节点文献
频域OCT对青光眼视网膜分层厚度检测的研究
Spectral Domain Optical Coherence Tomography Measurements of Retinal Layers Thicknesses in Glaucoma
【作者】 樊宁;
【导师】 黄丽娜;
【作者基本信息】 暨南大学 , 眼科学, 2011, 博士
【摘要】 频域相干光断层扫描术(spectral domain optical coherence tomography, SDOCT)被认为是最有发展前途的OCT技术。相对于时域(time domain, TD) OCT,它极大地提高了扫描速度和轴向分辨率,视网膜的活体断层成像更加清晰、直观。在临床应用中,我们还有很多疑问。频域OCT与时域OCT的测量的视网膜神经纤维层(retinal nerve fiber layer, RNFL)厚度,两者有无差异?对早期原发性开角型青光眼(primary open angle glaucoma, POAG)的诊断能力如何?频域OCT对RNFL局部缺损的检测与眼底彩照的结果是否一致?是否比时域OCT更灵敏、准确?黄斑区视网膜神经节细胞复合体(ganglion cell complex, GCC)层在青光眼人群有何变化?与视功能参数的改变是否一致?青光眼是否累及光感受器细胞?关于这些问题,目前国外学者研究不多,国内基本未见报道。本研究采用频域OCT检测了视网膜的RNFL、GCC和光感受器细胞层的厚度,分析了青光眼对视网膜不同层次的结构损害特征,希望解答上述问题,提高我们对青光眼的认识能力和诊断水平。第一部分频域OCT对视网膜神经纤维层厚度的检测目的1.比较频域与时域OCT测量早期、视野前期青光眼和非青光眼人群RNFL厚度的差异和联系,及其对青光眼的早期诊断效能。2.评价频域、时域OCT与眼底彩照检测青光眼RNFL局部缺损的相关性和一致性。方法1.分别用Cirrus HD-OCT和Stratus OCT测量非青光眼人群62例(62眼)、早期POAG患者47例(47眼)、视野前期POAG患者30例(30眼)的RNFL厚度,比较两种OCT测量的三组人群4个象限及全周平均RNFL厚度的差异、相关性和一致性;比较各参数对早期、视野前期青光眼的诊断效能;比较两种OCT根据系统自带的正常人数据库,对青光眼人群检测的阳性率。2.选择经眼底彩照确认有RNFL局部缺损的POAG患者55例(55眼)和非青光眼人群41例(41眼)纳入研究,分析眼底彩照、Cirrus HD-OCT、Stratus OCT分别检测的RNFL局部缺损的位置和宽度,进行三者检测结果的一致性和相关性研究,评价OCT对RNFL局部缺损的诊断价值。结果1. Cirrus HD-OCT测量的三组人群4个象限和全周平均RNFL厚度均较Stratus OCT薄,差异有统计学意义;在非青光眼和视野前期、早期青光眼人群,两种OCT测量的全周平均RNFL厚度的差值分别是:(11.6±10.6)μm、(6.9±17.5)μm和(6.0±13.1)μm。2.两种OCT测量的4个象限和全周平均RNFL厚度的相关系数为0.676~0.935(P均为0.000),其中下方、上方象限和全周平均测量值之间相关系数均达到0.9以上。3. Cirrus HD-OCT和Stratus OCT测量的视野前期青光眼全周平均RNFL厚度的AROC为0.951和0.881,差异有统计学意义(P=0.006);而其他4个象限RNFL厚度值的AROC的差异没有统计学意义(P>0.05)。两种OCT对早期青光眼测量的全周平均及各象限RNFL厚度的AROC均没有显著差异(P>0.05)。两种OCT测量的全周平均、下方及上方RNFL厚度的AROC较大,在视野前期青光眼组均>0.87,在早期青光眼组均>0.95。4.根据系统自带的正常数据库,Cirrus HD-OCT、Stratus OCT检测视野前期青光眼人群的阳性率分别是83.4%、43.3%,而对早期青光眼,两种OCT的阳性率分别是97.9%、95.8%。5.55例(55眼)POAG患者共有RNFL局部缺损75处,分别位于颞上和颞下象限。以RNFL厚度落在系统正常数据库的红色区域为标准,Cirrus HD-OCT和Stratus OCT检测RNFL局部损害的敏感度分别是88.0%和69.3%,特异度分别是92.7%和97.6%。6. Cirrus HD-OCT和Stratus OCT检测RNFL局部缺损的位置均与眼底彩照的检测结果高度正相关(r=0.993、0.992,P均=0.000);两种OCT检测的RNFL局部缺损宽度与眼底彩照检测的差异均没有统计学意义(Cirrus HD-OCT:P=0.114, Stratus OCT:P==0.074),并且OCT与眼底彩照的检测值之间有一定正相关(Cirrus HD-OCT:r=0.420, P=0.019; Stratus OCT:r=0.432, P=0.002). Cirrus HD-OCT对RNFL局部缺损宽度的检测值大于Stratus OCT (P=0.002)。结论1.对于视野前期、早期青光眼和非青光眼人群,Cirrus HD-OCT检测的各象限和全周平均RNFL厚度值普遍比Stratus OCT显著偏小,但是相关性较好。2.两种OCT测量的全周平均、下方和上方RNFL厚度值均能很好的区分视野前期、早期青光眼和非青光眼人群,Cirrus HD-OCT对视野前期青光眼的诊断能力高于Stratus OCT.3.两种OCT均具有较好的诊断RNFL局部缺损的价值,与眼底彩照检测值一致性较好。第二部分频域OCT对视网膜神经节细胞复合体厚度的检测目的1.探讨频域OCT检测的黄斑区视网膜神经节细胞复合体(ganglion cell complex, GCC)厚度在青光眼人群的分布特征。2.分析GCC厚度与视功能检测参数在青光眼与非青光眼人群的变化关系,同时结合RNFL厚度进行评价,探讨青光眼对视网膜结构与功能损害的一致性。方法1.采用频域OCT (RTVue-100)和时域OCT (Stratus OCT)分别检测非青光眼人群41例(41眼)和POAG患者101例(101眼)的黄斑区GCC厚度和视盘全周平均RNFL厚度,比较GCC的5个厚度参数值在早期和视野前期青光眼与非青光眼人群的差异,以及在早期、进展期和晚期青光眼人群的差异;计算并比较GCC厚度与RNFL厚度的受试者工作特征曲线下面积(area under the receive operating characteristic curve, AROC).2.