Toric intraocular lenses (IOLs) are becoming more commonly available, allowing more predictable, precise, and stable correction of astigmatism than corneal or limbal relaxing incisions.1,2 Optimum astigmatic correction requires precise IOL axis alignment with the meridian of corneal astigmatism. In addition to visual acuity, refraction, and keratometry, studies tend to assess IOL rotation subjectively3 using a slitlamp biomicroscope eyepiece graticule4 or slit-beam protractor,5 although the method is often not specified.6 However, these subjective techniques rely on the patient maintaining a stable and vertical head position at each assessment and they estimate rotation to approximately the nearest 1 to 5 degrees only.
Digital imaging has been applied to toric IOL rotation assessment. Original studies used generic7,8 or custom image-analysis software9 to assess the rotation of a line drawn to join features on the IOL. However, this axis was compared with the image horizontal plane, ignoring the effect of head or eye rotation between assessments.
Viestenz et al.3 evaluated the eye's rotational stability during photography over at least 6 months using fundus image analysis. They found that it changed, on average, by 2.5 degrees between visits, although the change was as high as 11.5 degrees, being greater in women, older patients, and those with worse visual acuity or higher astigmatism. The authors note that the deviation in the measured orientation of the eye between visits resulted from a combination of cyclotorsion, head rotation, and autorotation during fixation of the positioning light. The study also estimated relatively large errors from the mounting of the camera and framing and projection of slides, which is less of an issue with cameras integrated into slitlamps. The latter usually have an external light source as well as the slit beam to allow illumination of the iris and bulbar conjunctiva at the same time as the retroillumination. Viestenz et al. recommend a digital overlay technique that uses conjunctival vessels, Axenfeld loops, or iris structure as references to account for these intrinsic rotations. Weinand et al.10 used this technique in 17 of 40 eyes immediately and 6 months after implantation of a single-piece acrylic IOL. The other images could not be analyzed due to insufficient dilation (IOL orientation required visibility of both haptic–optic junctions) and poor image quality. In addition, a different camera was used on each occasion and repeatability of analysis and image capture was not assessed. Patel et al.11 also compensated for head and eye rotation by rotating the retroilluminated image to align preoperatively made corneal ink markings on a surgical video frame with the 6 o'clock position. This technique had an intraobserver variability of 2.3 to 3.1 degrees. More recently, Shah et al.12 calculated the center of the IOL as the center of a rectangle with the toric IOL marks as the opposite corners. They overlaid a radial grid on the center of the IOL to assess the axis of a line joining the toric marks to 0.1-degree precision. The axis of a line joining the center of the IOL to a single prominent episcleral vessel was used to compensate for eye and head rotation. However, this complex method is susceptible to error if the IOL changes centration.
Optimum alignment is a major issue when toric correction or compensation for ocular aberrations are to be incorporated into the IOL optic.13 Intraocular lens centration has been assessed by image analysis in which an oval is fitted to the IOL optic margin and the limbus and the centers are compared.8,14,15 However, the repeatability of analysis and image capture has not been assessed and although image quality is considered an important factor, the effect of this poor image quality has not been determined.
This study, therefore, examined the repeatability of objective analysis of IOL rotation and centration and the effect of image quality.
PATIENTS AND METHODS
This study assessed eyes that had unilateral implantation of an Akreos AO aspheric IOL (Bausch & Lomb) with orientation marks. Surgery was performed at 1 of 6 hospital sites across Europe. All patients provided informed consent before IOL implantation, and the ethical committee at each site approved the study.
The eyes were dilated with phenylephrine 2.5% and tropicamide 1.0%. The IOL was imaged at ×10 magnification in retroillumination using a CSO SL-990 digital slitlamp biomicroscope (Costruzione Strumenti Oftalmici). This was repeated 1 to 2, 7 to 14, 30 to 60, and 120 to 180 days after IOL implantation.
The axis of rotation of the IOL was determined by drawing a line to join the IOL orientation marks. This was normalized for rotation of the eye in front of the slitlamp between visits by comparing the axis of a line joining 2 consistent conjunctival vessels or iris features on opposite sides of the pupil margin (Figure 1). The reference markers had to be visible on the images captured at every follow-up visit. The center of ovals overlaid to circumscribe the IOL optic edge, pupil margin, and limbus were compared to determine IOL centration (Figure 1). A clinician masked to the IOL rotation and centration subjectively graded the images. The clinician was familiar with digital slitlamp imaging but did not take part in the image capture. The following were rated: iris feature quality, including illumination consideration (0 = poor/ungradeable; 1 = moderate; 2 = good; 3 = excellent); scleral blood vessel clarity, including underillumination (0 = poor/ungradeable; 1 = moderate/partially obscured markings; 2 = good; 3 = excellent); and marking clarity of the toric IOL, including under illumination and with dilation (0 = at least 1 not visible; 1 = indistinct; 2 = clear; 3 = sharp).
