Fitting Accuracy of an
Arc Step-Based Contact Lens Module

BY KENNETH A. LEBOW, O.D., F.A.A.O
NOV. 1997

Computer-assisted, videokeratoscopic RGP contact lens fitting modules are intended to improve the accuracy of initial contact lens selection, simulate on-eye fluorescein patterns and reduce overall fitting time. The ideal software program is one that can determine a contact lens prescription that's based on a predetermined base curve-to-cornea fitting relationship and has an appropriate power, lens diameter and peripheral curve system, while being performed by a technician for verification by the practitioner.

Can a computer do what you do? This study investigates the accuracy of computerized contact lens fitting using the Humphrey Atlas Corneal Topography System.

Humphrey Instruments' MasterVue corneal topography software (Revision A6) includes the MasterFit contact lens fitting module, which uses an arc-step algorithm for calculating corneal elevation data and generating tear thickness profiles. While contact lens designs based on axial topography data produce accurate fitting results, tangential or instantaneous radius of curvature readings can provide more accurate data regarding the location of the corneal apex and may be even more accurate in predicting appropriate contact lens fitting parameters. Topography systems that utilize advanced arc-step algorithms may be even more accurate in locating an exact corneal position because they acquire radius of curvature data by measuring the distance between each of the Placido rings as an arc triangulated on the measured data using the points in an iterative analysis. Using the precise location of the corneal surface when generating tear film thickness profiles is critical to the accuracy of base curve selection and subsequent lens power calculations.

The following is a retrospective evaluation of the clinical efficacy and accuracy of base curve and lens power selection for MasterFit compared to my actual fitting preference.

METHODS

The Humphrey Atlas corneal topographer was used to map the eyes of 47 previously adapted, daily wear RGP patients (13 men, 34 women) who participated in an extended wear RGP clinical investigation. Their refractive errors ranged from -8.50D to +5.50D sph and -0.50D to -2.50D cyl. Other baseline findings recorded were keratometry, post-contact lens refraction, slit lamp evaluation of the anterior segment and on-eye fluorescein patterns with the investigative lens design.

Some patients were fit using uniform corneal alignment pressure gradients. For moderate to marked corneal astigmatism, I accepted slight astigmatic corneal touch patterns rather than fitting back toric designs to achieve uniform pressure gradients. The final lens design was based solely upon previous contact lens parameters, on-eye fluorescein evaluation, lens position, movement and subjective visual acuity. Patients wore these lenses during the one-year extended wear clinical trial.

Subsequently, I retrospectively compared MasterFit contact lens parameters to these clinically determined values for accuracy and consistency in base curve and lens power selection. After entering each patient's vertex corrected manifest spectacle correction, MasterFit generated a proposed contact lens design. I evaluated the simulated fluorescein patterns, but made no adjustments to the suggested computer-generated parameters to optimize the appearance of the simulated patterns.

MasterFit software offers a customized base curve selection based on the practitioner's fitting philosophy and the manufacturer's suggested guidelines. Since no predetermined manufacturer's fitting nomogram was available for this investigational lens, the software calculated base curve and power data according to an alignment fitting philosophy. Generally, this resulted in the computer selecting a contact lens base curve based on the flat K reading when the corneal astigmatism was less than 0.75D, and as much as 1.00D steeper than the flat K reading when the corneal astigmatism exceeded 2.50D.

RESULTS

Comparing the base curve selections determined by MasterFit software with those from the on-eye clinical evaluation shows that the average MasterFit base curve selection did not significantly differ from those determined clinically (Table 1). Statistically, the description of both groups is virtually identical for base curve, correlation coefficient 0.961 (Table 2) and lens power distributions, correlation coefficient 0.985 (Table 3).

The frequency distribution of MasterFit base curves selected shows a normal distribution, and within �0.25D (0.05mm) of the actual clinical base curve selected for a patient, MasterFit accurately predicted the proper base curve 72 percent of the time (Fig. 1). When the range is increased to �0.50D (0.10mm), which is generally considered clinically significant in determining a base curve change based upon fluorescein observation, the accuracy increased to 90 percent. Since computer-predicted lens powers are determined directly from the base curve selection, incorrect base curve parameters will adversely influence power determinations. MasterFit lens power calculation averaged 0.14D less minus than that determined clinically, a clinically insignificant difference. Within a �0.50D range of lens power, MasterFit was 88 percent accurate, which is not only clinically significant, but also correlates well with its 90 percent base curve accuracy (Fig. 2).

Simulated fluorescein patterns generated by the MasterFit software accurately predicted and generally reproduced the on-eye fluorescein pattern. Figure 3 represents a simulated fluorescein pattern demonstrating overall lens alignment with slight 3 and 9 o'clock midperipheral bearing. Figure 4 shows the on-eye fluorescein pattern. While all of the simulated fluorescein patterns demonstrated clinically acceptable results (either aligned, flat, steep or with-the-rule astigmatism), 71 out of 94 eyes exactly matched the on-eye fluorescein patterns.

There were 12 eyes that MasterFit predicted as either with-the-rule bands or apical touch but were all clinically observed as aligned, and of 11 eyes predicted for alignment, nine were actually steep, one was flat, and one was a with-the-rule band. The base curves the software predicted for these 23 eyes were only 0.01 � 0.07mm flatter than those clinically observed. Only two eyes showed flatter-than-expected actual fluorescein patterns, while the other 21 eyes were all slightly steeper than MasterFit had predicted. Figure 5 shows a simulated fluorescein pattern demonstrating apical touch in spite of the overall lens alignment visible in the actual fluorescein pattern for this patient (Fig. 6).

EFFICIENT CONTACT LENS FITTING

Placido-based corneal topography systems, especially those using sophisticated arc-step iterative algorithms, are capable of a high degree of accuracy in predicting base curve and lens power parameters from corneal topography data. Contact lenses designed from these parameters clinically align well to the corneal surface and provide practitioners with an efficient method of fitting RGP materials.

When variations occurred between clinical and computer-selected contact lens parameters, the range of parameters determined by the MasterFit software appeared to be slightly flatter than and less minus than those determined clinically. This correlates well with the slightly flatter range of base curve selections offered by the MasterFit software. CLS

References are available upon request to the editors at Contact lens Spectrum. To receive them via fax, call (800) 239-4684 and request document #30.

Dr. Lebow is in private practice in Virginia Beach, Va., specializing in contact lenses and clinical research. He is a member of the AOA, a past chair of the AOA contact lens section and is also a consultant for Humphrey Instruments, Inc.

SPECTRUM NOV. 1997