Bach M, Kommerell G
Universitäts-Augenklinik Freiburg.
Klin Monbl Augenheilkd. 1998 Apr;212(4):190-5. doi: 10.1055/s-2008-1034863.
According to the European standard EN ISO 8596 the Landolt-C in 8 different orientations has to be used to measure visual acuity. With decreasing size of the Landolt-C the hit rate declines from 100% to the chance level of 12.5%. This gradual transition is described by the "psychometric function". The steepest point of the psychometric function is in the middle between 100 and 12.5, i.e., at 56.25%. This point of the psychometric function (approximated by 5 of 8 Landolt-Cs) has been selected as the threshold for visual acuity, because it is there that the visual acuity is influenced least by (incidental) fluctuations. The subject has to answer by forced choice; a response like "I cannot detect anything" is not acceptable. "NORMAL" VISUAL ACUITY: Cannot be assigned to a certain value, like 1.0 or 6/6. With the standard test procedure, visually healthy, young subjects achieve a visual acuity of about 2.0 or 12/6, while in senior subjects 0.5 (3/6) may be "normal". AVERAGING VISUAL ACUITY: Logarithmic, not arithmetic, scaling of visual acuity approximates the perceptual metric. Consequently, visual acuity values may not be averaged arithmetically. Instead, three steps are required: all values have to be converted to logarithms, then averaged, and finally the average can be reconverted. Geometric averaging is equivalent. "MINIMUM ANGLE OF RESOLUTION" NOT NECESSARY: MAR is the reciprocal of visual acuity. In many studies, clinical outcome has been assessed using log(MAR). Though statistically correct, this term is unnecessary, as log(acuity) has identical statistical properties. Furthermore, log(MAR) is contra-intuitive as its value becomes smaller when vision improves. COMPUTER-ASSISTED INSTRUMENTATION: Facilitates complying with the EN ISO 8596. For instance, the Freiburg Visual Acuity Test relieves the examiner from observing whether 5 responses have been correct, and that not more than 8 tests are given per level.
根据欧洲标准EN ISO 8596，必须使用8种不同方向的Landolt-C视标来测量视力。随着Landolt-C视标尺寸减小，命中率从100%逐渐下降至12.5%的随机水平。这种逐渐过渡由“心理测量函数”描述。心理测量函数最陡的点位于100%和12.5%之间的中间位置，即56.25%处。心理测量函数的这一点（由8个Landolt-C视标中的5个近似）被选作视力阈值，因为在此处视力受（偶然）波动的影响最小。受试者必须通过强制选择作答；像“我什么都看不见”这样的回答是不可接受的。“正常”视力：不能指定为某个特定值，如1.0或6/6。采用标准测试程序时，视力健康的年轻受试者视力约为2.0或12/6，而老年受试者中0.5（3/6）可能为“正常”。视力平均值：视力的对数标度而非算术标度近似于感知度量。因此，视力值不能进行算术平均。相反，需要三个步骤：所有值都必须转换为对数，然后求平均，最后将平均值再转换回来。几何平均是等效的。“最小分辨角”不必要：最小分辨角（MAR）是视力的倒数。在许多研究中，临床结果使用log(MAR)进行评估。尽管从统计学角度正确，但这个术语是不必要的，因为log(视力)具有相同的统计特性。此外，log(MAR)违反直觉，因为视力改善时其值会变小。计算机辅助仪器：便于符合EN ISO 8596标准。例如，弗莱堡视力测试使检查者无需观察5个回答是否正确，以及每个级别是否进行了不超过8次测试。
