|Year : 2013 | Volume
| Issue : 4 | Page : 36-39
Quantitative computerized color vision testing in diabetic retinopathy: A possible screening tool?
Rashid Al Saeidi, Marcus Kernt, Thomas C Kreutzer, Guenther Rudolph, Aljoscha S Neubauer, Christos Haritoglou
Department of Ophthalmology, Ludwig-Maximilians-University, Mathildenstr, Munich, Germany
|Date of Web Publication||30-Nov-2013|
Rashid Al Saeidi
Consultant Ophthalmology and President, Oman Ophthalmic Society and Sultanate of Oman
Source of Support: None, Conflict of Interest: None
| Abstract|| |
Purpose: To evaluate the efficacy of a computerized color vision testing (Arden color contrast test) as a screening test for detection of diabetic macular edema (DME).
Materials and Methods: A consecutive, prospective case series of 83 eyes of 42 diabetic patients with and without macular edema was enrolled. Macular edema was assessed clinically by stereoscopic grading and by central retinal thickness measurement with optical coherence tomography (OCT). Additionally, a computerized chromatest for the protan- and tritan-axis was performed. Analysis of test characteristics included receiver operating characteristic (ROC) curves and calculated sensitivity and specificity.
Results: Sixty-one eyes had clinically significant macular edema (CSME). OCT yielded an area under the ROC curve (AUC) of 0.92. Color vision testing yielded an AUC of 0.82 for the tritan- and 0.80 for the protan-axis. Using a cut off of 199 microns OCT resulted in a 100% sensitivity at 39% specificity. With a cut-off of 4.85, color testing yielded a sensitivity of 100% at a specificity of 8% on the tritan-axis, respectively. Considering OCT instead of clinical examination as a reference standard resulted in a comparable high sensitivity, but low specificity for color vision testing. Disturbance of the tritan axis was more pronounced than for the protan axis in present macular edema and also better correlated (r = 0.46) with retinal thickness measured with OCT.
Conclusions: Computerized, quantitative color testing using the chromatest allows detection of diabetic maculopathy with high sensitivity. However, only a low specificity exists for retinal macular edema, as in diabetic retinopathy (DR) frequently abnormalities of the tritan axis exist before any retinal thickening occurs.
Keywords: Color vision, diabetic retinopathy, diabetic macular edema, screening, OCT
|How to cite this article:|
Al Saeidi R, Kernt M, Kreutzer TC, Rudolph G, Neubauer AS, Haritoglou C. Quantitative computerized color vision testing in diabetic retinopathy: A possible screening tool?. Oman J Ophthalmol 2013;6, Suppl S1:36-9
|How to cite this URL:|
Al Saeidi R, Kernt M, Kreutzer TC, Rudolph G, Neubauer AS, Haritoglou C. Quantitative computerized color vision testing in diabetic retinopathy: A possible screening tool?. Oman J Ophthalmol [serial online] 2013 [cited 2020 Jun 2];6, Suppl S1:36-9. Available from: http://www.ojoonline.org/text.asp?2013/6/4/36/122294
| Background|| |
Diabetes mellitus is the leading cause of legal blindness and moderate visual loss amongst working age population in developed countries. Besides, a number of studies revealed that diabetes also causes impaired color discrimination. , Reduced color discrimination may lead to significant impairment in many daily life activities. In addition, the debilitating nature of untreated diabetic retinopathy (DR) promotes the need for cost-effective screening methods. Macular edema is the leading cause of visual impairment due to DR. Although seven field 30 degree stereoscopic color fundus photographs as well as fluorescence angiography, during the last years often supported by OCT, are well-established and represent the gold standard for diabetic screening, they remain either relatively difficult to obtain or expensive and invasive. Therefore, numerous programs for DR screening utilize non-stereo digital photography as a cheaper alternative with adequate sensitivity and specificity.  Non-stereoscopic fundus imaging is also easier to obtain, but may have limitations in detecting macular edema. There is evidence that tritan color vision is reduced in patients with diabetic macular edema (DME). , Therefore, testing color vision may be an additional helpful tool for (early) detection of DME. But testing with the FM100 hue and Farnsworth-Lanthony D-15 tests are labor intensive and time consuming. Automated, digital color vision testing with a computer graphics system may be an effective alternative. This study assesses the ability of the chromatest, an automated, digital color contrast sensitivity program as screening tool for detection of DME.
