mEYEstro software: an automatic tool for standardized refractive surgery outcomes reporting

mEYEstro is programmed and compiled in MATLAB R2023a (MathWorks Inc., Natick, MA, USA) using the MATLAB runtime compiler (MathWorks Inc.). mEYEstro is therefore an executable file (*.exe) that can be run as an independent Desktop application. mEYEstro requires the MATLAB runtime compiler (MRC) to be correctly installed on the computer. The MRC installs automatically with the mEYEstro install. mEYEstro has been tested on Windows 10 Home and Professional, with a 64-bit-operating system and both with 1920 × 1080 and 3840 × 2160 screen resolutions. mEYEstro and the demonstration trial datasets are available to download from https://www.lasikmd.com/media/meyestro. A tutorial video is available at this link (https://www.youtube.com/watch?v=NFlRRHx6ZaI) and a tutorial guideline in Supplementary File C.

mEYEstro can be used to automate producing all of the standard refractive surgery graphs, as recommended by various ophthalmology journals [10,11,12,13,14]. The tool was developed specifically for academic research and teaching purposes but can also be used by surgeons looking to understand and improve their clinical outcomes. mEYEstro can be used to examine the visual and refractive outcomes of any corneal or intraocular refractive procedure. The corneal procedures include LASIK, PRK, and SMILE as well as collagen crosslinking, incisional keratotomy, intracorneal rings segments, LASEK, etc. The lens-based procedures include cataract surgery, refractive lens exchange, phakic IOL, etc. mEYEstro can also be utilized to study outcomes of procedures used to treat the various refractive surgery complications that exist today, or any other surgical procedure involving the eye [15,16,17,18,19,20,21,22,23,24]. The use of mEYEstro is completely free provided that the user cites the current manuscript when using mEYEstro results in publications, presentations, or other public communications.

To automatically generate the figures, mEYEstro reads data files in Microsoft Excel format (e.g., Datafile.xlsx). Excel was used due to its widespread use and simplicity. There are 20 columns, including 15 that are mandatory for proper mEYEstro functioning. The first five columns are 1) preoperative refraction sphere, 2) preoperative refraction cylinder, 3) preoperative refraction axis, 4) preoperative refraction vertex distance, and 5) preoperative corrected distance visual acuity (CDVA). The next four columns are 6) intended postoperative refraction sphere target, 7) intended postoperative refraction cylinder target, 8) intended postoperative refraction axis target, and 9) intended postoperative refraction vertex distance. If the intended postoperative refraction is plano, columns 6, 7 and 8 should be reported as 0, 0, and 0, respectively. The six next columns are 10) postoperative refraction sphere, 11) postoperative refraction cylinder, 12) postoperative refraction axis, 13) postoperative refraction vertex distance, 14) postoperative CDVA and 15) postoperative uncorrected distance visual acuity (UDVA). The last five columns (columns 16 to 20) are optional and allow the user to report the postoperative spherical equivalent (SEQ) at up to five different time points to generate a standard stability graph. Refraction data must be provided in the point decimal format (e.g., -1.50, 0.75) and using the negative cylinder (-ve) nomenclature. The negative cylinder notation was chosen since it is by far the most widespread notation used among refractive surgeons. For calculation mEYEstro will automatically convert the negative cylinder (-ve) to positive notation (+ ve). The UDVA and CDVA data must be provided as the 20/XX Snellen denominator (e.g., 20–1, 15, 20 + 2, 25, 30–1). An example of a representative mEYEstro data file is presented in Supplementary File A. For users that use LogMAR notation in their charting, a LogMAR to Snellen denominator automatic conversion table is included in Supplementary File B. This automatic conversion table can be used as needed to make automatic conversion of LogMAR values to 20/XX Snellen denominator values. The converted value can simply be pasted in a mEYEstro data file. Users can report their refraction data at any vertex distance (12 mm, 10 mm, 0 mm, etc.). mEYEstro will automatically convert the refractive astigmatism, generally measured at a vertex distance of 12 mm, to the corneal plane (0 mm). Data exclusion is at the user’s discretion prior to data importation. Upon entering your data in the mEYEstro datafile, if Excel is automatically converting Snellen denominator like 25–2 to a date “25-Feb”, please instead type ‘25–2 or set the column number format to “Text”.