对非青光眼人群41例(41只眼)和POAG患者95例(95只眼)分别进行RTVue-100OCT检测黄斑区整体平均GCC厚度、Cirrus HD-OCT检测视盘全周平均RNFL厚度、Humphrey视野计检测平均偏差(mean deviation, MD)、闪光视网膜电图检测明视负波反应(photopic negative response, PhNR)的振幅。通过曲线拟合、建立回归模型分析以上4个参数在青光眼病程中的变化关系。结果1.GCC的5个厚度参数(整体平均厚度GCC-Avg、上方平均厚度GCC-Sup、下方平均厚度GCC-Inf、局部丢失容积GCC-FLV、整体丢失容积GCC-GLV)在视野前期、早期青光眼与非青光眼人群比较,差异均有统计学意义(P<0.001)。2.在青光眼的视野前期、早期、进展期与晚期,整体平均厚度GCC-Avg、上方平均厚度GCC-Sup和下方平均厚度GCC-Inf的测量值逐渐减少;而局部丢失容积GCC-FLV、整体丢失容积GCC-GLV的测量值逐渐增大,差异有统计学意义(P<0.001)。3. 5个GCC厚度参数诊断青光眼的AROC分别是0.965,0.950,0.949,0.967,0.972。而全周平均RNFL厚度的AROC为0.978,与GCC各参数比较,其差异没有统计学意义(P>0.05)。4.黄斑区整体平均GCC厚度与MD值在青光眼病程中呈曲线变化关系,回归模型的决定系数为0.595;与PhNR振幅呈直线变化关系,回归模型决定系数为0.437。视盘全周平均RNFL厚度与MD、PhNR振幅也分别呈曲线、直线变化关系,决定系数分别是0.606,0.454。结论1.频域OCT测量的黄斑区GCC厚度,随青光眼病情进展而逐渐变薄,具有很好的区分青光眼与非青光眼人群的能力。2.GCC厚度和RNFL厚度均与MD呈曲线变化关系,与PhNR振幅呈直线变化关系,GCC厚度与功能参数的回归模型决定系数略低于RNFL厚度。第三部分频域OCT对视网膜光感受器细胞层厚度的检测目的1.用频域OCT辅助手工测量正常人视网膜光感受器细胞各层的厚度,评价检测方法的可行性和可重复性。2.以该方法测量青光眼和非青光眼人群的视网膜光感受器细胞层的厚度,探讨光感受器细胞层在青光眼病程中的变化。方法1.用频域OCT对正常人40例40眼的黄斑区扫描获得视网膜断层图像,同一位操作者采用Sigma图像分析软件盲法测量黄斑中心凹和旁中心凹处(中心凹外1.5 mm)视网膜光感受器细胞核、内节和外节层的厚度。2.检测方法的重复性检验:随机选择30张正常眼的SDOCT图像,由同一位检测者在3次非连续时间内、盲法测量其黄斑中心凹光感受器细胞各层的厚度。计算组内相关系数(intraclass correlation coefficient,ICC)、组内变异系数(within-subject coefficient of variation, CVw)和可重复性指数(repeatability coefficient)3.以该方法测量POAG患者(青光眼组)48例48眼和非青光眼人群(对照组)38例38眼的黄斑中心凹和旁中心凹处光感受器细胞各层厚度,比较其在青光眼与非青光眼人群的差异,分析光感受器细胞层厚度与视盘旁视网膜神经纤维层厚度间的关系。结果1.频域OCT测量正常人黄斑中心凹视锥细胞核层、内节层、外节层的平均厚度分别是:(96.1±10.5)μm、(24.2+2.1)μm、(32.1±3.2)μm;旁黄斑中心凹光感受器(以视杆细胞为主)核层、内节层、外节层的平均厚度分别是:(69.2±13.1)μm、(22.0+2.4)μm、(23.9±4.1)gm。2.测量黄斑中心凹视锥细胞核层、内节层、外节层的重复性:组内相关系数分别为0.964、0.816、0.824,组内变异系数分别为3.11、4.98、4.91,可重复性指数分别为8.475、8.137、8.132。3.早期青光眼组的黄斑中心凹处视锥细胞层、细胞核层厚度分别是(165.9+16.7)μm、(105.7±13.9)μm、较对照组明显增厚(P=0.026,0.020),内节和外节层厚度为(60.4±5.6)μm),与对照组比较差异没有统计学意义(P=1.000)。而中晚期青光眼组的黄斑中心凹处视锥细胞层、细胞核层、内节和外节层厚度分别是(160.0+14.8)μm、(101.8±12.7)μm、(58.5±5.3)μm,与对照组比较差异没有统计学意义(P=0.751,0.350,1.000)4.青光眼组的黄斑中心凹处视锥细胞核层厚度和视盘旁RNFL厚度之间有二元线性关系:Y=-0.019X2+2.73X+10.34,R2=0.211,P=0.005。5.青光眼组和对照组在中心凹外3 mm处光感受器细胞核层、内节层、外节层的厚度与对照组比较,差异均没有统计学意义(P=0.410,0.445,0.198)。结论1.频域OCT联合手工测量的方法对视网膜光感受器细胞各层的检测,具有较高的重复性,是研究活体光感受器细胞结构较可靠的方法。2.早期青光眼的黄斑中心凹视锥细胞层及细胞核层显著增厚,并随病程进展呈现出动态、曲线性变化。
【Abstract】 The spectral domain OCT (SD OCT) is deemed to be the most promising OCT technology. Compared with time domain (TD) OCT, it shows much faster scan speed and higher axial revolution, and improves visualization of the intra-retinal structures in vivo more clear.In clinical use of SDOCT, we have many questions:Is there any difference between the retinal nerve fiber layer (RNFL) thicknesses measured using SDOCT and TDOCT? And whether the diagnostic power of SDOCT is better? Whether the topographic profiles of localized RNFL defects determined by SD-OCT and TD-OCT agree with that determined by fundus photography? And is the SDOCT more sensitive and accurate? How the macular ganglion cell complex (GCC) thickness changes in glaucoma patients? Whether the GCC thickness changes correlate