| Materials and Methods|| |
Patients visiting the DR unit's outpatient center at the university eye hospital of Ludwig-Maximilians University in Munich were recruited for this study. Inclusion criteria were as follows : Corrected visual acuity (VA) of 20/60 or better, type 2 diabetic patients, untreated nonproliferative DR (NPDR) and untreated clinically significant macular edema (CSME). Exclusion criteria included type 1 diabetes, PDR, previous laser photocoagulation or intravitreal pharmacologic therapy, concurrent ocular pathologies including infection, trauma, amblyopia, optic nerve and/or nondiabetic macular disease, vascular occlusive disease, as well as known color blindness. Medical history including duration of diabetes with recent HbA1c, hypertension, renal disease, and smoking were recorded. Examination of best corrected logMAR visual acuities (BCVA) was followed by color contrast sensitivity testing of each eye separately. The diabetes module of the chromatest, a software program analyzing the age-corrected tritan (TCCT) and protan color contrast thresholds (PCCT) was used for our testing. Patients were intensively instructed before starting the testing procedure. Both eyes were tested separately and in all patients the right eye was tested first followed by the left eye.
Chromatest procedure and heterochromic flicker test
For color threshold testing, the patient is seated in front of a computer monitor at a fixed distance of 1.5 m. An alphabetical letter is presented in the center of the screen at a constant angle, testing the central 6.5 degrees of the retina [Figure 1].
The system uses an individual spectral sensitivity function and calculates the needed voltage to present colors of equal luminance but of different spectrum. Therefore, it is important to know the aberration in spectral sensitivity of each test subject. This can be identified with the "heterochromic flicker test". A 4 degree square field is presented in the center of the screen and illuminated for 40 ms each by a red and green channel. This causes the impression of a flickering yellow-orange field, which has to be adjusted for least flickering. The test subject is asked - independently of the color of the field - to perform the above mentioned adjustment with a computer mouse. With this task the test subject increases or decreases the green fraction within the colored field. The test subject informs the machine when the state of least flickering is reached. The test is performed several times in order to receive a mean value. If the standard deviation is below 5% (usually after 3-4 test runs) the test is terminated and the computer compares the obtained results with a normal population. The same test procedure is also conducted for green-blue color. In the end the computer calculates for the specific test subject those colors which are of isoluminance, even if the subject has sclerotic cataract or wears partially tinted glasses. In previous analyses a tinted glass of up to 20% showed no effect on the acquired results with the heterochomic flicker test. Despite this test subjects in our study were not allowed to use tinted glasses with the test.
Measurement of central color contrast threshold values
Distance between the patient and the monitor was 1.5 m. The stimulus consisted of options of 10 letters (A, E, H, M, O, T, U, V, X, and Y) with about 8 degrees and appeared for a duration of 200 ms at the center of the screen. Depending on the individual previously identified spectral sensitivity the letters had a uniform background with equally distributed light density. The color contrast between letter and background was defined with 0% if no difference between the background and the letter was seen, and 100% when the letter was seen with maximum contrast. The three color axes red (protan), green (deutan), and blue (tritan) were investigated. The initial color contrast was set at 12% and the repetition was set to eight times per contrast level. If the test subject could recognize the first letter with 12% contrast, then it was next presented with a 6% contrast. If this was recognized again, the contrast was reduced to 3%; if not, it was increased to 9%. With this modified, binary search approach the threshold was determined within a few steps (about 7-9).
Macular thickness measurement
Central retinal thickness was measured by a certified operator using the Stratus OCT3 system using the automated macular thickness map protocol of the OCT software (Carl Zeiss Meditec, Jena, Germany). Retinal morphologic features were assessed using a 5-level grading scale for cystic spaces and a 3-level grading scale for the presence or absence of central subretinal fluid.