The mEYEstro software adheres to terminology, calculations, and graphical representations originally described by Waring and Reinstein, as well as Editorials by Reinstein et al. [1,2,3, 5, 6, 9]. All vectorial analyses adhere to terminology, calculations, and graphical representations originally described by Alpins [10, 11, 13, 14]. The efficacy index is calculated as the ratio of postoperative UDVA (converted to decimal format) to the mean preoperative CDVA (converted to decimal format). The safety index is the ratio of postoperative CDVA (converted to decimal format) to mean preoperative CDVA (converted to decimal format). The SEQ was calculated by adding the sum of the sphere power with half of the cylinder power. The defocus equivalent was calculated as the absolute value of the SEQ plus half the absolute value of the cylinder. Negative cylinder (-ve) to positive notation (+ ve) conversion and vertex distance conversions of refraction data adhere to the methodology described by Alpins [10]. Statistical methodologies are presented in the statistical analyses reporting section of the current paper. For additional methodological details, please contact the corresponding author.

The flow chart of the mEYEstro workflow is shown in Fig. 1. mEYEstro is entirely controlled via a few simple steps and each triggered as the user progresses through the program workflow (Fig. 1). Upon starting the application, the user must choose the type of refractive surgery procedures (LVC, RLE, ICL, CAT) (Fig. 1A), the study design (single group, unpaired groups, paired groups) (Fig. 1B), the name of the group(s), the color of the graphs, and the analysis parameters (Snellen lines to display on the UDVA/CDVA graphs, LogMAR threshold for each Snellen optotypes, efficacy & safety index levels, etc.) (Fig. 1C). If the user wants to include a stability graph, additional choices are presented (number of time points, selection of time points to compare, etc.). Finally, the user is invited to select the Excel data file for each group (Fig. 1D). The selected graphs are then generated and automatically saved in a folder as high resolution 400 dpi TIFF images (Fig. 1E). These individual images are ideal for PowerPoint presentations and scientific articles. In addition to the 400 dpi TIFF images, the one-page figure with all 10 standard graphs (Fig. 1F) is exported as an ultra-high definition 1200 dpi TIFF image for journals with higher image quality criteria, such as the Journal of Cataract & Refractive Surgery.

Flow chart of the mEYEstro workflow. A-B Upon starting the application, the user is invited to select (A) the type of refractive surgery and (B) the type of statistical grouping. C The user is then invited to enter the group(s) name(s), select the colors of the graphs, and parameters. D The user is next invited to select the data file. E The figures are automatically displayed and saved as high-resolution TIFF images in a folder at the same location as the original Excel data file. This folder automatically opens once the graphs are saved. F The user can visualize the generated standard figures

Efficacy analyses include the preoperative and postoperative cumulative Snellen uncorrected (UDVA) and corrected visual acuity (CDVA) graph (Panel A in Figs. 2, 3 snd 4) and the difference between postop UDVA and preop CDVA graph (Panel B in Figs. 2, 3 and 4). These two graphs allow the user to visualize and report standard visual outcomes. Panel A also includes the average (± standard deviations) of the preoperative and postoperative UDVA and CDVA in LogMAR values. The Panel B graph also reports the average efficacy index. The number of eyes per group is also displayed in Panel B. If two groups are analyzed, the p-value and effect size between groups is also displayed. For cataract surgery, the postoperative UDVA is compared to postoperative CDVA instead of preoperative CDVA, in agreement with current journal standards.

Graphs that are automatically generated by mEYEstro from the provided Trial 1 dataset. The first simulated trial dataset (Trial 1) includes two Excel files (Group A and Group B) and investigates the outcomes of a laser vision correction contralateral eye study comparing two excimer lasers in hyperopic eyes with astigmatism. To generate this standard figure, please use the provided tutorial guideline in Supplementary File C. A tutorial video is available at this link (https://www.youtube.com/watch?v=NFlRRHx6ZaI)

Graphs that are automatically generated by mEYEstro from the provided Trial 2 dataset. The second dataset (Trial 2) is comprised of simulated data from a single group in order to investigate the outcomes of a toric Phakic IOL (PIOL) in hyperopic eyes with moderate to high astigmatism. To generate this standard figure, please use the provided tutorial guideline in Supplementary File C. A tutorial video is available at this link (https://www.youtube.com/watch?v=NFlRRHx6ZaI)