Two images of each eye were captured immediately after surgery in a subgroup of patients and analyzed to assess intrasession repeatability of the technique. Images from 2 patients were analyzed 10 times to assess the repeatability of the analysis.
Subjectively rated image-quality elements and their relationship to apparent IOL rotation were compared with the Spearman rank correlation and between visits and sites, with the Friedman chi-square test. Head rotation, assessed by 1 set or the mean of 2 sets of blood vessel or iris features on either side of the pupil, was compared with a t test. Standard deviations were calculated to assess the intrasession and repeated analysis variability in IOL rotation and centration.
One hundred seven eyes were evaluated. The subgroup analysis of intrasession repeatability included 40 eyes.
Table 1 shows the mean subjectively rated image quality. Overall, subjectively rated iris and blood vessel clarity were strongly correlated (r = 0.487, P<.001); these assessments were moderately related to the clarity of the IOL axis marks (iris, r = 0.237; vessel clarity, r = 0.184; P<.001). The clarity of blood vessels was rated best, generally increasing between visits (Friedman chi-square = 14.782, P = .002). Iris features were rated as least clear, with a countertendency to decrease in clarity between visits (chi square = 7.349, P = .062) along with the IOL axis marks (chi square = 10.811, P = .013;). The sites significantly differed in their ability to capture clear images of the blood vessels (chi square = 29.148, P<.001), iris features (chi square = 28.611, P<.001), and toric IOL marks (chi square = 19.677, P = .001).
Due to the relationship between blood vessel and iris feature clarity and the reliance on only 1 of the features to assess head rotation, the maximum score of these 2 ratings was taken. Inability to detect either anterior eye feature or the IOL toric marks resulted in an image that could not be graded; therefore, the 2 were multiplied and divided by the maximum value of 9 to give the percentage quality. The mean image quality between 1 to 2 days and 120 to 160 days after IOL implantation was significantly correlated with absolute apparent toric IOL rotation when compensated for head movements (r = −0.449, P<.001I (Figure 2).
Head rotation assessed by the rotation of blood vessel or iris features on either side of the pupil between visits was similar when 1 set (mean 2.23 degrees ± 1.84 [SD]; 10% >5 degrees rotation) rather than the mean of 2 sets (mean 2.03 ± 1.66 degrees; 7% >5 degrees rotation) of markers was used (P = .126). The measured changes in IOL centration were similar, whether compared to the pupil center or to the limbal center, for 7 to 14 days (mean 0.24 ± 0.18 mm versus 0.20 ± 0.15 mm; P = .152), 30 to 60 days (mean 0.24 ± 0.21 mm versus 0.27 ± 0.22 mm; P = .147), and 120 to 180 days (0.22 ± 0.17 mm versus 0.24 ± 0.19 mm; P = .370) compared with 1 to 2 days after surgery. However, the absolute difference in anatomic center between the pupil and the limbus varied greatly, being significantly larger in the vertical meridian (mean 1.89 ± 1.82 mm) than in the horizontal meridian (mean 0.18 ± 0.19 mm; P<.001).
The intrasession SD was ±0.79 degrees for rotation, ±0.10 mm for horizontal centration, and ±0.10 mm for vertical centration. The SD of repeated analysis of the same image IOL was ±0.70 degrees, ±0.20 mm, and ±0.31 mm, respectively.
This study examined the repeatability of objective analysis of IOL rotation and centration and the effect of image quality. As expected, iris feature and blood vessel clarity were related. Although there was evidence of an improvement in imaging with time for blood vessel clarity, the reverse was found for iris features, perhaps due to the limited depth-of-focus of imaging systems. Some digital systems have an aperture control that can be reduced to enlarge the depth of focus, thus allowing simultaneous imaging of the iris, conjunctiva, and IOL toric markings. However, the illumination has to be increased to compensate for the reduced aperture, which can cause patient discomfort, or the sensor gain's increased, which can cause a grainy image. The decrease in visibility of the toric IOL markings with time may also be related to fibrosis of the anterior capsule when the toric marking fell outside the capsulorhexis border. The clinical sites significantly differed in their ability to capture clear images of the IOL and anterior eye features, and no site had consistently high performance; these findings emphasize the need for imaging training and support. Despite the low rotation of the Akreos AO IOL platform, apparent image rotation increased with poorer rated image quality. This confirms the importance of high image quality for objective analysis of IOL rotation and supports the image quality metric devised.
