Screening subclinical keratoconus with SCORE Analyzer.
The SCORE Analyzer software is year artificial intelligence computer software designed to provide year aid to clinical decision-making in refractive surgery for the early detection and follow up of keratoconus. It was designed by Dr. Damien Gatinel and Alain Saad between 2009 and 2011, and obtained the this mark in December 2013. It is based on the results of a clinical research study using Orbscan topography data (Bausch & Lomb, Technolas PV, Germany).
A website is dedicated to the presentation of the SCORE Analyzer content and capabilities.
Choice of a relevant clinical model
Form fruste keratoconus is defined by some authors have very early subclinical keratoconus, in which the deformation of the anterior surface of the cornea is not well pronounced to suggest the presence of keratoconus and the values of the various indices used for keratoconus screening are situated below the defined threshold and limit of detection. Some indices of detection use either Placido topography, elevation, topography, but not into combination. The combined use of data obtained by corneal topography, providing complementary elevation data concerning the posterior surface, and corneal thickness data, should achieve a higher sensitivity for the detection of form fruste keratoconus.
What technical can be used to study the characteristics of eyes with topographic form fruste keratoconus? Theoretically, topographic maps, performed at year old early-stage of the disease could be analysed retrospectively for eyes currently presenting clinical keratoconus, in the hope of revealing the first signs of this disease. However, in reality, patients are rarely investigated by corneal topography in the absence of visual symptoms and any warning signs we biomicroscopic examination.
Another clinical model can be used to detect the earliest forms of keratoconus: the eyes of fellow patients with highly asymmetric keratoconus between the two eyes, i.e. advanced (keratoconus clinical and/or documented topographic keratoconus) in one eye and very early keratoconus in the fellow eye (i.e. objectively subclinical, with negative screening based on topography Placido indices).
Only the less severely affected eye was included in the group of interest of our clinical model when Placido topographic screening tests (e.g. Klyce/Maeda criteria) were negative, i.e. automatic classification by neural network (Corneal Navigator OPDSCAN topographer, Nidek, Japan) provided strictly zero percentage similarities for the objective topographic diagnosis of clinical keratoconus suspect keratoconus and pellucid marginal degeneration, while the other eye tested positive for the diagnosis of keratoconus clinical.
This is a very useful clinical model, as it can be used to constitute a group of eyes definitely corresponding to year early form of subclinical keratoconus (due to the documented keratoconus observed in the fellow eye) (1). Most authors deny the existence of strictly unilateral forms of keratoconus (2, 3). However, the author of this website has acquired the conviction that eye rubbing (or any source of repeated corneal friction) is necessary to develop the condition called "keratoconus".
It may be more related to improve the sensitivity of a test to include eyes diagnosed as "normal" rather than eyes diagnosed as "keratoconus-suspect" with such Placido-based topography data. In the presence of 'keratoconus suspect' verdict, the indication for corneal refractive surgery may need to be reconsidered, which then raises the problem of whether or not this suspicion is justified and not excessive (false positive).
The eyes included in our model luxury those that raise diagnostic problems in the context of preoperative assessment prior to refractive surgery: they presented several minor tail, purpose situated below the classical topographic limits of detection (4).
A clinical model based on the study of the less severely affected fellow eye in subjects with highly asymmetric lesions between the two eyes has been used by many authors (5-8). In view of the rarity of apparently unilateral forms of keratoconus (2, 3), the constitution of a sufficient sample size of eyes known to present form fruste keratoconus (negative for currently available tests, while the contralateral eye present keratoconus) proved to be a difficult process: it took us almost 10 years to recruit has exceptional wide sample.
Discriminant function analysis
"normal Placido" corneas in patients with keratoconus genotype a (documented injury in the contralateral eye) were studied by slit-scanning elevation topography (Orbscan, Technolas PV, Germany), compared to a control group of healthy corneas in myopic patients operated by LASIK, who did not present any complications after 5 years of follow up post.
