两种回归方法推算扫描视觉诱发电位客观推断视力的比较

Comparison of two methods of extrapolating sweep pattern visual evoked potential acuity

背景扫描视觉诱发电位（SVEP）是一种客观视力的检测方法，虽然SVEP客观推断视力与主观视力之间存在着明显的相关性，但是两者之间并不完全吻合。研究显示，改进SVEP客观推断视力的推算方法可以提高推断视力的准确性。目的研究SVEP客观推断视力两种推算方法，即振幅-空间频率（A—SP）回归方法与振幅一视角对数（A-logVA）回归方法在客观视力评估中的准确性，探讨影响SVEP客观推断视力准确性的因素。方法对64例具有不同视力的113眼进行SVEP检测、主观小数视力和LogMAR视力检测，分别计算A—SP回归方法推算出的小数视力和LogMAR视力及A．10gVA回归方法推算出的小数视力和LogMAR视力，并对检测结果行相关性分析。使用重庆国特公司GT-2000n型视觉电生理仪检测SVEP，选用0．99～12．89cpd的10个空间频率水平正弦光栅连续刺激视网膜，经离散傅里叶分析获得A—SP回归曲线和A—logVA回归曲线，并将相关直线外推至0振幅处，获得SVEP客观推断视力。结果在小数视力评价系统中，A—SP回归方法和A-logVA回归方法推算出的回归线相关指数分别为0．663和0．705，主观小数视力与由A-SP回归方法推算所得的小数视力间呈正相关（r=0．620，P〈0．01），主观小数视力与由A—logVA回归方法推算所得的小数视力间呈正相关（r=0．540，P〈0．01）。SP小数视力与VA小数视力间差异有统计学意义（Z=-8．688，P〈0．01）；在LogMAR视力评价系统中，A-SP回归方法和A—logVA回归方法推算出的回归线相关指数分别为0．733和0．715，主观LogMAR视力与由A-SP回归方法推算所得的LogMAR视力间呈正相关（r=0．710，P〈0．01），主观LogMAR视力与由A-logVA回归方法推算所得的LogMAR视力间呈正相关（r=0．700，P〈0．01）。两种推算方法获得的LogMAR视力间差异有统计学意义（Z=-8．748，P〈0．01）。不同性别、眼别、病种以及年龄段中VALogMAR视力测试值的偏差差异均无统计学意义（χ2=2．171，P=0．338；χ2=0．976，P=0．614；χ2=6．032，P=0．420；χ2=14．720，P=0．257）。结论SVEP是检测客观视力较为有效的方法。但SVEP变异较大，就目前检测技术的稳定性与准确性而言，A—logVA回归方法推算出的客观视力较A—SP回归方法准确。

Background Sweep pattern visual evoked potential （SVEP） is an objective method of visual test. There is a clear correlation between SVEP acuity and subjective vision,but they are not identical. Recent studies showed that new regression method can improve the accuracy of SVEP acuity. Objective This trial was to investigate and compare the outcome between amplitude-spatial frequency （A-SP） regression method and amplitude- logVA （A-logVA） regression method in extrapolating the SVEP acuity. Methods SVEP was recorded in 113 eyes of 64 subjects using GT-2000 （ Guo Te ,China） with the gratings of 10 different spatial frequency from 0. 99 to 12.89 cpd as stimulus. The 113 eyes included cataract, glaucoma, corneal disease, optical neuropathy, retinal disease, ocular trauma,refractive error and normal eyes. The correlation were analyzed of SVEP acuity, decimal visual acuity and LogMAR visual acuity. The response were averaged and DFT on the monitor display. SVEP acuity was calculated by extrapolating 0 response amplitude. Results The correlation indices of decimal visual acuity curves obtained by the A-logVA function was 0. 663, and that obtained by the A-SP function was 0. 705. The positive correlation was seen between subjective decimal visual acuity and A-logVA decimal visual acuity （ r = 0. 540, P 〈 0.01 ） and between subjective decimal acuity and decimal acuity calculated by the A-SP regression method （r = 0. 620,P〈0.01 ）. SVEP decimal acuity calculated by the A-SP function regression method was significantly different from the that calculated by the A-logVA function regression method （Z =-8. 688, P〈0.01 ）. And the correlation indices of LogMAR visual acuity curves obtained by the A-IogVA function was 0. 733 and that obtained by the A-SP function was 0. 715. The positive correlation was found between the subjective LogMAR acuity and that calculated by the A-SP regression method （r=0. 700,P〈 0.01 ） and between the subjective LogMAR acuity and LogMAR acuity calculated by the A-logVA regression method （r=0. 710 ,P〈0.01 ）. SVEP LogMAR acuity from A-SP function regression method was significantly different from the LogMAR acuity from A-IogVA function regression method （Z =-8. 748, P〈0.01 ）. No significant differences of VA LogMAR were found in gender, eyes, type of disease and age （χ2 = 2. 171, P = 0. 338 ; χ2=0. 976,P=0. 614;χ2=6.032,P=0. 420;xi=I4.720,P=0.257）. Conclusions SVEP can obtain the visual outcome in human. The amplitude-logVA function regression method is more accurate in extrapolating SVEP acuity.