Clinical fundus examination
After performing chromatest and OCT all patients underwent dilated funduscopy with slit lamp biomicroscopy using a 78 D lens by a vitroretinal consultant to confirm the grading of CSME according to the Early Treatment Diabetic Retinopathy Study extension of the modified Airlie House Classification with CSME being defined as any retinal thickening within 500 microns of the center of the fovea, hard, yellow exudates within 500 microns of the center of the fovea with adjacent retinal thickening or at least one disc area of retinal thickening, any part of which is within one disc are of the center of the fovea. Since this is the first study of NPDR using the chromatest compared to OCT, threshold levels were derived using the same data set for both training and test.
| Results|| |
Clinically 22 eyes showed no edema, while 61 eyes revealed edema (edema ist nicht gleich CSME. Hatten die garkein #214;dem oder nur kein CSME?). OCT yielded an area under the reciever operating characteristic (ROC) curve (AUC) of 0.92. Color vision testing yielded an AUC of 0.82 for the tritan- and 0.80 for the protan-axis [Figure 2]. This means that OCT with a cut-off at 199 microns showed 100% sensitivity at 39% specificity. Analogously, with a cut-off at 4.85 for color contrast threshold chromatest for tritan-axis yielded a sensitivity of 100% at a specificity of only 8% for detecting CSME. Considering OCT instead of clinical examination as a reference standard resulted in a very similar picture of high sensitivity, but low specificity for color vision testing. Disturbance of the tritan axis was more pronounced than for the protan axis and also better correlated (r = 0.46) with retinal thickness measured by OCT [Figure 3]. The average testing time was 5 min.
|Figure 2: Area under the receiver operating characteristic (ROC) curve (AUC) was 0.92 for optical coherence tomography (OCT). Color vision testing yielded an AUC of 0.82 for the tritan- and 0.80 for the protan-axis|
Click here to view
|Figure 3: Disturbance of the tritan axis was more pronounced than for the protan axis and also better correlated (r = 0.46) with retinal thickness measured by OCT|
Click here to view
| Discussion|| |
DR is a potentially sight threatening disease and the major cause of blindness in patients under the age of 60 years. Therefore, it represents a strong socioeconomic challenge. DME is the leading cause for visual impairment in this group of patients. Increasing numbers of type 2 diabetes patients due to dietary and lifestyle changes underline the need for an effective and affordable ophthalmologic screening method for detection of DR and DME. Therefore, such screening methods are needed not only from a therapeutic but also from an economic point of view. It has been previously described that abnormal protan and especially tritan color vision but also blue-yellow defects may be seen in patients with DR 5 and glaucoma. ,,,
This study examined patients with NPDR using the macular modules of the chromatest. Although the exact pathomechanism of altered color vision in patients with DR is not completely understood, there is some evidence that reduced retinal oxygen saturation is associated with impaired color vision in diabetics.  In addition it has been shown that error scores in color vision are directly correlated to the severity of macular edema.  A correlation between selective loss of short wavelength pathway sensitivity and the severity of diabetic macular edema has been demonstrated. 
Our study focused on the assessment of patients with untreated CSME to ascertain the viability of such a screening method. Laser photocoagulation was an exclusion criterion in our study as it is known to affect tritan color vision.  One may argue that the lens status has an impact on the results of the color test performed in our study. We are aware that cataract formation may cause preretinal absorption of short-wavelength light resulting in tritan deficits  and may have influenced the overall sensitivity and specificity of the study. However, patients with lens opacities up to grade 2 (and pseudophakic patients) were not excluded from our study, as both conditions are common in both, our study population and in "real life" cohort of type 2 diabetic patients. Therefore, an exclusion would have interfered with the intention of our study to assess the chromatest as a screening tool.
It is worth mentioning that some patients with a highly reduced color threshold test result for either protan or tritan or tritan alone had normal central retinal thickness in OCT. This may be due to severe NPDR with or without ischemic maculopathy. Thirteen cases had ischemic maculopathy, which may have contributed to highly reduced test results in color contrast thresholds especially for Tritan. In addition, 22 cases had either severe maculopathy and edema or severe NPDR which may have also contributed to the highly reduced color threshold test results we noticed.