with the changes of visual functional parameters? Is photoreceptor involved in glaucoma? About the above problems, we found few reports conducted by some overseas researchers, but they are far from solved, and no report in Chinese.In this study, we use SDOCT to measure the RNFL, GCC and photoreceptor layer thickness, and analyze the intra-retinal structural damage in glaucoma. We hope to give answers to these clinical questions, and help us understanding the pathophysiology of glaucoma and assisting clinicians in glaucoma management.Part I Retinal nerve fiber layer thickness measurement by SD-OCTPurpose1. To compare the RNFL thickness measurements in mild glaucoma, preperimetric glaucoma patients and non-glaucoma subjects between TD-OCT and SD-OCT, to assess the diagnostic values of RNFL measurements by the two types OCT.2. To evaluate the correlation and agreement between topographic profiles of localized RNFL defects determined by SD-OCT and TD-OCT with fundus photography. Methods1. Cirrus HD-OCT and Stratus OCT were used to measure the RNFL thickness of 62 eyes of 62 non-glaucoma subjects,47 eyes of 47 mild glaucoma patients, and 30 eyes of 30 preperimetric glaucoma patients. The measurements of RNFL thickness parameters, their correlation, and diagnostic value between the two types OCT was compared and evaluated. The positive rates in glaucoma patients determined by the standard normal databases of the OCT systems were calculated.2. Forty-one eyes of 41 non-glaucoma subjects and 55 eyes of 55 glaucomatous patients with localized, wedge-shaped RNFL defects identified by two glaucoma specialists in fundus photographs were enrolled in the study. The angular location and width of RNFL defects determined on the images of fundus photography, Cirrus HD-OCT and Stratus OCT were analyzed respectively using Pearson’s correlation coefficient and linear regression analysis, to assess the diagnostic values of RNFL defects determined by two types OCT.Results1. The measuring values of the four quadrants and global average RNFL thickness parameters on the three groups by Cirrus HD-OCT were thinner than those by Stratus OCT with significant difference. The difference of the global average RNFL thickness between the two types OCT were 11.6±10.6μm,6.9±17.5μm and 6.0±13.1μm respectively in the non-glaucoma, preperimetric glaucoma and mild glaucoma groups. The correlation coefficients (r) of the RNFL thickness parameters between the two types OCT were 0.676-0.935 (P=0.000). Inferior and superior quadrant, global average RNFL thickness had the largest r value more than 0.9.2. The area under the receive operating characteristic curve (AROC) of global average RNFL measured by Cirrus HD-OCT and Stratus OCT on preperimetric glaucoma patients were 0.951 and 0.881 (P=0.006), and the AROC of the four quadrants RNFL thickness between the two types OCT without