University of Groningen. Defocus-specific contrast sensitivity Nio, Ying-Khay

Transcription

1 University of Groningen Defocus-specific contrast sensitivity Nio, Ying-Khay IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record Publication date: 2005 Link to publication in University of Groningen/UMCG research database Citation for published version (APA): Nio, Y-K. (2005). Defocus-specific contrast sensitivity s.n. Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons). Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Downloaded from the University of Groningen/UMCG research database (Pure): For technical reasons the number of authors shown on this cover page is limited to 10 maximum. Download date:

2 RIJKSUNIVERSITEIT GRONINGEN Defocus-Specific Contrast Sensitivity A psychophysical investigation of aberrations of the eye studied in healthy subjects and patients after cataract extraction and various types of refractive surgery Proefschrift ter verkrijging van het doctoraat in de Medische Wetenschappen aan de Rijksuniversiteit Groningen op gezag van de Rector Magnificus, dr F. Zwarts, in het openbaar te verdedigen op woensdag 1 juni 2005 om uur door Ying-Khay Nio geboren op 29 augustus 1966 te Zürich (Zwitserland)

3 Promotor: Prof. dr. A.C. Kooijman Copromotor: Dr. N.M. Jansonius Beoordelingscommissie: Prof. dr. G. van Rij Prof. dr. P. Artal Prof. dr. D. Stavenga ISBN:

4 Ter nagedachtenis aan Opa en Oma Nio.

5 The studies presented in this thesis were performed at the Laboratory of Experimental Ophthalmology of the State University of Groningen and at the Department of Ophthalmology of the University Medical Center Groningen, the Netherlands. Financial support of this Ph.D.-project by Pharmacia Groningen B.V. is gratefully acknow ledged. The printing of this thesis was financially supported by the Prof. Mulder Stichting, Ophtec B.V., AMO Groningen B.V. and Oculenti B.V.

6 Defocus-Specific Contrast Sensitivity A psychophysical investigation of aberrations of the eye studied in healthy subjects and patients after cataract extraction and various types of refractive surgery Defocus Specifieke Contrastgevoeligheid Een psychofysische studie naar de aberraties van het oog bij gezonde mensen en bij patiënten na een staaroperatie en diverse soorten refractiechirurgie

7 Het is te hopen voor de presbyopen dat iemand snel iets vindt en ze laat zien als kind. Maar voor het zover is gekomen en we nu alleen maar dromen, zullen aberraties ons leren dat scherptediepte ontberen het zien niet zal bekoren, asthenopie ons kan verstoren en is de moraal van dit verhaal bi-, tri-, of multifocaal.

8 Contents Page Chapter 1 9 General introduction Chapter 2 17 Influence of the rate of contrast change on the quality of contrast sensitivity assessment: a comparison of three different psychophysical methods Chapter 3 37 Age-related changes of defocus-specific contrast sensitivity in healthy subjects Chapter 4 61 Spherical and irregular aberrations are important for the optimal performance of the human eye Chapter 5 85 Effect of intraocular lens implantation on visual acuity, contrast sensitivity, and depth of focus Chapter Effect of methods of myopia correction on visual acuity, contrast sensitivity, and depth of focus Chapter summary Chapter samenvatting Dankwoord en Curriculum Vitae 144

9

10 Chapter 1: General introduction Contents 1.1 Introduction 1.2 Contrast sensitivity 1.3 Contrast sensitivity function and modulation transfer function 1.4 Optical aberrations of the human eye 1.5 From contrast sensitivity to aberrations: the eye model of Jansonius and Kooijman Scope of this thesis 9

11

12 General Introduction 1. General introduction 1.1 Introduction Good eyesight has always been important to mankind. Ancient activities like hunting and fishing require the same eye quality as driving a car or flying an airplane. A significant loss of eyesight invalids people, resulting in a poor quality of life. It is feared, therefore, by most people and, unfortunately, encountered by many. Until the beginning of the 20 th century, not much could be done to restore diminished eyesight. Nowadays, however, ophthalmology has a top ranking among other medical specialities with regard to diseases that can be cured safely. Cataract is a prominent example. Other areas that still need investigating include the prevention and treatment of eye infections in the Third World and refractive surgery, which increases the quality of life of ametropic individuals. The definition of good eyesight has varied over time. In the previous century, for example, the challenge of cataract surgery was to remove the cataractous lens. Since Ridley s famous operation in 1949, it has become possible to replace an opaque lens with an intraocular lens thus rendering the patient emmetropic or with the desired refractive state. Today, ophthalmologists are trying to treat presbyopia and will not be satisfied until optical aberrations are under control, hopefully resulting in a supernaturally high visual acuity. Good does not seem to be good enough. Obviously, new surgical techniques in cataract and refractive surgery have made it possible for us to set new standards. Yet, is a supernaturally high visual acuity the new definition of good eyesight? An aberration-free optical system will image an unusually sharp, highcontrast picture on the retinal mosaic of cones. The image we then perceive, however, will not necessarily be improved: its quality is limited by the retina and the brain, that permit a theoretical foveal acuity of approximately Moreover, there might be a side effect of aberration-free optics in the form of aliasing, a deteriorating effect of the optical image and the retinal mosaic. 3 Good results in hunting, fishing, car driving, and other activities of daily life do not, however, depend on just the image at optimal focus. Compared to the situation in the natural eye, the stronger deterioration of the image outside the focal plane in the aberration-free eye will result in a small depth of focus and may cause uncomfortable eyesight in presbyopic individuals and complaints of asthenopia in younger people. Contrast sensitivity, visual field, and viewing behaviour are also important for the performance of daily activities. 4,5 Thus, visual function is not just acuity but a combination of the complex optical and neural aspects of our visual system. The evaluation of visual function after cataract and refractive surgery should, therefore, not be limited to the objective measurement of optical aberrations of the in-focus image. It should also contain measurements that assess neural performance and optical 11