In contrast, we found cases with high or border line reduced color thresholds in patients suffering from early NPDR with OCT results being within normal limits.
In color contrast testing, higher TCCT or PCCT scores are correlated to more abnormal test results compared to age-matched healthy individuals. In our study, 40% of patients with NPDR had TCCT scores above normal levels.
In addition, a high percentage of CSME patients revealed TCCT scores above normal levels.
A potential limitation of our study may be that normal threshold levels were obtained from our study population by analyzing cases without CSME, as only limited data on diabetic patients is available. Therefore, the results may be biased.
Although this study investigated a large numbers of untreated CSME eyes for color vision abnormalities with the chromatest, still more data is required to substantiate our findings. However, data from this first, limited, pilot-study implicate that color contrast analysis using the chromatest may become a valuable tool for screening for DME, despite the limitations of the results; there was no discrimination for age and VA due to the ease of the test, which all patients were able to perform in a very short period of time with no pupil dilation being required as for fundus photography and fluorescein angiography.
To offer more comprehensive screening programs, cost-effectiveness is highly warranted. Currently digital photography supplemented by biomicroscopy is proposed as standard for DR and DME screening. Furthermore, OCT has become a powerful tool in screening and monitoring CSME with sensitivity and specificity rates of near 80 and 90%, respectively. Color testing is a potential additional approach to improve screening practice. Indeed, further studies are needed, but the presented early data implicate that TCCT testing may become a supplement for detecting and monitoring sight threatening pathology without the need of much equipment or trained technicians.
| References|| |
|1.||Bresnick GH, Condit RS, Palta M, Korth K, Groo A, Syrjala S. Association of hue discrimination loss and diabetic retinopathy. Arch Ophthalmol 1985;103:1317-24. |
|2.||Fong DS, Barton FB, Bresnick GH. Impaired color vision associated with diabetic retinopathy: Early Treatment Diabetic Retinopathy Study Report No. 15. Am J Ophthalmol 1999;128:612-7. |
|3.||Welty CJ, Agarwal A, Merin LM, Chomsky A. Monoscopic versus stereoscopic photography in screening for clinically significant macular edema. Ophthalmic Surg Lasers Imaging 2006;37:524-6. |
|4.||Barton FB, Fong DS, Knatterud GL. Classification of Farnsworth-Munsell 100-hue test results in the early treatment diabetic retinopathy study. Am J Ophthalmol 2004;138:119-24. |
|5.||Giusti C. Lanthony 15-Hue Desaturated Test for screening of early color vision defects in uncomplicated juvenile diabetes. Jpn J Ophthalmol 2001;45:607-11. |
|6.||Alvarez SL, Pierce GE, Vingrys AJ, Benes SC, Weber PA, King-Smith PE. Comparison of red-green, blue-yellow and achromatic losses in glaucoma. Vision Res 1997;37:2295-301. |
|7.||Nuzzi R, Bellan A, Boles-Carenini B. Glaucoma, lighting and color vision. An investigation into their interrelationship. Ophthalmologica 1997;211:25-31. |
|8.||Yamagami J, Koseki N, Araie M. Color vision deficit in normal-tension glaucoma eyes. Jpn J Ophthalmol 1995;39:384-9. |
|9.||Dean FM, Arden GB, Dornhorst A. Partial reversal of protan and tritan colour defects with inhaled oxygen in insulin dependent diabetic subjects. Br J Ophthalmol 1997;81:27-30. |
|10.||Ulbig MR, Arden GB, Hamilton AM. Color contrast sensitivity and pattern electroretinographic findings after diode and argon laser photocoagulation in diabetic retinopathy. Am J Ophthalmol 1994;117:583-8. |
|11.||Knowles PJ, Tregear SJ, Ripley LG, Casswell AG. Colour vision in diabetic and normal pseudophakes is worse than expected. Eye (Lond) 1996;10:113-6. |
| Authors|| |
[Figure 1], [Figure 2], [Figure 3]