significant difference (P>0.05). The differences of the two OCTs’AROC of all the RNFL thickness parameters on mild glaucoma patients were all no significant (P>0.05). The AROC of the global average, inferior and superior RNFL thickness were larger than 0.87 on preperimetric glaucoma group,0.95 on mild glaucoma group.3. Based on the standard normal database of Cirrus HD-OCT and Stratus OCT systems, the positive rates in preperimetric glaucoma patients were 83.4% and 43.3%, in mild glaucoma patients were 97.9% and 95.8%.4. Seventy-five RNFL defects were identified in 55 glaucomatous eyes by two glaucoma specialists unanimously with the defect position at superior-temporal and inferior-temporal quadrants. If the RNFL thickness in the red color band of the normal database defined as the defect borderline, the sensitivity of Cirrus HD-OCT and Stratus OCT to determining RNFL defects were 88.0% and 69.3% respectively and their specificity were 92.7% and 97.6% respectively.5. The angular locations of RNFL defects by Cirrus HD-OCT and Stratus OCT were highly correlated with those by fundus photography (r=0.993,0.992, P<0.001). No significant differences were found in the defect width of RNFL between Cirrus HD-OCT or Stratus OCT and fundus photography (Cirrus HD-OCT:P=0.114; Stratus OCT:P=0.074), and the angular widths of RNFL defects by Cirrus HD-OCT and Stratus OCT were moderately correlated with those by fundus photography(r=0.420,0.432, P=0.019,0.002). The angular widths of RNFL defects by Cirrus HD-OCT was larger than that by Stratus OCT (P=0.002).Conclusions1. RNFL thickness measurements generally were thinner by Cirrus HD-OCT than by Stratus OCT in mild glaucoma, preperimetric glaucoma patients and non-glaucoma subjects, but the measurements of the two types OCT correlated well.2. The global average, superior and inferior quadrant RNFL thickness measured by both of the two types OCT, could distinguish between the mild glaucoma or preperimetric glaucoma patients and non-glaucoma subjects. The diagnostic power of Cirrus HD-OCT was higher than that of Stratus OCT.3. Both of the two types OCT could localize RNFL defects with high sensitivity and specificity. The measure value of Cirrus HD-OCT and Stratus OCT for RNFL defects showed a good diagnostic agreement with fundus photography. Part□Ganglion cell complex thickness measurement by SD-OCTPurpose1. To explore the macular GCC thickness feature in POAG patients measured with SD-OCT.2. To analyze the association between GCC, RNFL thickness and the visual function parameters, and evaluate the relationship between structural and functional damage of retina in glaucoma.Methods1. The macular GCC thickness and peripapillary RNFL thickness of 41 eyes of 41 non-glaucoma subjects and 101 eyes of 101 POAG patients were measured using RTVue-100 and Stratus OCT respectively. The measurements of the 5 