13 Chapter 1 performances outside the focal plane. The main objective of this research was to introduce psychophysical defocus-specific contrast sensitivity measurements as a tool for the evaluation of eye optics. The following paragraphs will explain how these measurements can provide a good estimation of aberrations of human eye optics. Section 1.6 (Scope of this thesis) briefly describes the research that was conducted in the search for appropriate defocus-specific contrast sensitivity parameters. It also characterizes the various populations that were studied in order to evaluate their specific eye optics. 1.2 Contrast sensitivity In the Collins Cobuild English Dictionary, the word contrast is defined as the degree of difference between the darker and lighter parts of a photograph or television picture. In ophthalmic science, contrast sensitivity is an individual s ability to distinguish relative differences in light intensity. It is usually measured psychophysically by offering stimuli with varying contrast. The subject is asked to indicate the stimulus showing the minimal amount of contrast he or she can perceive. This amount of contrast is the subject s contrast threshold; contrast sensitivity is defined by its reciprocal. There are several stimuli currently in use that test contrast sensitivity. They can be roughly subdivided into two types: letter-like optotypes and gratinglike stimuli. Tests that use letter-like optotypes are usually clinical tests, with prominent examples being the Pelli-Robson 6 and the Regan and Neima charts. 7 These tests can be performed quickly and do not require extensive instructions. Moreover, they are inexpensive because the optotypes are printed on a chart. The letters, however, consist of spatial frequencies that have variable orientations. The letters H and E, for example, require contrast sensitivity in a different direction than the letters K and W. So, even though a quick impression of a patient s contrast sensitivity can be obtained with these tests, they are not useful in scientific projects where contrast sensitivity has to be measured in an exact and standardized way. Such projects mostly use sinusoidally modulated gratings displayed on a cathode ray tube. Depending on the aim of the study, these gratings can be varied with regard to their mean luminance, spatial frequency, orientation, temporal flicker, spatial movement, and number of gratings presented. In contrast to tests using letter-like optotypes, grating tests are more laborious and need extensive instructions. Contrast can be defined in different ways. This thesis uses the Michelson contrast, which is commonly used for sinusoidally modulated gratings. The Michelson contrast is formulated as: Lmax - Lmin Contrast = L max + L min 12

14 General Introduction where L max is the maximal luminance of the bright bars and L min the minimal luminance of the dark bars. 1.3 Contrast sensitivity function and modulation transfer function The contrast sensitivity function (see, e.g., Chapter 3, Figure 2) shows contrast sensitivity as a function of spatial frequency. It depends on both eye optics and the neural processing of light that reaches the retina. In other words, it measures both optical and neural performance. The quality with which eye optics convey the sinusoidally modulated gratings originating from the cathode ray tube is represented by the optical modulation transfer function. This function can be measured objectively in different ways: double-passing techniques, 8,9 the crossed-cylinder aberroscope technique, 10 and the Hartmann-Shack wave-front sensor. 11 Neural processing of the image is characterized by the neural modulation transfer function. It can be measured using interference fringes, generated on the retina with a laser. Campbell and Green 12 compared the neural modulation transfer function with the psychophysically measured contrast sensitivity function in order to reconstruct the optical modulation transfer function. Another way to describe human eye optics using psychophysical measurements is with the relative modulation transfer at a certain spatial frequency as defined by Charman: 13 the ratio of contrast sensitivity at a certain level of defocus to contrast sensitivity focussed optimally. This parameter is determined by eye optics only, because neural modulation transfer is eliminated in the ratio of the two contrast sensitivity values. Thus, psychophysical measurements of contrast sensitivity outside the focal plane enable the evaluation of optical characteristics and give an estimate of the depth of focus. 1.4 Optical aberrations of the human eye The optical performance of the human eye is determined by diffraction and aberrations. Diffraction is inherent in the wave characteristics of light: it is the limiting factor for pupil sizes smaller than approximately 3mm. Aberrations that are due to the imperfection of eye optics diminish optical performance, especially at larger pupil sizes. There are several types of aberration. Myopia, hyperopia and astigmatism are common optical aberrations that can be corrected with spectacles. Chromatic aberration is caused by the diverse refractions of different colours of light. A comparison between monochromatic and polychromatic optical modulation transfer functions has revealed that chromatic aberration does not have a great influence. 14 This is also true for the relative modulation transfer function. 1 Besides, it is practically impossible to correct chromatic aberration of the eye, in contrast to the different monochromatic aberrations, such as spherical aberration, coma-like aberration, and irregular aberration. 13