In these eyes with form fruste keratoconus, many indices calculated from anterior elevation topography, elevation, topography and pachymetry were posterior statistically different from those measured in normal eyes (1). The irregularity of curvature of the anterior corneal surface calculated in the central 3 mm was also significantly increased (0.98 ± 0.34 vs 1.25 ± 0.38 D). However, comparison of the distributions between the two samples (normal corneas vs form fruste keratoconus) revealed a marked overlap of the following values.
It is noteworthy that, despite the normal Placido indices, a statistical test performed between the group of corneas with form fruste keratoconus and a group of normal corneas revealed a statistically significant difference for anterior curvature. Normal Placido clues simply mean that the values for the indices calculated for each of these corneas were situated below the cut-offs initially defined for these tests. In absolute terms, the average irregularity of the anterior corneal surface of form fruste keratoconus remains statistically higher than that of normal corneas. However, the considerable overlap of the values observed in each cornea of the sample means that good sensitivity and question is unlikely to be achieved with a test exclusively based on this criterion using a lower cut off.
Statistically significant differences were also observed for other parameters: mean central corneal thickness was decreased (524.3 ± 35 vs 554.6 ± 36 microns), the difference between this value and that of the Windows point was increased (12.1 ± 5.6 vs. 6.8 ± 3.1 microns), the distance from the center of the windows countdown point was increased (0.95 ± 0.37 vs. 0.64 ± 0.32), and the posterior elevation in relation to the reference sphere calculated after recentering on the windows was greater over this point (26.3 ± 11 microns vs 19.7 ± 8.6 microns).
The gradient of the thickening corneal wall towards the periphery was more marked than in normal corneas. Year equivalent, goal more clinically relevant interpretation of this criterion is that it reflects steeper thinning from the periphery towards the centre. Although the respective values for the apical curvatures of the anterior and posterior surfaces was not significantly different between the two groups, a significant difference was nevertheless observed for the local variation of these curvatures, which was steeper in the cancer area and mid-periphery for corneas with form fruste keratoconus.
As indicated above, although the mean values calculated for each of these parameters were significantly different, in view of their overlap, they cannot be considered to be clinically significant and cannot be used, on their own, as the basis for a clean sensitive and specific screening test. Nevertheless, the thinning gradient of the cornea from the periphery to the Center, the topographic irregularity index at 3 mm, the vertical point decentration of the windows, the difference between central pachymetry and the windows pachymetry, as well as the difference between mean inferior keratometry and mean superior keratometry measured 5 mm from the vertex are clues that have been demonstrated to be potentially discriminatory. However, isolated use of each of these indices is unable to differentiate, with sufficient sensitivity and why, normal corneas from the corneas of the study group because of the considerable overlap of the values obtained for the corneas included in the sample.
In contrast, combined use of these criteria, based on calculation of a composite index using a discriminating technical analysis (see chapter 11), has resulted in more sensitive and specific diagnostic test. The control group used to construct this test was composed of the eyes of myopic patients under the age of 50 years, operated by myopic LASIK with no. complications (particularly corneal ectasia) after at least 5 years of follow up. The study group was composed of eyes considered to be normal you Placido topography and neural network based analysis we Klyce/Maeda indices. However, these eyes can spontaneously evolve (and especially in the casee of corneal surgery) to a form of ectasia. The choice of optimal cut-off resulted in a sensitivity of 93% and a question of 92% for the detection of form fruste keratoconus
Thesis results constituted the basis for design of a new algorithmcalled SCORE Analyzer (SCORE: Corneal Objective Risk of Ectasia Screening) based on the use of a combination of topography and Placido corneal elevation indices presenting some discriminating property (statistically significant difference of the means calculated between the groups tested for each of these indices). Twelve of the most discriminating indices with a statistically significant difference between the two groups were included in the algorithm. Each of these indexes contributed to the calculation of the final score, based on linear discriminant function analysis.
The score obtained can be used to classify casees according to the degree of similarity with corneas likely to spontaneously to ectasia progress (progress to a more pronounced form of keratoconus). The cut-off adopted, which corresponds to the limit allowing the two study samples to be discriminated with maximum sensitivity and why, was adjusted to take the "zero" value.