The objective methodology had a repeatability of less than 1 degree in the assessment of the IOL rotation. Head rotation between measures was, on average, approximately 2 degrees, significantly reducing the variability in the measured IOL rotation when taken into account. This finding was consistent with a study of IOL rotation using fundus photography,3 which found a mean rotation of 2.3 ± 1.7 degrees in 400 eyes. Using the pupil or limbal center to calculate changes in IOL centration gave an equivalent result, with repeatability of approximately 0.1 mm, an order of magnitude better than subjective estimation. The pupil is not anatomically central to the limbus, particularly in the vertical meridian and the center may vary with dilation. This could cause further variability in subjective estimation of IOL centration unless the reference anatomical feature is clearly defined.
In conclusion, objective analysis of digital retroillumination images at different postoperative periods allowed sensitive assessment of the stability of IOL rotation and centration. Eye rotation between images can lead to significant errors if not taken into account. The quality of the images also significantly affects the accuracy of objective assessment. The aspheric IOL with orientation marks was stable in the eye 3 to 6 months after implantation.
1. Sun X-Y, Vicary D, Montgomery P, Griffiths M. Toric intraocular lenses for correcting astigmatism in 130 eyes. Ophthalmology. 2000;107:1776-1781. discussion by RM Kershner, 1781–1782.
2. Mendicute J, Irigoyen C, Ruiz M, Illarramendi I, Ferrer-Blasco T, Montés-Micó R. Toric intraocular lens versus opposite clear corneal incisions to correct astigmatism in eyes having cataract surgery. J Cataract Refract Surg. 2009;35:451-458.
3. Viestenz A, Seitz B, Langenbucher A. Evaluating the eye's rotational stability during standard photography; effect on determining the axial orientation of toric intraocular lenses. J Cataract Refract Surg. 2005;31:557-561.
4. Ruhswurm I, Scholz U, Zehetmayer M, Hanselmayer G, Vass C, Skorpik C. Astigmatism correction with a foldable toric intraocular lens in cataract patients. J Cataract Refract Surg. 2000;26:1022-1027.
5. De Silva DJ, Ramkissoon YD, Bloom PA. Evaluation of a toric intraocular lens with a Z-haptic. J Cataract Refract Surg. 2006;32:1492-1498.
6. Chang DF. Early rotational stability of the longer Staar toric intraocular lens; fifty consecutive cases. J Cataract Refract Surg. 2003;29:935-940.
7. Nguyen TM, Miller KM. Digital overlay technique for documenting toric intraocular lens axis orientation. J Cataract Refract Surg. 2000;26:1496-1504.
8. Becker KA, Auffarth GU, Völcker HE. Messmethode zur Bestimmung der Rotation und der Dezentrierung von Intraokularlinsen. [Evaluation of rotation and decentration of intraocular lenses]. Ophthalmologe. 2004;101:600-603.
9. Bender L, Spalton DJ, Uyanonvara B, Boyce J, Heatley C, Jose R, Khan J. POCOman: new system for quantifying posterior capsule opacification. J Cataract Refract Surg. 2004;30:2058-2063.
10. Weinand F, Jung A, Stein A, Pfützner A, Becker R, Pavlovic S. Rotational stability of a single-piece hydrophobic acrylic intraocular lens: new method for high-precision rotation control. J Cataract Refract Surg. 2007;33:800-803.
11. Patel CK, Ormonde S, Rosen PH, Bron AJ. Postoperative intraocular lens rotation; a randomized comparison of plate and loop haptic implants. Ophthalmology. 1999;106:2190-2195. discussion by DJ Apple, 2196.
12. Shah GD, Praveen MR, Vasavada AR, Rampal NV, Vasavada VA, Asnani PK, Pandita D. Software-based assessment of postoperative rotation of toric intraocular lens. J Cataract Refract Surg. 2009;35:413-418.
13. Tassignon M-J. 2007. Technology and needs for tomorrow's treatment of cataract. In: Manns F, Soederberg PG, Ho A, Stuck BE, Belkin M, editors., Ophthalmic Technologies XVII. Proceedings SPIE 6426, SPIE, Bellingham WA, 64260E.
14. Becker KA, Holzer MP, Reuland AJ, Auffarth GU. Genauigkeit der Linsenstärkenberechnung und Zentrierung einer asphärischen Intraokularlinse. [Accuracy of lens power calculation and centration of an aspheric intraocular lens]. Ophthalmologe. 2006;103:873-876.
15. Pérez-Torregrosa VT, Menezo JL, Harto MA, Maldonado MJ, Cisneros A. Digital system measurement of decentration of Worst-Fechner iris claw myopia intraocular lens. J Refract Surg. 1995;11:26-30.