GCC thickness parameters (GCC-Avg, GCC-Sup, GCC-Inf, GCC-GLV and GCC-FLV) were compared between mild glaucoma or preperimetric glaucoma patients and non-glaucoma subjects, and among the mild, advanced and end-stage glaucoma patients. The area under the receive operating characteristic curve (AROC) of the GCC thickness and RNFL thickness were calculated and compared.2. Ninety-five eyes of 95 POAG patients and 41 eyes of 41 non-glaucoma subjects were enrolled in the study. Macular GCC-Avg thickness and peripapillary average RNFL thickness were measured using RTVue-100 OCT and Cirrus HD-OCT respectively, mean deviation (MD) of visual field was examined using Humphrey VF analyzer, photopic electroretinograms were elicited by white stimuli on a white background and the amplitude of photopic negative response (PhNR) were measured. The associations between the GCC or RNFL thickness and MD or amplitude of PhNR were evaluated with linear and curvilinear regression models.Results1. There were significant differences of the 5 GCC thickness measurements between mild glaucoma or preperimetric glaucoma patients and non-glaucoma subjects (P<0.001). 2. The measurements of GCC-Avg, GCC-Sup and GCC-Inf thickness were decreased, while the measurements of GCC-FLV and GCC-GLV were increased, from mild and advanced to end-stage glaucoma with significant difference (P<0.001).3. The AROC of the 5 GCC thickness parameters (GCC-Avg, GCC-Sup, GCC-Inf, GCC-GLV and GCC-FLV) were 0.965,0.950,0.949,0.967 and 0.972 respectively. The AROC of global average RNFL thickness parameter was 0.978. There were no significant differences of the AROC between the 5 GCC thickness and RNFL thickness paramenters.4. The curvilinear regression model better described the relationship between GCC thickness and MD with coefficient of determination (R2=0.595), and the linear regression model better fit the relationship between GCC thickness and amplitude of PhNR with coefficient of determination (R2=0.437). RNFL thickness showed the similar regression models with MD and amplitude of PhNR as GCC thickness, but the coefficients of determination were higher between the RNFL thickness and MD or amplitude of PhNR (R2=0.606,0.454).Conclusions1. The macular GCC thickness measured using SDOCT decreased with the severity of glaucoma, it could well differentiate between non-glaucoma subjects and glaucomatous patients.2. Both of the GCC thickness and RNFL thickness showed a curvilinear relationship with MD and a linear relationship with amplitude of PhNR, the coefficients of determination of GCC thickness were lower than RNFL thickness.Part□Photoreceptor layer thickness measurement by SD-OCTPurpose1. To explore the feasibility and repeatability of manual measure photoreceptor layer thickness using SDOCT.2. To measure and compare photoreceptor layer thickness between glaucoma and non-glaucoma subjects