Unresolved Issues in Prediction of Subjective and Objective Refraction from Wavefront Data

Size: px

Start display at page:

Download "Unresolved Issues in Prediction of Subjective and Objective Refraction from Wavefront Data"

Pierce York
6 years ago
Views:

1 Wavefront Congress Symposium Feb, 2008 Unresolved Issues in Prediction of Subjective and Objective Refraction from Wavefront Data Larry N. Thibos School of Optometry, Indiana University, Bloomington, IN Slideshow in public domain at

2 6 questions without answers 1) Which subjective criterion is to be emulated: perceived or performance IQ? 2) Do metrics need to include the neural system, or is Rx just optics? 3) What are the sources of discrepancy between subjective and objective Rx? 4) How close is close enough? 5) Mono vs. poly chromatic IQ: does it matter? 6) Is there a systematic error due to depth of retinal beacon? If so, does it depend on wavelength?

$Traditional subjective refraction$ Retinal Image (inaccessible) Vision

3 Traditional subjective refraction Caveman Refractionist Wow! Retinal Image (inaccessible) Vision (perceivable) Patient decides which lens is better, #1 or #2

$The subjective refraction paradigm Retinal image #1 Visual Object Filter #1$

4 The subjective refraction paradigm Retinal image #1 Visual Object Filter #1 task Retinal image #2 Observer Filter #2 Which optical filter (i.e. lens) is better?

5 Objective refraction Visual target Mini-clinician assesses image quality Adjust distance Find the viewing distance (or correcting lens) for which retinal image quality is maximized.

Measure wavefront aberration and compute the correcting

6 Objective wavefront refaction Maxiclinician inspects reflected wavefront Δy Δx Aberrated rays P object point Measure wavefront aberration and compute the correcting lens needed to optimize an external image of the fundus.

$Wavefront refraction: finding the best$ Fourier of math optics Computed Image IQ

7 Wavefront refraction: finding the best correcting lens Aberration map Lots Fourier of math optics Computed Image IQ Is IQ max? Yes No Add sphero-cylindrical wavefront

8 6 questions without answers 1) Which subjective criterion is to be emulated: perceived IQ or performance IQ? 2) Do metrics need to include the neural system, or is Rx just optics? 3) What are the sources of discrepancy between subjective and objective Rx? 4) How close is close enough? 5) Mono vs. poly chromatic IQ: does it matter? 6) Is there a systematic error due to depth of retinal beacon? If so, does it depend on wavelength?

9 What does "better" mean?

10 The subjective refraction paradigm Retinal image #1 Visual Object Filter #1 task Retinal image #2 Observer Filter #2 Choice of metric for ranking optical filters depends on the task. Task A: read the letters Task B: judge perceived quality

11 6 questions without answers 1) Which subjective criterion is to be emulated: perceived or performance IQ? 2) Do metrics need to include the neural system, or is refraction just optics? 3) What are the sources of discrepancy between subjective and objective Rx? 4) How close is close enough? 5) Mono vs. poly chromatic IQ: does it matter? 6) Is there a systematic error due to depth of retinal beacon? If so, does it depend on wavelength?

12 Objective prediction of perceived image quality Granger & Cupery (1972) Barten (1987) Area = SQF Area = SQRI 3 12

13 6 questions without answers 1) Which subjective criterion is to be emulated: perceived or performance IQ? 2) Do metrics need to include the neural system, or is Rx just optics? 3) What are the sources of discrepancy between subjective and objective Rx? 4) How close is close enough? 5) Mono vs. poly chromatic IQ: does it matter? 6) Is there a systematic error due to depth of retinal beacon? If so, does it depend on wavelength?

14 How large is the discrepancy? Defocus (M) Astigmatism (J 0 or J 45 ) Objective best lens (D) R R Subjective best lens (D) Cheng et al., 2004 (J. Vision) computationally-blurred images, monochromatic light 3 levels of 3rd & 4th order Zernike aberrations letter size & lens power (M, J 0 or J 45 ) varied to maximize VA

15 Real-world clinical refractions Subjective refraction uses polychromatic light Whereas, optical measurements of wavefront aberrations are monochromatic, typically with "invisible light" (infra-red) Subjective hyperfocal refractions deliberately leave the eye myopic (i.e. undercorrected) Whereas, wavefront refraction aims to optimize image quality for infinity, not the hyperfocal distance. How large is the discrepancy between subjective and objective refractions in the clinic?

16 6 questions without answers 1) Which subjective criterion is to be emulated: perceived or performance IQ? 2) Do metrics need to include the neural system, or is Rx just optics? 3) What are the sources of discrepancy between subjective and objective Rx? 4) How close is close enough? 5) Mono vs. poly chromatic IQ: does it matter? 6) Is there a systematic error due to depth of retinal beacon? If so, does it depend on wavelength?

17 Quantifying error of astigmatism predictions J 45 Subjective Prediction Error J 0 Objective J 0 = 0.5*Cyl*cos(2*axis) = c 2 +2 /r 2 Scatterplot of prediction errors J 45 J45 + J0 J 0 J 45 = 0.5*Cyl*sin(2*axis) = c 2-2 /r 2 Accuracy (systematic error) = distance of population mean from origin Precision (random error) = mean radius of 95% confidence ellipse

18 Astigmatism results (Coe, WFC 2004) The good news: Many metrics of WQ, IQ, and VQ make highly accurate predictions of subjective Rx (systematic error < 0.05D) Suggests errors in predicting astigmatism are random. For the most precise metrics, 95% of objective predictions are within 0.25 D of subjective measurements. This might be as good as we can expect, given variability in the gold standard of subjective refractions. It is hard to hit a moving target.