Graphs that are automatically generated by mEYEstro from the provided Trial 3 dataset. The third simulated dataset (Trial 3) includes two files (Group A and Group B), used to investigate the outcomes of two cataract surgery groups, comparing two biometers, in myopic-astigmatism eyes. To generate this standard figure, please use the provided tutorial guideline in Supplementary File C. A tutorial video is available at this link (https://www.youtube.com/watch?v=NFlRRHx6ZaI)

Safety analyses include the change in lines of CDVA (Panel C in Figs. 2, 3 and 4). This graph allows the user to visualize procedures safety in terms of corrected visual acuity line gain and line loss from preop to postop. The graph also reports the average safety index. The number of eyes per group is also displayed in Panel C. If two groups are analyzed, the p-value and effect size between groups is also displayed.

SEQ accuracy analyses include the accuracy of SEQ to intended target histogram (Panel D in Figs. 2, 3 and 4) and the achieved SEQ vs attempted SEQ scattergram (Panel E in Figs. 2, 3 and 4). These two graphs allow the user to visualize accuracy outcomes. Panel D displays the percentage of eyes within 0.25, 0.50, 0.75 and 1.00 D of intended target, as well as the average postop SEQ to intended target. The number of eyes per group is also displayed in Panel D. If two groups are analyzed, the p-value and the effect size between groups is also displayed. Panel E also displays the linear regression equation, the R2, the average attempted SEQ, and the range of attempted SEQ.

For longitudinal studies where stability over time is an important part of the analyses, a standard SEQ stability graph is required. This graph shows the preoperative SEQ and the postoperative SEQ at up to 5 time points for each group. For example, the user can provide the 1, 3, 6, 12, and 24 months SEQ data in the 5 last columns of the data file and mEYEstro will generate the graph (Panel F in Fig. 2) from the provided data, automatically calculating the mean SEQ at each time point. mEYEstro will also calculate the percentage of eyes with a SEQ change greater than ± 0.50 D between two selected time points. For example, between the 3 months and 24 months postop. Since not all research questions require a longitudinal analysis, the stability analyses are optional and the user may leave the last five columns in the data file blank. The number of eyes at each time point and each group is also displayed at the bottom of Panel F.

When the user does not include a stability graph, a DEQ accuracy graph will automatically be included instead. Defocus equivalent (DEQ) accuracy analyses include the postoperative DEQ histogram (Panel F in Figs. 3 and 4). By default, this graph shows the percentage of eyes with a DEQ within 0.25, 0.50, 0.75, 1.00 and 2.00 D. The average postoperative DEQ and the number of eyes per group are also displayed in Panel F. If two groups are analyzed, the p-value and the effect size between groups is also displayed.

Standard astigmatism analyses include the postoperative refractive astigmatism graph (Panel G in Figs. 2, 3 and 4), the target-induced astigmatism (TIA) vector vs surgically-induced astigmatism (SIA) vector scattergram (Panel H in Figs. 2, 3 and 4), the Correction Index histogram (Panel I in Figs. 2, 3 and 4), and the Angle of Error histogram (Panel J in Figs. 2, 3 and 4). Panel G display the percentage of eyes within 0.50, 0.75 and 1.00 D of plano postoperative astigmatism, as well as the average postop refractive astigmatism. Panel H also displays the linear regression equation, the R2, the average TIA and SIA, and the range of attempted SEQ. The number of eyes per group is also displayed in Panels G, H, I and J. If two groups are analyzed, the p-value and the effect size between groups is also displayed. For more detailed formulas and calculations, the interested reader can consult previous literature [10,11,12,13,14]. For advanced standard vector graphs, we have described and provided AstigMATIC tool, available at www.lasikmd.com/media/astigmatic.

When comparing two groups, the mEYEstro software automatically selects and uses the appropriate statistical hypothesis tests. The Kolmogorov–Smirnov test is first used to test if the preoperative and postoperative variables are normally distributed. Unpaired sample T-tests and non-parametric Mann Whitney U-tests are then used where applicable to compared outcomes between two independent groups. Paired samples T-tests or non-parametric Wilcoxon signed-rank test tests are used where applicable to compare two paired groups. Statistical significance is set at p