using SDOCT, and explore the change of photoreceptor in glaucoma. Methods1. The macular areas of 40 eyes from 40 normal subjects were imaged by SDOCT. The outer nuclear layer (ONL) and inner and outer segments (IS/OS) layer thicknesses in fovea and parafovea (1.5 mm from the fovea) were measured by a single masked observer using an image analysis software (SigmaScan Pro version 5.0).2. Repeatability test:the measurements were repeated 3 times in a random sample of 30 normal macular images to determine the intraclass correlation coefficient (ICC), within-subject coefficient of variation (CVw) and repeatability coefficient.3. Forty-eight eyes of 48 POAG patients and 38 eyes of 38 non-glaucoma subjects were measured the photoreceptor layer thickness in fovea and parafovea using SDOCT. The measurements between glaucoma and non-glaucoma groups were compared. The association between photoreceptor layer thickness and RNFL thickness was evaluated.Results1. The ONL、IS and OS layer thickness in normal subjects were 96.1±10.5μm,24.2±2.1μm and 32.1±3.2μm respectively in fovea, and 69.2±13.1μm,22.0±2.4μm and 23.9±4.1μm respectively in parafovea.2. Repeatability test results:the ICC of ONL, IS and OS layer thickness measurement was 0.964,0.816 and 0.824 respectively, the CVw of the 3 layer thickness measurement was 3.11%,4.98% and 4.91%, and the repeatability coefficient of the 3 layer thickness measurement was 8.475,8.137 and 8.132, respectively.3. The total photoreceptor layer and the ONL thickness in the fovea were 165.9±16.7μm and 105.7±13.9μm respectively in the mild glaucoma group, which were higher than that of the non-glaucoma group (P=0.026,0.020). While the IS+OS layer thickness in the fovea were 60.4±5.6μm, which was no significant difference compared with the non-glaucoma group (P= 1.000). The total photoreceptor layer, the ONL, and the IS+OS layer thickness in the parafovea were 160.0±14.8μm,101.8±12.7μm and 58.5±5.3μm respectively in the advanced glaucoma group, which was no significant difference compared with the non-glaucoma group (P=0.751,0.350,1.000).4. The relationship between ONL in fovea and RNFL thickness in glaucoma group was best described with a second order polynomial regression model association (Y=-0.019X2+ 2.73X+10.34,R2=0.211, P=0.005).5. There were no significant differences between the glaucoma and non-glaucoma group of the ONL, IS and OS layer thickness in parafovea (P=0.410,0.445,0.198).Conclusions1. The manual measure photoreceptor layer thickness using SDOCT was feasible and repeatable for in vivo study.2. The foveal ONL thickness was increased in mild glaucomatous eyes. The curvilinear alteration of foveal ONL thickness was associated with the severity of the disease.
【Key words】 glaucoma; optical coherence tomography/spectral domain; retinal nerve fiber layer; retinal ganglion cell; macula; photoreceptor;