15 Chapter 1 Spherical aberrations are caused by a relative stronger refraction of light at the periphery of the eye optics. Coma-like aberrations are mainly due to a decentration of the pupil from the optical axis. Irregular aberrations are caused by the segmented character of human eye optics and result in multiple foci. 15 Excimer lasers, one of the modern techniques of refractive surgery, are used to attempt to correct these monochromatic aberrations. 1.5 From contrast sensitivity to aberrations: the eye model of Jansonius and Kooijman 1 Each monochromatic aberration has a specific effect on the modulation transfer of the human eye. 1 Simple defocus reduces the amplitude of sinusoidally modulated gratings imaged on the retina and, therefore, reduces modulation transfer and contrast sensitivity. Spherical aberration decreases modulation transfer at optimum focus more than outside the focal plane. Therefore, it increases the relative modulation transfer. It also causes a specific myopic shift in the optimum focus for lower spatial frequencies compared to that for higher spatial frequencies. 16 Coma-like aberration substantially affects neither modulation transfer at optimum focus nor relative modulation transfer. 1 Irregular aberration decreases modulation transfer at optimum focus and increases relative modulation transfer. 1 An increase in relative modulation transfer caused by spherical and irregular aberrations results in a beneficial larger depth of focus. So, the decrease of modulation transfer at optimum focus caused by aberrations may have a positive aspect. In 1998, Jansonius and Kooijman designed an eye model for their investigation into the effect of different kinds of aberrations on (relative) modulation transfer. That eye model was used to calculate relative modulation transfer after the implementation of various amounts of spherical and irregular aberration. Depth of focus and myopic shift were then calculated from the relative modulation transfer. For the sake of convenience, two amounts of spherical aberration were explored: eye (1), which approximated the upper limit of spherical aberration, and eye (2), which represented an average amount of spherical aberration. The depth of focus and the myopic shift of these two theoretical eyes were determined for a range of irregular aberration values. By comparing these values with experimental results, one can estimate the amount of spherical and irregular aberration of a particular eye. 1.6 Scope of this thesis The aim of this investigation was to study the possibility of evaluating eye optics in a psychophysical way. First, a suitable psychophysical method to measure vast amounts of contrast sensitivity functions in a relatively short time span had to be found. Chapter two describes a study in which two existing psychophysical methods, using different rates of contrast change, were compared with a golden standard in order to find the most efficient 14

16 General Introduction method. Then, contrast sensitivity was measured in a large healthy population, subdivided into different age categories (range: years). Chapter three reports the effect of age on contrast sensitivity in the same population. The results were used later as a reference in a study of pseudophakic subjects. Chapter four compares the experimental results found in the reference population with the results from the eye models of Jansonius and Kooijman. Contrast sensitivity at optimum focus, myopic shift, and depth of focus were found to be suitable parameters for the evaluation of eye optics. These parameters were used in later studies described in Chapters five and six. Chapter five compares a group of eleven pseudophakic subjects with an agematched reference group, derived from the population described in Chapter three. Chapter six presents the results of measurements concerning different techniques of refractive surgery. Low myopic individuals with either soft contact lenses or Intacs were compared with low myopic subjects whose eyesight was corrected with spectacles. High myopic individuals with either LASIK, Artisan claw lenses, or rigid contact lenses were compared with a matching group wearing spectacles. 15

17 Chapter 1 References 1. Jansonius NM, Kooijman AC. The effect of spherical and other aberrations upon the modulation transfer of the defocussed human eye. Ophthalmic Physiol Opt 1998; 18: Schwiegerling J. Theoretical limits to visual performance. Surv Ophthalmol 2000; 45: Williams DR, Collier R. Consequences of spatial sampling by a human photoreceptor mosaic. Science 1983; 221: Coeckelbergh, T. R. M. Thesis: Effect of compensatory viewing strategies on practical fitness to drive in subjects with visual field defects caused by ocular pathology. Rijksuniversiteit Groningen, The Netherlands, Hess R, Woo G. Vision through cataracts. Invest Ophthalmol Vis Sci 1978; 17: Pelli DG, Robson JG, Wilkins AJ. The design of a new letter chart for measuring contrast sensitivity. Clin Vision Sci 1988; 2: Regan D, Neima D. Low-contrast letter charts as a test of visual function. Ophthalmology 1983; 90: Campbell FW, Gubisch RW. Optical quality of the human eye. J Physiol 1966; 186: Artal P, Navarro R. Monochromatic modulation transfer function of the human eye for different pupil diameters: an analytical expression. J Opt Soc Am A 1994; 11: Howland B, Howland HC. Subjective measurement of high order aberrations of the eye. Science 1976; 193: Liang J, Grimm B, Goelz S, Bille JF. Objective measurement of wave aberrations of the human eye with the use of a Hartmann-Shack wave-front sensor. J Opt Soc Am A 1994; 11: Campbell FW, Green DG. Optical and retinal factors affecting visual resolution. J Physiol 1965; 181: Charman WN. Effect of refractive error in visual tests with sinusoidal gratings. Br J Physiol Opt 1979; 33: Bour LJ. MTF of the defocused optical system of the human eye for incoherent monochromatic light. J Opt Soc Am 1980; 70: Van den Brink G. Measurements of the geometrical aberrations of the eye. Vision Res 1962; 2: Green DG, Campbell FW. Effect of focus on the visual response to a sinusoidally modulated spatial stimulus. J Opt Soc Am 1965; 55:

18 Chapter Two Influence of the rate of contrast change on the quality of contrast sensitivity assessment: a comparison of three psychophysical methods Y. K. Nio 1,2, N. M. Jansonius 1,3, P. Lamers 1, A. Mager 1, J. Zeinstra 1 1, 3, 4, and A. C. Kooijman 1 Laboratory of Experimental Ophthalmology, University of Groningen, PO Box , 9700 RB Groningen, the Netherlands 2 Department of Ophthalmology, Het Spittaal Gelre Ziekenhuizen, Zutphen, the Netherlands 3 Department of Ophthalmology, University Hospital Groningen, the Netherlands 4 Visio National Foundation for the Visually Impaired and Blind, the Netherlands

19

20 Influence of the Rate of Contrast Change on Contrast Sensitivity Assessment Ophthalmic and Physiological Optics 2005; 25: Abstract Criterion-free forced-choice procedures for measuring contrast sensitivity with a cathode ray tube (CRT) have low within-subject, intersubject, and test-retest variabilities, but a long test time compared with psychophysical methods that rely on the subject s criterion to determine threshold. Test time and variability of criterion-dependent methods depend on the rate at which the contrast changes on the CRT display. This study compared two criterion-dependent psychophysical methods for measuring contrast sensitivity (the method of increasing contrast and the von Békésy tracking method) with a criterion-free two-alternative forced-choice procedure. A range of rates of contrast change was studied: 0.1, 0.3, 0.5, 0.7, and 1.0 log unit s -1. Contrast sensitivity, within-subject variability, intersubject variability, testretest variability, and test time of the three methods were compared. The 2-AFC procedure performed best with regard to within-subject, intersubject, and test-retest variabilities. A time-efficient alternative was the von Békésy tracking method at rates between 0.1 and 0.5 log unit s

21 Chapter 2 Introduction Contrast sensitivity measurements have been used by scientists and practitioners for almost four decades. Over the years, these tests have proved their worth in evaluating different kinds of pathology, including glaucoma, 1 cataract, 2 and multiple sclerosis. 3 An interesting application today is the subjective study of the eye s optics after intraocular lens implantation or refractive error correction. 4,5,6,7,8 Contrast sensitivity measurements at different spatial frequencies, yielding the so-called contrast sensitivity function, evaluate visual function more comprehensively than the Snellen visual acuity because the latter indicates only the maximal resolution of the eye for high-contrast stimuli. The various reasons for contrast-threshold testing result in different requirements for the measuring method used. Screening tests, for example, require a high sensitivity and specificity, whereas tests meant to monitor changes in patients demand a low test-retest variability. Practitioners have a strong preference for fast, simple tests that anyone can perform and that are easy to interpret, while scientists usually want to invest more time and effort in order to obtain a more reliable result. Statistical considerations and practical features are important elements in any method. Contrast sensitivity measured using a computer-driven cathode ray tube (CRT) depends on several test parameters. A few examples are luminance, spatial frequency, defocus, and pupil diameter. Even when these test parameters are kept constant, threshold levels of contrast determined using different psychophysical methods are not equivalent. 9 In selecting a method, both total test time and accuracy should be considered. Psychophysical methods that are often applied in contrast sensitivity measurements include the method of increasing contrast, the von Békésy tracking (VBT) method, the method of adjustment, and forced-choice procedures. Of these, only the time-consuming forced-choice procedures are criterion-free, i.e. independent of a subject s judgement. With the other methods, contrast levels vary and the subjects are requested to indicate when the contrast has reached a perceptible level. Thus, the results of these tests vary with the visibility criterion of the subjects. The advantage of these methods over forced-choice procedures, however, is the shorter test time. Wetherill and Levitt 10 designed the Up-and-Down Transformed Response Rule for a two-alternative forced-choice (2-AFC) procedure, a short version of a forced-choice procedure, which was used in the present study. A short version of this procedure was reported by Higgins et al. 11 It needs about 12 minutes to measure the contrast sensitivity function at five spatial frequencies, using a limited number of 35 trials per spatial frequency. The duration of noncriterion-free psychophysical tests can be shortened by increasing the rate with which the stimulus contrast changes, by decreasing the number of trials, and 20