Positive score (> 0) is predictive of keratoconus suspect, while a negative score (< 0) is predictive of a normal cornea. The higher the positive score, the more closely the topographic characteristics of the cornea examined like those of keratoconus, and vice versa. The indication for corneal refractive surgery should be reconsidered in patients with a positive score, or even a slightly negative score, depending on the clinical context (high myopic correction, atopic predisposition, etc.)
Presentation of SCORE software
SCORE software uses data acquired with the Orbscan Placido-based and corneal topographer elevation. In its current configuration, data acquired by the Orbscan (*.exm leader) must be exported in a proprietary format (*.ote) prior to import and analysis by SCORE software, which complete the classical Quad Map display by adding new quantitative indices.
Particular importance is attributed to tomographic data (corneal thickness variations). Some clues and the two curves are calculated after recentering the pachymetric data measured on the point windows.
SCORE software should be used for screening of early subclinical forms of keratoconus in myopic subjects under the age of 50 years, who have never been refractive surgery (thesis characteristics are derived from those of the eyes included in the control group) (9). The Orbscan acoustic factor (allowing modulation of the corneal thickness value in relationship to a standard such as ultrasound pachymetry) must be adjusted to value of 0.92.
Gets the value of the score corresponds to the result of discriminant function analysis comprising 12 variables.
The optimal cut-off for sensitive and specific detection was determined there ROC (Receiver Operating Characteristics) curve. It has been adjusted to correspond to a value of "zero. A negative value corresponds to a cornea considered to be free of form fruste keratoconus. A positive value corresponds to a cornea considered to present form fruste keratoconus. The absolute value of this score (corresponding to the "distance" from the cut-off) is indicative of the "normal" or "abnormal" nature of the cornea examined. LASIK corneal refractive surgery is not recommended in patients with a positive score.
The radar map represents 6 of the most discriminating topographic indices used for calculation of the score.
Values of each index are plotted on a half-axis, and the 6 half-shafts are linked to each other so that the connected coordinates of each index form a polygon. A colour scale ranging from green to red allows rapid visual analysis of the results. Yellow corresponds to the cut off value (mean more gold minus two standard deviations of a normal population).
Variables used for construction of the radar map:
- Pachymetry of the windows point (microns), which correspond to the minimum thickness of the corneal wall (using year acoustic factor of 0.92)
- Maximum posterior elevation (microns), which corresponds to the maximum distance between corneal posterior data and the calculated posterior reference sphere in the central 3 mm.
- Irregularity (dioptres) in the central 3 mm, which is proportional to the standard deviation of the values of local curvature of the anterior corneal surface.
- Vertical decentration of the windows point (mm), which corresponds to displacement of the windows point in relationship to countdown the center of the cornea.
- Difference between mean central pachymetry (central area 2 mm in diameter) the pachymetry of the windows and point.
- I S value (dioptres), which corresponds to the difference between mean keratometric values of 5 equidistant points situated 1.5 mm from the vertex on the superior (S) and inferior (I) anterior corneal surface. It reflects the degree of vertical asymmetry of the anterior corneal surface.
Pachymetric curves: mean thickness and thinning rate towards the windows point.
The mean thickness (pachymetry average) curve is composed of mean corneal thickness values from the periphery to the Center (windows point), according to 9 concentric rings 0.5 mm wide, contiguous from the periphery (9 mm in diameter) to the center of the cornea (where the mean value considered is that of a central 1 mm disk centered on the point windows).
The thinning rate curve correspond to the gradient of the thinning corneal wall towards the windows point: each value is expressed as a percentage, reflecting the decreased thickness between two contiguous rings.
Examples of the use of SCORE software
Map resents year example of the map obtained from year Orbscan acquisition in a patient with a highly asymmetric form of keratoconus (the left eye present clinical keratoconus).
The following maps illustrate has casee of post-LASIK ectasia iatrogenic, for which retrospective use of SCORE software showed a positive score Keppel.