19 Defocus results (Coe, WFC 2004) The bad news: Accurate and precise predictions of spherical defocus have proven elusive, even when using unbiased "infinity" refractions designed to make the retina optically conjugate to infinity, rather than the hyperfocal distance used clinically. This discrepancy might be due to various features of ocular chromatic aberration.

20 6 questions without answers 1) Which subjective criterion is to be emulated: perceived or performance IQ? 2) Do metrics need to include the neural system, or is Rx just optics? 3) What are the sources of discrepancy between subjective and objective Rx? 4) How close is close enough? 5) Mono vs. poly chromatic IQ: does it matter? 6) Is there a systematic error due to depth of retinal beacon? If so, does it depend on wavelength?

21 Predicting defocus (M) is more challenging Accurate and precise prediction of defocus (M) is hampered by ocular chromatic aberration for several reasons: 1. Wavefront aberrometers using infrared light are subject to potential measurement bias. 2. Wavefronts measured in infrared must be converted to a visible wavelength using some model of ocular chromatic aberration (OCA). The model could be wrong. 3. Although OCA models should include higher-order chromatic aberrations, HOCAs are probably less important than individual variation in lower-order OCA. 4. Polychromatic calculations of defocus (M) require knowledge of wavelength-in-focus when white-light targets were optimally focused subjectively.

22 6 questions without answers 1) Which subjective criterion is to be emulated: perceived or performance IQ? 2) Do metrics need to include the neural system, or is Rx just optics? 3) What are the sources of discrepancy between subjective and objective Rx? 4) How close is close enough? 5) Mono vs. poly chromatic IQ: does it matter? 6) Is there a systematic error due to depth of the retinal beacon used to measure the eye? If so, does it depend on wavelength?

$A possible explanation for myopic bias in IR Classical, refractive CA IR light$

23 A possible explanation for myopic bias in IR Classical, refractive CA IR light may reflect from deeper layers of fundus, making eye appear myopic. Reflective CA??

24 A possible artifact of aberrometry Melanin and hemoglobin are more transparent for IR than for visible light. If IR light of aberrometer penetrates into choroid, eye will appear longer, hence more myopic, than for visible light. IR Visible

1D @ 543 & 780 nm Our measurements at Indiana (N=30) indicate significant bias (-0.

25 Myopic bias of infrared aberrometry (Warren, 2006) Previous reports indicate fundus reflection is near cone apertures: Williams et al. (1994) 633nm Lopez-Gil & Artal (1997) < 543 & 780 nm Our measurements at Indiana (N=30) indicate significant bias 850nm, suggesting the retinal beacon is near Bruch's membrane. RPE Bruch's memb. Cone apertures 0.25 ±0.16 D = 92 ±59 µm 850nm retinal beacon

26 Even more unanswered questions: Computational Is the dynamic range of the metrics large enough for Rx? Are the metrics robust enough for iterative solutions to converge to global maximum IQ? Is wavefront reconstructiona and Fourier optics calculations necessary? Or can we go straight from measured wavefront slopes to the prescription? Is compensation for individual variation necessary? Ocular chromatic aberration models used for polychromatic analysis Wavelength-in-focus for polychromatic objects Measurement bias due to depth-of-penetration of fundus beacon

27 Even more unanswered questions: Peripheral refractions Do foveal methods of objective refraction work also for peripheral vision? What is the best way to deal with elliptical entrance pupils that occur off-axis? Clinical issues Might cyclopean metrics of visual quality prove useful for performing binocular wavefront refractions? Will wavefront refractions replace clinical trials for new designs of refractive therapy? Will objective wavefront refractions replace subjective refractions as the "gold standard" of clinical practice?

28 The Visual Optics Group at Indiana University Larry Thibos, PhD Arthur Bradley, PhD Steve Burns, PhD Ann Elsner, PhD Donald Miller, PhD Carolyn Begley, OD Rowan Candy, OD PhD Jacob Rubinstein, PhD Jayoung Nam, Phd Nikole Himebaugh, OD Charles Coe, OD Haixia Liu, MD Jingyun Wang, BS Pete Kollbaum, OD Sowmya Ravikumar, BS Toco Chui, MS Xin Wei, BS Danielle Warren, OD Weihua Gao, BS Kevin Haggerty, BS Benno Petrig, PhD Ravi Jonnal, MS Jungtae Rha, PhD Yan Zhang, PhD Barry Cense, PhD Zhangyi Zhong, BS Jie Shen, BS Hongxin Song, BS Support: National Institutes of Health / NEI National Science Foundation Borish Center for Ophthalmic Research Vision Research at

Accuracy and Precision of Objective Refraction from Wavefront Aberrations

$Accuracy and Precision of Objective Refraction from Wavefront Aberrations$ Accuracy and Precision of Objective Refraction from Wavefront Aberrations Larry N. Thibos Arthur Bradley Raymond A. Applegate School of Optometry, Indiana University, Bloomington, IN, USA School of Optometry,