental results show that the proposed algorithm can achieve satisfactory results. At the same time, in order to adapt to different types of late product equipment, according to the traditional Placido plane light source and the principle of related geometry optics, this paper designs a new type of Placido ring light source structure, which makes the detection equipment more close to the eyes of the patient. The imaging range is enlarged. Finally, according to the detection algorithm module and patient information management, the paper designs the initial application software model of the upper computer equipment, and establishes the patient data model, which caters to the current trend of cloud big data. This also provides a guarantee for more comprehensive medical records and more accurate medical diagnosis.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】