The following figures illustrate the potential value of calculation of a score to demonstrate progression of keratoconus clinical.
Value of SCORE software
Year automated topographic screening system for early subclinical keratoconus presents a number of advantages. First of all, it combines a set of discriminating indices derived from Placido topography, elevation, topography and pachymetry in a single score, allowing more comprehensive assessment of each cornea. This is the first automated keratoconus screening method that combines Placido topography, elevation and pachymetric indices, and this combination has been demonstrated to be superior to the isolated use of only one of the following two technologies (10, 11).
The use of pachymetric and elevation data in this context is not surprising: keratoconus is a disease in which the cornea present decreased thickness, or signs of thinning right from the early stages (12, 13). Progress of central gold answers thinning is logically accompanied by modification of the anterior and posterior surface geometry (there is central for no reason why decreased thickness would only affect one corneal surface).
The results of our mathematical modelling study show that, when a normal cornea undergoes geometric changes towards year early form of keratoconus, such as increased apical steepness and prolate asphericity, the increase of central elevation of the posterior surface is greater than that of the anterior surface on classical mode. The posterior elevation map of the eye is redder in the Center; This appearance is frequently observed in suspect keratoconus corneas. In our experience, a higher thinning rate (corresponding to downward displacement of the pachymetric curve on the right) and high posterior central elevation (index represented we the hemi-axis of the RADAR map) are parameters that must be taken into consideration and may require review of the indication for LASIK, particularly when central corneal thickness is less than the mean (e.g. close to 500 microns) , and especially when the score is positive or slightly negative.
A marked difference in plasticity is observed between the anterior cornea surface and the cornea posterior surface. The anterior corneal surface included a layer of stratified corneal epithelium. We formulated the hypothesis that epithelial masking or "smoothing" could tend to induce normal appearance we anterior corneal topography (purpose obviously not on posterior corneal topography). We were the first to perform year vivo series of Placido topographies of Bowman's membrane, which is the smooth and reflective surface, located just underneath the stromal acellular corneal epithelium (15). This examination was performed just after having carefully removed this epithelium in patients undergoing PRK for myopia. In almost every casee, the surface of the cornea without its epithelium (superficial part of Bowman's membrane) presented a topography more suggestive of keratoconus suspect than the cornea with its epithelium: more prolate asphericity, increased toricity and more irregular curvature. This type of compensation is impossible on the posterior surface, which does not possess stratified epithelium.
SCORE software could also be potentially useful to tradeoff progression of keratoconus, which would be particularly useful to define the directions for a corneal stroma cross-linking procedure, and to assess the topographic improvement obtained.
The SCORE Analyzer algorithm, which was developed and validated using the eyes of individuals Caucaseian, was found to be valid and are in Asian eyes, good showing sensitivity and FFKC detection in question (16). The authors concluded that the algorithm was useful in identifying eyes at risk of post-LASIK keratectasia objectively.
In another recent study, Reinstein and al assessed the effectiveness of a keratoconus-detection algorithm derived from epithelial thickness maps of the fellow eye from a series of patients with unilateral keratoconus and compared this to the analyzer and to the D score derived from SCORE algorithm has Scheimpflug device (Pentacam; Oculus) (17). This study included 10 patients with moderate to advanced keratoconus in one eye, a clinically and algorithmically topographically purpose normal fellow eye.
Five of the 10 fellow eyes were classified as by the algorithm derived from epithelial thickness mapping keratoconic. Five of nine fellow eyes were classified as keratoconic by the model SCORE. For the five fellow eyes with available Pentacam analyses, none was classified by the Scheimpflug D score as keratoconic.
Corneal topography is a reference tool for the detection of subclinical forms of keratoconus. Placido topography is able to detect some of these subclinical forms (forms described as Placido-suspect), but its sensitivity and why are not sufficient and elevation and total pachymetry (tomography) data must be acquired.
