【作者】 刘瑞雪；

【导师】 宣丽；

【作者基本信息】 中国科学院研究生院(长春光学精密机械与物理研究所) ， 光学工程， 2014， 博士

【摘要】 通过视网膜微血管形态检查可以早期诊断内分泌疾病。但是，由于人眼像差的存在，使临床所用检眼镜的分辨率大幅下降，不能直接看到10μm以下的微细血管。自适应光学技术能够实时补偿人眼像差，得到接近衍射极限的视网膜视觉细胞图像，但对视网膜微细血管成像时还存在一些问题。本论文针对准确定位微细血管困难、图像对比度难以达到临床要求等问题，结合人眼的光学特性，进行了一系列光学系统的创新设计。采用视标盯视方法进行眼底方位定位。根据人眼焦深、调节误差、视觉锐度和对颜色的敏感度，设计视标的位置、形状、照明波长等：确定眼前1D处的视标满足准确性、普适性的要求；设计马耳他十字视标，调节支臂可满足不同视力的人眼盯视；采用人眼最敏感的绿光照射视标，有利于提高人眼盯视能力。根据视网膜结构，定位方法分三步：1）选择小凹中心为定位的基准，2）横向定位之后，按照系统放大倍率，将成像相机进行轴向位移，聚焦到位于视网膜内部几层的血管，3）找到较粗血管后，可沿血管走势有意识控制横向定位，直至寻找到10μm以下微细血管。设计了具有微细血管搜寻定位的自适应光学成像系统。根据视网膜血管的吸收波长，确定了高对比度视网膜血管的成像方法：1）先用785nm红外光搜寻，以避免绿光长时曝光对人眼的强烈刺激，2）再使用561nm绿光数毫秒曝光成像。通过理论计算得到具体的有利于提高对比度的途径。利用血液与周围组织结构偏振性的差异，设计液晶偏振自适应光学系统，利用微细血管的消偏光成像，对比度达到0.25，较原来提高约1.2倍。针对实际光路中由于光学器件的加工和光学系统装调中的误差，极限分辨率远远达不到理想设计值的问题，设计简易化自适应光学系统，大大减少所使用的光学元器件的个数。在保证系统校正功能的基础上，提高了系统的能量利用率和调制传递函数。利用图像配准与叠加消除噪声以提高对比度，使得获得的图像具有更高的临床使用价值。更多还原

【Abstract】 Fundus examination is a necessary step to diagnos many diseases, for that themorphology of retinal vessels can early represent many endocrine diseases.However, due to the existence of ocular aberrations, the resolution of the traditionalophthalmoscope is not so high to observe changes of vascular subtle timely.Adaptive optics is a novel technology to compensate aberrations of human eye forretinal imaging, which can reach close to the diffraction limit. This research grouphas been employed a liquid crystal adaptive optics system successfully for highresolution imaging of retinal photoreceptor cells. However, there are still someproblems on human vascular imaging, including accurate positioning of microvascular difficulties, not specifically for vascular imaging adaptive optics system, theimage contrast is difficult to achieve the clinical application. In this paper, combinedwith the optical properties of the eye, several measures would be taken on theinnovation design a series of optical system.For non-paralysis eye, a target was employed for eyes staring. According tohuman focal depth, the eye accommodation error, visual acuity and color sensitivity,the visual target location, shape, illumination wavelength were identified anddesigned. The distance of1D target from the human pupil met the accuracyrequirements of the system. The design of Malta cross target shape, regulating arm can meet different visual acuity eyes stare. The most sensitive wavelength is greenlighting. Even if a faint green light was illuminated to the target, the human eye canalso clearly distinguish details, which is beneficial to improve the ability to stare.Based on retinal structures, the location method was determined to divide into threesteps:1） selecting foveal most center positioning reference,2） lateral positioning,in accordance with the system magnification, the imaging camera axial displacement,focusing in the inner layers of the retina blood vessel; this is a process of axial.3）Find the vessel position, along the vascular trend conscious control of lateralpositioning, in order to find the location of the lesion, splicing and processing alsohas the benefit of late vascular.An adaptive optical imaging system with micro vascular searching and locatingwas designed. According to the distribution characteristics of vessels, a stare systemcan realize large range for searching vessels. According to the optical characteristicsof retinal blood vessels, the illumination system contains two sources, includingdetection and imaging light source. In order to achieve good imaging results, thelinkage of light source and camera should be realized. With the analyzed statistics ofthe human eye axis data and calculated axial magnification, the movement of sourceand camera can be confirmed accurately.Specific ways to improve images contrast had been theoretical calculated anddiscussed. Using the difference polarization characteristics between blood and othertissues, a polarization liquid crystal adaptive optics system had been designed. Withthe help of polarized light imaging of micro blood vessels, the imaging contrast canbe improved to0.25, which was increased by about1.2times and basically reachedthe requirements for clinical use. In view of the actual optical path due to theprocessing and optical devices installed in the resolution limit error adjustment, asfar from the ideal design value, design of simple adaptive optical system, greatlyreduce the number of optical components used in the. Based on obtained retinalvascular images, a variety of image processing methods were used for improvingcontrast. Image registration was employed for noise elimination to improve contrast. In accordance with the blood flow, the vascular morphology was reconstructed,which made the obtained images with higher clinical value.更多还原