22 Influence of the Rate of Contrast Change on Contrast Sensitivity Assessment by starting the test at an initial contrast level that lies close to the expected contrast threshold. These measures, however, may have adverse effects on the accuracy of the measurement. Usually, an individual is subjected to several trials in order to assess his/ her final contrast sensitivity, i.e. the mean of the contrast sensitivity derived from each trial. The within-subject variability can then be expressed as the standard deviation of this mean. Ginsburg and Cannon 9 determined that the within-subject variability using the VBT method was 1.5 db greater than that using the method of increasing contrast. Similar conclusions concerning the VBT method were drawn by Corwin and Richman, 12 Long and Tuck, 13 and Gilmore et al. 14 Tweten et al., 15 however, discovered that the Nicolet CS-2000 Vision Tester used by the authors mentioned above gives an artifactually high variability for the VBT method because it calculates the variability for the combined appearance and disappearance thresholds. In this respect, it is worthwhile to reconsider the VBT method. A second way to express quality of measurements is with intersubject variability. One measure for this variability is the standard deviation of the data derived from different subjects. The intersubject variability of the data obtained with the VBT method approximately equals that of the method of increasing contrast. 15 As the rate of contrast change determines the test time and influences the assessed contrast sensitivity, it would be interesting to study intersubject variability as a function of this rate. To our knowledge, no such study has been performed to date. A third aspect of measurement quality is the test-retest variability. This variability describes how closely measurements, which are repeated over several days under identical conditions, match in one subject. It can be expressed as the standard deviation of contrast sensitivity values measured on different days. The goal of the present study was to collect information about the trade-off between the total test time and the quality of different methods of measuring contrast sensitivity with a CRT. For this purpose, a comparison was made between the within-subject variability, intersubject variability, and testretest variability of three psychophysical measuring methods: the method of increasing contrast, the VBT method, and the criterion-free 2-AFC procedure. The latter method was used as a reference. When studying the method of increasing contrast and the VBT method, measurements were taken with a range of contrast change rates. Methods Subjects The study population comprised six students of the University of Groningen, aged between 19 and 21 years. All were free of any ocular pathology and all had a visual acuity of 6/6 (20/20) or more. Contrast sensitivity measurements 21

23 Chapter 2 were performed monocularly with natural pupils. The eye with the highest best-corrected visual acuity was chosen. When visual acuity was equal in both eyes, the dominant eye was selected. Test equipment The CRT (Joyce DM4 monitor, Cambridge Research Systems Ltd., Rochester, Kent, UK, P31 phosphor, peak wavelength 520 nm) used to display the vertical sinusoidally-modulated gratings had a mean luminance of 200 cd m -2 and a size of degrees. It was surrounded without seams by an equiluminant screen measuring degrees. The sinusoidally-modulated gratings were created by a software-driven video card (Cambridge Research Systems Ltd., Rochester, Kent, UK: VSG 2/3 version 4.02). The subjects viewed the monitor at a viewing distance of two meters for which they were optically corrected. Their heads were stabilized by means of a chin rest and a head cushion. A more detailed description of the eye-target alignment can be found in Nio et al. 16 As noted above, three different psychophysical methods were compared: the 2-AFC procedure, the VBT method, and the method of increasing contrast. Each method yielded contrast sensitivity functions composed of five spatial frequencies: 1, 2, 4, 8, and 16 cpd (because of software limitations, the highest spatial frequency used with the 2-AFC procedure was 14.9 cpd). The order in which the spatial frequencies were measured was randomized per contrast sensitivity function. To determine the effect of the different rates of contrast change on the results of the criterion-dependent methods, measurements were performed at rates of 0.1, 0.3, 0.5, 0.7, and 1.0 log unit s -1. An increase of 0.3 log unit s -1 meant a doubling of contrast each second. The order in which these rates of contrast change were used was randomized per method and the measurement of a contrast sensitivity function at a certain rate of contrast change was finished before another rate was applied. Likewise, the order in which the methods were used was random and all measurements using a certain method were finished before another method was employed. All three methods were tested in one day. Total test time per subject per day was usually about two hours, including pauses between measurements. To study the effect of repeated measurements, the entire procedure was repeated on two other days within a period of two weeks. The resulting log contrast thresholds and the duration of the tests at different rates of contrast change were compared with each other and with the criterion-free 2-AFC procedure. Contrast sensitivity was defined as the inverse of the contrast threshold measured. Thus, log contrast sensitivity equals log contrast threshold. 22

24 Influence of the Rate of Contrast Change on Contrast Sensitivity Assessment Forced-choice procedure A spatial 2-AFC procedure was used in which the monitor was divided into an upper and a lower screen. The subjects had to indicate in which half the gratings were located by pressing a button. The presentation time of the gratings was 4 seconds, preceded by a gradual appearance and followed by a gradual disappearance over a period of one second in order to avoid any influence of temporal flickering on the log contrast threshold. First, a quick impression of the subject s log contrast threshold was obtained with a one-down (0.50 log unit)-one-up (0.25 log unit) staircase test, starting at an initial contrast level of 100%. In total, the individuals were subjected to 25 trials per spatial frequency. This resulted in a log contrast threshold value, which was used as the initial value of contrast for a second 2-AFC procedure which determined the subject s threshold level of contrast more precisely: the Up-and-Down Transformed Response Rule for two-alternative forced-choice procedures.10 This second procedure, consisting of 80 trials per spatial frequency, was a two-down-one-up staircase in which contrast was decreased 0.05 log units with two correct responses and increased 0.05 log units with one incorrect response. Compared with the one-down-one-up staircase, this test is less vulnerable to guessing as confirmation is required before the contrast is lowered. The two-down-one-up rule method determines the 71% point of seeing on the psychometric function 10. A three-down-one-up rule withstands guessing even better and, therefore, determines a higher point on the psychometric function. However, it takes much more time. Not all 80 trials were considered in the calculation of the log contrast threshold for a certain spatial frequency. In the course of the test, each reversal of contrast resulted in a peak or valley. Peaks were formed when the course of contrast changed from an increase to a decrease because of two or more successive correct responses. Valleys were formed when two or more correct responses were followed by an incorrect response (Figure 1a). In order to avoid any influence from the initial contrast setting, the first four reversals were omitted from the calculation of the log contrast threshold. A large number (n=80) of trials was necessary to obtain at least 10 reversals (5 peaks and 5 valleys) from which the log contrast threshold could be calculated. If there was an unequal number of peaks and valleys, the last reversal was also ignored. The mean of the remaining reversal points, expressed in log contrast, was the log contrast threshold for the spatial frequency concerned. Method of increasing contrast The log contrast threshold was determined by increasing the contrast level at a constant logarithmic rate, starting from a level well below the human contrast threshold: (i.e %). When the subjects detected the gratings, they were instructed to respond immediately by pressing a button on the response box. After a response was made, the contrast returned to its initial 23