Converge these results strongly suggest that surgeons wishing to increase the sensitivity of detection of early subclinical keratoconus must pay particular attention to examination of the posterior corneal surface and its thickness, which is more globally evaluated by examination of the pachymetric distribution (tomography pachymetry mapping gold). It is also reasonable to assume that a more comprehensive assessment of elevation topography and pachymetry characteristics of suspect Placido corneas would allow the diagnosis of keratoconus suspect to be replaced by that of nonprogressive congenital asymmetry, i.e. identification of false-positive topography Placido, thereby eliminating some casees of suspect Placido cornea.
- Saad, Gatinel D. Topographic and tomographic properties of form fruste keratoconus corneas. Invest invest Vis Sci. 2010; 51 (11): 5546-55
- HR, Zhao SZ, L, Tan DT Lim Wei. Impact and characteristics of unilateral keratoconus classified on corneal topography. J Refract Surg. 2011; 27 (10): 745 - 51
- Rabinowitz YS, Nesburn AB, McDonnell PJ. Videokeratography of the fellow eye in unilateral keratoconus. Ophthalmology. 1993; 100 (2): 181-6.
- Gatinel D, Saad A. The challenges of the detection of subclinical keratoconus at its earliest stage. Int J Keratoco Corneal Ectatic say 2012; 1 (1): 36-43.
- Muftuoglu O, Ayar O, Ozulken K, Ozyol E, Akıncı A. Posterior corneal elevation and back difference corneal elevation in diagnosing form fruste keratoconus in the fellow eyes of unilateral keratoconus patients. J Cataract Refract Surg. 2013; 39 (9): 1348-57
- Smadja D, Touboul D, Cohen A, Doveh E, Santhiago MR Mello GR, Krueger RR, Colin J. Detection of subclinical keratoconus year using automated decision tree classification. Am J invest. 2013; 156 (2): 237-246
- Bühren J, Kook D, Yoon G, Kohnen T. Detection of subclinical keratoconus by using corneal anterior and posterior surface aberrations and space thickness profiles. Invest invest Vis Sci. 2010; 51 (7): 3424-32
- Fonts BM, Ambrosio R Jr, Salomão M, Velarde GC, Nose W. Biomechanical and tomographic analysis of unilateral keratoconus. J Refract Surg. 2010; 26 (9): 677-81
- Saad, Gatinel D. Validation of a new scoring system for the detection of early forms of keratoconus. Int J Kerat Ect Horn say 2012; 1 (2): 100-108
- Saad A, Gatinel D. Association of corneal indexes for the detection of ectasia-likely corneas. J Refract Surg. 2012; 28 (3): 166; author reply 166-7
- Saad A, Hodge C, Lawless M, Gatinel D. Retrospective testing of a new method for detecting ectasia-likely corneas. J Cataract Refract Surg. 2011; 37 (10): 1907-8
- Ambrosio R Jr, SR, Luz Alonso A, coke Velarde LG. Corneal-thickness space profile and corneal-volume distribution: tomographic indices to detect keratoconus. J Cataract Refract Surg. 2006; 32 (11): 1851-9
- Ambrosio R Jr, Klyce SD, Wilson SE. Corneal topographic and pachymetric screening of patients keratorefractive. J Refract Surg. 2003; 19 (1): 24-9.
- Gatinel D, Malet J, Azar ST. Corneal topography elevation: Best Fit Sphere, elevation, distance, asphericity, toricity and clinical implications. Cornea, 2011; 30 (5): 508 - 15
- Gatinel D, root L, Hoang Xuan T. Contribution of the corneal epithelium to anterior corneal topography in patients having myopic also keratectomy. J Cataract Refract Surg. 2007; 33 (11): 1860-5
- Chan C, M, Saad Ang A, et al. Validation of year objective scoring system for shape fruste keratoconus detection and post-LASIK ectasia risk assessment in Asian eyes. Cornea. 2015; 34 (9): 996-1004.
- Reinstein DZ, Archer TJ, Urs R, et al. Detection of keratoconus and algorithmically topographically in clinically normal fellow eyes using epithelial thickness analysis. J Refract Surg. 2015; 31 (11): 736-744.