Could the coefficient of variation (COV) of the corneal endothelium be overestimated when a centre-dot method is used?

Agreement of Corneal Endothelial Cell Analysis Between Konan-Noncon Robo SP-6000 and Tomey EM-3000 Specular Microscopes in Healthy Subjects

PURPOSE

To compare corneal endothelium parameters taken by two common noncontact specular microscopes in healthy subjects.

METHODS

Healthy participants visiting the outpatient eye clinic at the Rabin Medical Center, Petah Tikva, Israel, were recruited prospectively. All participants underwent three consecutive corneal endothelial cell photographs with both the Konan-Noncon Robo SP-6000 and the Tomey EM-3000 specular microscopes. Endothelial cell density (ECD) was evaluated using the manual center technique in both machines. Bland-Altman graphs were used to assess the agreement between the devices, and intraclass correlation coefficient (ICC) served to assess intraobserver variability for each device.

RESULTS

Recruited were 49 healthy subjects with a mean age of 48.9±15.6 years, 49 right eyes were included. The mean ECD was comparable between the Tomey EM-3000 and the Konan-Noncon Robo SP-6000 (2,713.2±242.4 vs. 2,700.8±300.5 cells/mm2, respectively, P=0.47) with a mean difference of 12.4 cells/mm2 (0.67%), a mean ECD absolute difference of 93.3 cells/mm2, and low 95% limits of agreement of -222.0 to +246.9 cells/mm2. A folded empirical distribution function curve showed that all differences fell within 525.4 cells/mm2, centered around a median of 13.3 cells/mm2. Intraclass correlation coefficient was high for both the Konan-Noncon Robo SP-6000 (0.93, 95% confidence interval [CI]: 0.89-0.95) and the Tomey EM-3000 (0.88, 95% CI: 0.82-0.93).

CONCLUSIONS

The difference in endothelial cell measurements between the Konan SP-6000 and the Tomey EM-3000 specular microscopes through the center and the L-count analyzing techniques, respectively, is clinically small and not statistically significant. Nevertheless, caution should be taken when used interchangeably because ECD difference between the two machines can be as high as 525.4 cells/mm2.

Endothelial morphometric measures to predict endothelial graft failure after penetrating keratoplasty

IMPORTANCE

Endothelial morphometric measures have potential value in predicting graft failure after penetrating keratoplasty.

OBJECTIVE

To determine whether preoperative and/or postoperative central morphometric measures (endothelial cell density [ECD], coefficient of variation [CV], and percentage of hexagonality [HEX]) and their postoperative changes are predictive of graft failure caused by endothelial decompensation after penetrating keratoplasty to treat a moderate-risk condition, principally Fuchs dystrophy or pseudophakic corneal edema.

DESIGN

In a subset of Cornea Donor Study participants with graft failure, a central reading center determined preoperative and postoperative ECD, CV, and HEX from available central endothelial specular images.

SETTING

Cornea Image Analysis Reading Center of the Specular Microscopy Ancillary Study.

PARTICIPANTS

Eighteen patients with graft failure due to endothelial decompensation and 54 individuals matched for most donor and recipient measures at baseline whose grafts did not fail.

MAIN OUTCOME MEASURE

Change in ECD, CV, and HEX values.

RESULTS

Preoperative ECD was not associated with graft failure (P = .43); however, a lower ECD at 6 months was predictive of subsequent failure (P = .004). Coefficient of variation at 6 months was not associated with graft failure in univariate (P = .91) or multivariate (P = .79) analyses. We found a suggestive trend of higher graft failure with lower HEX values at 6 months (P = .02) but not at the established statistical significance (P < .01). The most recent CV or HEX values, as time-dependent variables, were not associated with graft failure (P = .26 and P = .81, respectively). Endothelial cell density values decreased during follow-up, whereas CV and HEX appear to fluctuate without an apparent trend.

CONCLUSIONS AND RELEVANCE

Endothelial cell density at 6 months after penetrating keratoplasty is predictive of graft failure, whereas CV and HEX appear to fluctuate postoperatively, possibly indicating an unstable endothelial population in clear and failing grafts.

TRIAL REGISTRATION

clinicaltrials.gov Identifier: NCT00006411.

In vivo confocal microscopy of the corneal endothelium: comparison of three morphometry methods after corneal transplantation

PURPOSE

The purpose of this study was to assess the endothelium of corneal grafts by in vivo confocal microscopy (IVCM), and to evaluate an automated endothelial software system in comparison with a manual cell count and planimetry.

PATIENTS AND METHODS

Overall, 40 corneal grafts (20 deep anterior lamellar keratoplasties (DALKs) and 20 penetrating keratoplasties (PKs)) were assessed by scanning-slit IVCM. The endothelial cell density (ECD) was estimated with the automated and the manual cell count method of the instrument's Nidek Advanced Vision Information System (NAVIS) software. The results were compared with planimetry as the reference method, and the agreement was assessed.