25 Chapter 2 level and the second trial was started. Six trials were performed per spatial frequency (Figure 1b) and there was a variable interval of 1-3 seconds between trials so that rhythmical responses were avoided. The results of the first trial were excluded in order to prevent noise arising from the habituation of the subject during the previous set of trials. Trials with the highest and lowest contrast thresholds were also excluded. The average of the remaining three trials, expressed in log contrast, was then considered to be the threshold for the spatial frequency concerned. Von Békésy tracking method This method is based on the upper and lower limits of contrast thresholds by measuring contrast levels at which the gratings appear and disappear, respectively. The initial level of contrast in this study was (i.e %). Then contrast was increased at a constant logarithmic rate. As soon as the subject detected a grating, he/she was instructed to push and hold a button after which the contrast decreased at the same rate as it had increased. The button was released when the subject doubted the presence of gratings, thus initiating a subsequent contrast increase. The possible influence of afterimages was minimized with these instructions. The log contrast threshold of each spatial frequency was based on six peaks and six valleys (Figure 1c). The first, highest, and lowest values of both the peaks and the valleys were excluded to reduce the influence of randomization of spatial frequencies and extreme values, respectively. The log contrast values of the remaining three peaks and three valleys were averaged to calculate the log contrast threshold. Figure 1a 24

26 Influence of the Rate of Contrast Change on Contrast Sensitivity Assessment Figure 1b Figure 1c Figure 1a-c: Simulations of the course of contrast during a two-down-one-up 2-AFC procedure (a), the method of increasing contrast (b), and the von Békésy tracking method (c). Trials that were not included in the calculation of the final contrast threshold are marked with an X. 25

27 Chapter 2 Within-subject variability Within-subject variability can be expressed as the standard deviation of the mean (i.e. the standard error) of the log contrast threshold measured in a number of trials. With the 2-AFC procedure and the VBT method, the log contrast threshold at a certain spatial frequency was calculated from the average of the included peaks and valleys. It must be kept in mind, however, that the calculation of the standard error of all data points, both peaks and valleys, will result in an overestimation. The peaks and valleys have to be considered as two variables that are averaged to calculate the log contrast threshold. A better estimation of the standard error of the log contrast threshold [S.E.(log CT)] is given by: S.E.(logCT)= S.E.(p)2 + S.E.(v) 2 2 where S.E.(p) is the standard error of the peaks and S.E.(v) the standard error of the valleys. The S.E.(log CT) for the method of increasing contrast is equal to the standard deviation of the three included trials divided by the square root of three. The six subjects underwent contrast sensitivity measurements at five spatial frequencies on 3 days, resulting in 90 values of log contrast thresholds (six subjects x five spatial frequencies x 3 days) and 90 values of S.E.(log CT) for each psychophysical method. The average of all 90 S.E.(log CT) values was then taken as the within-subject variability for a certain method. The results of the VBT method and the method of increasing contrast were compared to the 95% confidence interval of the within-subject variability of the criterion-free 2-AFC procedure, which we took as reference. Intersubject variability The intersubject variability of a test is a measure of the distribution of the individual test results. This study used the standard deviation of the log contrast threshold values of all six subjects at a certain spatial frequency as the measure of the intersubject variability. Each psychophysical method was used on three different days to measure the log contrast thresholds of the five spatial frequencies. In total, 15 standard deviations (3 days five spatial frequencies) were averaged per psychophysical test as a measure of its intersubject variability. The results of the VBT method and the method of increasing contrast were compared with the 95% confidence interval of the intersubject variability of the 2-AFC procedure. 26

