EyeWorld is the official news magazine of the American Society of Cataract & Refractive Surgery.
Issue link: https://digital.eyeworld.org/i/618732
EW INTERNATIONAL 60 January 2016 by Matt Young and Gloria D. Gamat EyeWorld Contributing Writers 10 years of experience doing the same analysis. Indeed, numerous clinical cri- teria for LASIK risk assessment have been developed through the years. "We see these criteria evolve, with potential benefits and criti- cism," said Iqbal "Ike" K. Ahmed, MD, assistant professor, University of Toronto. However, artificial intelligence, Dr. Ahmed said, is a powerful and interesting concept. "With the right diagnostics and right demographics, you could churn these into a scoring system," he said. "At the end of the day, we look at risk stratification, and having an automated algorithm could be handy," he concluded. EW Reference 1. Saad A, et al. Topographic and tomograph- ic properties of forme fruste keratoconus corneas. Invest Ophthalmol Vis Sci. 2010 Nov;51(11):5546–55. Editors' note: Drs. Ahmed, Dean, and Rosman have no financial interests related to this article. Contact information Ahmed: ike.ahmed@utoronto.ca Dean: info@eyeinstitute.co.nz Rosman: mohamad.rosman@snec.com.sg of cases with this system, we suggest that SCORE values falling within the –0.6 and 0.5 interval be charac- terized as equivocal and addition- al testing for confirmation of its significance be at the discretion of the attending ophthalmologist," the researchers said. According to Dr. Rosman, if the score is grossly abnormal, it indi- cates a high risk of post-LASIK ecta- sia. Therefore, LASIK or other forms of corneal refractive surgery should not be performed. "If the SCORE is borderline, it increases our index of suspicion; the surgeon may opt to perform addi- tional scans with different tomogra- phers to confirm or elect to perform surface ablation," he said. Impact in clinical practice "Algorithms are built into most topographers that map the front of the eye, providing doctors with an index of suspicion," said Simon J. Dean, FRANZCO, specialist eye sur- geon, Eye Institute, Auckland, New Zealand. A physician can't pick all cases of keratoconus by just looking at the scans, Dr. Dean said. Detect- ing keratoconus is all about pattern recognition, he said, and perhaps the time will come when a system with enough built-in algorithms can analyze keratoconus risk that would be equivalent to a doctor who has single value; this makes it easier for surgeons to determine whether there is a risk of post-LASIK ectasia," Dr. Rosman explained. According to Dr. Rosman, the SCORE analyzer has been used in patients at SNEC since January 2014. Its main benefit, he noted, is greater patient safety. "We are able to detect more subtle abnormal corneas, and it gives us a higher index of suspicion in certain patients," he said. The use of the SCORE analyzer depends on whether the clinic has the Orbscan machine. "Some clinics use the Pentacam [Oculus, Wetzlar, Germany], which has its own set of post-LASIK ectasia risk predictive values; SNEC has both the Orbscan and Pentacam," Dr. Rosman said. Development of SCORE analyzer algorithm In 2010, Saad and Gatinel 1 reported the finding that when elevation and tomography data are added in when evaluating corneas, the sensitivity and specificity in detecting FFKC are increased. "The indices generated from corneal thickness and curvature measurements over the entire cornea centered on the thinnest point, and calculations of percentage changes in the thickness and anterior and posterior curvature variations are able to identify very mild forms of keratoconus undetected by Placido-based topography alone," they reported. This was the very concept that the artificial intelligence system in- corporated into the SCORE analyzer was based upon and designed for the detection of FFKC. Although origi- nally designed for Caucasian eyes, recent validation data indicated that the algorithm's discriminatory functions need not be adjusted for Asian eyes. "However, descriptive analyses of SCORE values for the false-neg- ative and false-positive groups showed that the values were close to zero, with the median at –0.6 and 0.5, respectively; based on these findings, to optimize the screening System is verified and employed rigorously in Singapore W hile LASIK has wowed patients and ophthal- mic surgeons alike for many years, keratec- tasia remains its most devastating postoperative corneal complication. A main culprit is undetected forme fruste keratoconus (FFKC), an elusive condition difficult to catch at its earliest stage. In recent years, various systems in topographic assessments have been described by many experts worldwide in the hope of catching the first signs of FFKC. However, no system has been able to detect all cases of FFKC, and there isn't one that can predict the risk of develop- ing post-LASIK keratectasia. A group of investigators from the Singapore National Eye Centre (SNEC), Singapore Eye Research Institute (SERI), and the Rothschild Foundation (Paris) validated the potential of the SCORE analyzer— software linked to the Orbscan IIz corneal topography system (Bausch + Lomb, Bridgewater, N.J.)—that would help address the keratoconus detection dilemma. SCORE analyzer validated in Asian eyes "The SCORE analyzer algorithm, developed and validated in eyes of white subjects, was found to be valid and consistent in Asian eyes, show- ing good sensitivity and specificity in FFKC detection, and to be useful in objectively identifying cases at risk of post-LASIK keratectasia," the investigators reported. "The SCORE analyzer provides a more visual and quantitative report compared to just looking at the stan- dard Orbscan topography printout," said Mohamad Rosman, FRCS(Ed), assistant professor, head and senior consultant, refractive surgery de- partment, Singapore National Eye Centre. "It quantifies several values that predict the presence of an abnormal cornea and summarizes them into a Hope in catching keratoconus early Acute hydrops in keratoconus Source: Karl Brasse, MD