RESULTS

The mean (±SD) automated ECD was 2278±524 cells/mm(2) (range 1167-3192 cells/mm(2)), whereas the manual cell count method gave significantly lower ECDs with a mean of 1213±677 cells/mm(2) (range 218-2440 cells/mm(2); P<0.001). The manual cell counts were also significantly lower than those by planimetry, with a mean ECD of 1617±813 cells/mm(2) (range 336-2941, P<0.001). Bland-Altman analyses indicated that the limits of agreement (LoA) between the automated and the planimetry method were -671 and +1992 cells/mm(2), whereas they were -1000 and +202 cells/mm(2) when comparing the manual cell counts with planimetry.

CONCLUSION

Following keratoplasty, the NAVIS automated method is likely to overestimate endothelial cell counts due to oversegmenting of the cell domains. Automated ECDs are substantially higher than those by the manual counting method or planimetry. The differences are considerably larger post-keratoplasty than for normal corneas, and the methods should not be used interchangeably.

Simulation of specular microscopy images of corneal endothelium, a tool for control of measurement errors

PURPOSE

We aimed at developing simulation software capable of producing images of corneal endothelium close to identical to images captured by clinical specular microscopy with defined morphometrical characteristics. It was further planned to demonstrate the usefulness of the simulator by analysing measurement errors associated with a trained operator using a commercially available semi-automatic algorithm for analysis of simulated images.

METHODS

Software was developed that allows creation of unique images of the corneal endothelium expressing morphology close to identical with that seen in images of corneal specular microscopy. Several hundred unique images of the corneal endothelium were generated with randomization, spanning a physiological range of endothelial cell density. As an example of the usefulness of the simulator for analysis of measurement errors in corneal specular microscopy, a total of 12 of all the images generated were randomly selected such that the endothelial cell density expressed was evenly distributed over the physiological range of endothelial cell density. The images were transferred to a personal computer. The imagenet-640 software was used to analyse endothelial cell size variation, percentage of hexagonal endothelial cells, and endothelial cell density.

RESULTS

The simulator developed allows randomized generation of corneal specular microscopy images with a preset expected average and variation of cell structure. Calculated morphometric information of each cell is stored in the simulator. The image quality can secondarily be varied with a toolbox of filters to approximate a large spectrum of clinically captured images. As an example of the use of the simulator, measurement errors associated with one trained operator using the imagenet-640 software, and focusing on endothelial cell density, were examined. The functional dependence between morphometric information estimated with the imagenet-640 software algorithm and real morphometric information as provided by the simulator was analysed with regression. It was demonstrated that that the estimations of endothelial cell size variation was associated with a scaling error and that the random error was strongly dependent on the operator.

CONCLUSION

The newly developed simulator for randomized generation of morphometrically defined corneal specular microscopy images for the first time makes it possible to estimate a spatial scaling error of an available semi-automatic algorithm and to determine the random measurement error of important morphometric estimates in a defined reference sample of images. It is anticipated that the simulator will be a valuable tool for the generation of a large set of morphometrically well-characterized corneal specular microscopy images that can be used for calibration among research centres, for minimization of random errors and for measurement of quality control. Simulated images will be useful for the development of fully automatic analysis of corneal endothelial cell morphometry.

A comparison of two methods for estimating polymegethism in cell areas of the human corneal endothelium

PURPOSE

To compare the outcome measures from two different methods for estimating the coefficient of variation (COV) in cell area for human corneal endothelia.

METHODS

A single non-contact specular micrograph was obtained from the central region of the corneal endothelium of 100 healthy non-contact lens-wearing white European subjects, aged from 32 to 62 years. The captured image file was either assessed using a machine-based algorithm in which 21 cells were marked and their areas reported (designated as the TOPCON 'centre-dot' method), or by an overlay and semi-automated computer-based image analysis system of the entire image of around 200 cells (designated as the AUTO method). The average cell area values were used to calculate the endothelial cell density (ECD), while the COV was calculated from the standard deviation of the cell area measures.

RESULTS

The TOPCON and AUTO methods yielded comparable data for average cell area (395 vs 391 microm(2)) and estimated ECD (2566 vs 2575 cells mm(-2)) that were not statistically different (p > or = 0.351), although there was a slight bias between the two methods. However, the mean COV values were very different at 43.5 and 31.4% (p < 0.001). The overestimation of the COV was related to the difference in the largest cell area domain identified by the two methods (p < 0.001, r = > or =0.748).

CONCLUSIONS

A centre-dot method using a small number of cells generates useful data on cell area and ECD, but it should be used cautiously for estimates of COV.