28 Influence of the Rate of Contrast Change on Contrast Sensitivity Assessment Test-retest variability Test-retest variability is a measure of how well the results of a certain test can be reproduced. Per method, we first calculated the standard deviation of the log contrast thresholds measured per subject over 3 days for each spatial frequency. This resulted in 30 standard errors (six subjects x five spatial frequencies). Then, these standard deviations were averaged to obtain an overall measure of test-retest variability for the method concerned. The results of the VBT method and the method of increasing contrast were compared with the 95% confidence interval of the test-retest variability of the 2-AFC procedure. Results All of the results of the contrast sensitivity tests are expressed in the common format of log contrast sensitivity, i.e. log contrast threshold. The mean contrast sensitivity functions ±1 S.E. obtained with the method of increasing contrast on the first day of measurement are shown in Figure 2 for different rates of contrast change. These functions are compared with that of the 2-AFC procedure. The error bars in the figure illustrate that the method of increasing contrast resulted in a lower contrast sensitivity than the 2-AFC procedure at almost every rate of contrast change. Moreover, contrast sensitivity shows a significant decrease with increasing rate and, for every spatial frequency, the contrast sensitivity is approximately 0.5 log units lower at a rate of 1.0 log unit s -1 compared with that at 0.1 log unit s -1. The measurements on the second and third day showed similar results. Figure 3 presents the mean contrast sensitivity functions ±1 S.E. measured on the first day using the VBT method and the 2-AFC procedure. Unlike the contrast sensitivity functions of the method of increasing contrast, the curves of the VBT method are located around that of the 2-AFC procedure; there is an exception at 8 cpd. In addition, contrary to the result obtained by the method of increasing contrast there is an increase in contrast sensitivity with an increasing rate of contrast change. The difference between the curves of 0.1 and 1.0 log unit s -1 is about 0.25 log units for every spatial frequency. The results on the second and third day were similar. 27

29 Chapter 2 Figure 2 Contrast sensitivity functions obtained with the method of increasing contrast at different rates of contrast change and with the 2-AFC procedure on the first day of measurement. The curves represent the mean contrast sensitivity ±1 S.E. of six subjects. Figure 3 Contrast sensitivity functions obtained with the von Békésy tracking method at different rates of contrast change and with the 2-AFC procedure on the first day of measurement. The curves represent the mean contrast sensitivity ±1 S.E. of six subjects. 28

30 Influence of the Rate of Contrast Change on Contrast Sensitivity Assessment Figure 4 Test time (minutes) to assess the log contrast threshold at five different spatial frequencies (i.e. one contrast sensitivity function) as a function of the rate of contrast change (log unit s -1 ). The curves represent mean test time ±1 S.E., averaged over 3 days and six subjects for both the method of increasing contrast and the von Békésy tracking method. Test time of the 2-AFC procedure was 31 minutes. Figure 4 presents the mean test time ±1 S.E. needed to assess one contrast sensitivity function as a function of the rate of contrast change. For both the method of increasing contrast and the VBT method, the test time was averaged over 3 days and over the six subjects per rate of contrast change. The largest decrease in test time was found between 0.1 and 0.3 log unit s -1. The method of increasing contrast required approximately the same test time as the VBT method at similar rates of contrast change. The average test time of the 2-AFC procedure for one contrast sensitivity function was 31 minutes. The mean within-subject variability ±1 S.E. for both the VBT method and the method of increasing contrast is presented in Figure 5 for each rate of contrast change. The upper and lower limits of the 95% confidence interval of the within-subject variability of the 2-AFC procedure are also shown. Regression analysis showed a significant linear increase of within-subject variability with an increasing rate of contrast change for the VBT method (p < 0.01). The within-subject variability of the method of increasing contrast was independent of the rate of contrast change. The within-subject variability of the VBT method was similar to that of the 2-AFC procedure at rates of contrast change below 0.5 log unit s -1, and comparable with that of the method of increasing contrast at rates of 0.7 log unit s -1 and higher. 29

31 Chapter 2 Figure 5 Mean within-subject variabilities ±1 S.E., averaged over 90 S.E. (log CT) values, of the method of increasing contrast and the von Békésy tracking method compared with the 95% confidence interval of the within-subject variability of the 2-AFC procedure. The mean within-subject variability of the 2-AFC procedure was Figure 6 Mean intersubject variabilities ±1 S.E., averaged over 3 days and five spatial frequencies, of the method of increasing contrast and the von Békésy tracking method compared with the 95% confidence interval of the intersubject variability of the 2-AFC procedure. The mean intersubject variability of the 2-AFC procedure was

32 Influence of the Rate of Contrast Change on Contrast Sensitivity Assessment Figure 6 compares the mean intersubject variability ±1 S.E. of the VBT method, the method of increasing contrast, and the 95% confidence interval of that of the 2-AFC procedure. The intersubject variability ranged between 0.14 and 0.22 for the VBT method and between 0.17 and 0.25 for the method of increasing contrast. The intersubject variability for the 2-AFC procedure was There was no significant difference between the VBT method and the method of increasing contrast at any rate of contrast change. Moreover, similar values for intersubject variability were found for all three test methods at all but the 1.0 log unit s -1 contrast change rate. Regression analysis showed a significant linear increase with an increasing rate of contrast change for both the VBT method and the method of increasing contrast (p < 0.05). Figure 7 Mean test-retest variabilities ±1 S.E., averaged over six subjects and five spatial frequencies, of the method of increasing contrast and the von Békésy tracking method compared with the 95% confidence interval of the test-retest variability of the 2-AFC procedure. The mean test-retest variability of the 2-AFC procedure was Figure 7 shows the mean test-retest variability results ±1 S.E. The 2-AFC procedure was the most consistent method during the 3 days of measurement. The method of increasing contrast and the VBT method had more or less comparable variabilities. However, the method of increasing contrast showed a significantly increasing test-retest variability with an increasing rate of contrast change (p < 0.05), while the test-retest variability of the VBT method was independent of the rate of contrast change. An ANOVA for repeated measures was performed on the contrast sensitivity data to study possible learning 31

33 Chapter 2 effects. No significant within-subjects effect of days was found for any of the methods nor for any rate of contrast change, i.e. there was no learning effect (p > 0.05). Discussion This study compared the within-subject variability, intersubject variability, and test-retest variability of two psychophysical methods of measuring contrast sensitivity (the VBT method and the method of increasing contrast) at different rates of contrast change to those of the 2-AFC procedure. The latter method was considered as a reference for the other methods because it is independent of the criterion of subjects. The 2-AFC procedure was not designed to be fast, but to deliver reliable and stable results. The other two methods require less test time. The purpose of this study was to discover the costs for this gain in time and to find a reasonable trade-off. The contrast sensitivity found with the method of increasing contrast decreased with an increasing rate of contrast change, as was also noted by Ginsburg and Canon. 9 Reaction time seems to influence the results of the method of increasing contrast more at higher rates. Reaction time is also known to increase with spatial frequency. 17,18 In our study, however, this decrease was less obvious than that caused by the rate of contrast change (Figure 2). With the VBT method the decreasing effect of reaction time on contrast sensitivity will be compensated by the reaction time needed to respond to the vanishing gratings. In fact, increased contrast sensitivity was found at higher rates of contrast change with the VBT method. According to Ginsburg and Cannon, 9 this effect could be caused by afterimages. It is less disturbing, however, than the effect of reaction time with the method of increasing contrast. On all 3 days, the 2-AFC procedure showed an apparently high contrast sensitivity at 8 cpd compared with the method of increasing contrast, the VBT method, and other studies using the 2-AFC procedure. 11,19 The monitor used with the 2-AFC procedure is divided into an upper and a lower screen, resulting in a smaller height (2.9 degrees) of the gratings shown. For the spatial frequencies we measured, Howell and Hess 20 found a critical grating height of 10 to 20 cycles, below which contrast sensitivity decreases. So, the relation between spatial frequency and grating height is relatively more disturbed at spatial frequencies of 4 cpd and lower, which may have resulted in a relatively low contrast sensitivity at these frequencies found in this study. Nevertheless, we do not think this influenced our conclusions regarding within-subject variability, intersubject variability, or test-retest variability. The within-subject variability of the method of increasing contrast was found to be significantly larger than that of the VBT method at the rates of 0.1 and 0.3 log unit s -1 (p < 0.01). This is in contrast to the intrasubject standard deviation noted in the study by Ginsburg and Cannon: 9 they showed a 1.5 db larger standard deviation for the VBT method than for the method of 32

34 Influence of the Rate of Contrast Change on Contrast Sensitivity Assessment increasing contrast. As mentioned earlier, the within-subject variability of the VBT method using the Nicolet CS-2000 Vision Tester in the study by Ginsburg and Cannon was erroneously high. In our study, the within-subject variability of the 2-AFC procedure was significantly smaller than that of the method of increasing contrast at all rates (p < 0.01). Only the VBT method had a comparable variability at rates of 0.1 and 0.3 log unit s -1 contrast change. So, if it is important to determine contrast sensitivity quickly with a small standard error, the VBT method at 0.1 and 0.3 log unit s -1 can be an alternative for the 2-AFC procedure. Intersubject variability is a measure of the distribution of contrast sensitivity values among subjects. Higgins et al. 21 found an intersubject variability of log units across the spatial frequencies they measured using the 2-AFC procedure. The intersubject variabilities described in the study by Ginsburg and Cannon 9 for the method of increasing contrast and the VBT method were larger: 0.2 and 0.4, respectively. In the latter study, the rate of contrast change was 0.14% contrast per second for the method of increasing contrast and 0.08% for the VBT method, corresponding to 0.06 and 0.03 log unit s -1, respectively. Thus, a rate of contrast change lower than the lowest value used in our study does not seem to guarantee a better intersubject variability. Similar values for intersubject variability (0.3 log units) for both methods were found in a study by Tweten et al. 15 These researchers changed the contrast exponentially from zero to 50% and back to zero in 60 seconds. No rate of log contrast change can be calculated from this information, however, as it is not possible to reach 0% contrast with an exponential rate of contrast change. The literature reports the intersubject variability of the VBT method to vary from to log units. The intersubject variability in our study seems small compared with that found in other studies: between 0.14 and 0.22 log units for the VBT method, between 0.17 and 0.25 log units for the method of increasing contrast, and 0.14 log units for the 2-AFC procedure. These differences might reflect differences in the research populations studied. The intersubject variability of a method can be influenced by the within-subject variability. In the present study, however, the within-subject variabilities were small compared with the intersubject variabilities. Therefore, it is unlikely that differences in the intersubject variability between the 2-AFC procedure and the other two methods were influenced much by the withinsubject variability. The instructions and training sessions were apparently sufficient to prevent a significant learning curve in 3 days of measurement. The test-retest results show that the 2-AFC procedure is the most reliable method with regard to repeated measurements. The test-retest variability of the method of increasing contrast increased with an increasing rate of contrast change. There was no clear relation between test-retest variability and the rate of contrast change using the VBT method. For both methods, however, the lowest rate of contrast 33