SPS-307 Keratoconus: Artificial Intelligence/Algorithms, CXL, PRK | ASCRS
2020 ASCRS Virtual Annual Meeting

presentations on demand

You do not have access to this library

This content is accessible only to 2020 ASCRS Virtual Annual Meeting attendees. If you attended the virtual meeting, please log in with the account associated with your virtual meeting registration.  To log in, click the teal "Login" button in the upper right-hand corner of this page.

Papers in this Session
Expand each tab below to view the paper abstract for each paper within this session.
Pattern of Success Post-Sequential Crosslinking and Topography-Guided PRK for Keratoconus
Authors
Alanna S. Nattis, DO
Eric D. Rosenberg, DO
Eric D. Donnenfeld, MD

Purpose
To evaluate visual and astigmatic improvement in keratoconus patients undergoing sequential collagen cross-linking (CXL) and topography-guided photorefractive keratectomy (PRK) through 18 months postoperatively. In addition, change in corneal thickness, keratometry values, and total higher order aberrations was analyzed.

Methods
This was a retrospective chart review of 14 eyes (13 patients) who underwent modified epithelium-on CXL followed by topography guided PRK at time of refractive & keratometric stability. Change in uncorrected and best-corrected visual acuity was measured from baseline, post-CXL, 6, 12, and 18-months post-PRK. Change in mean keratometry, steepest keratometry, keratometric astigmatism, central corneal thickness, and total higher order aberrations was analyzed at each time point. Statistical analysis was performed to determine significant relationships between demographic and baseline variables and final visual and astigmatic outcomes.

Results
14 eyes (13 patients) successfully underwent sequential CXL & topography-guided PRK. Seven eyes had only topographic irregularities treated, 7 eyes had at least some degree of refractive error treated during PRK. 18 months post-PRK, 100% of eyes had best-corrected visual acuity of 20/40 or better (this was true of both those eyes treated for refractive error and those treated for topographic irregularities alone). Visual acuity and astigmatic values continued to improve over the 18 month postoperative period. In addition, corneal thickness appeared to continue to modestly decrease over this period. Total higher order aberrations appeared to decrease and then level off at 18 months post-PRK.

Conclusion
Continued improvement of vision & astigmatic values, in conjunction with decreasing corneal thickness over an 18 month postoperative period, suggests corneal remodeling post-sequential CXL & topography guided PRK. This treatment protocol appears efficacious for keratoconic eyes post-CXL, regardless of baseline values for vision or astigmatism.
Improving Outcomes for Crosslinking Keratoconus Eyes: Results of the New Minneapolis Protocol for the Treatment of Keratoconus Eyes
Authors
Hui Zhao, MD, PhD
Paul Hammond, OD
Mark C. Lobanoff, MD, ABO

Purpose
Investigate the visual and keratometric outcomes of keratoconus patients by using topographic-guided PRK while simultaneously applying new improved crosslinking techniques.

Methods
A retrospective study was conducted in KCN patients who underwent topography-guided PRK, treating the topography and the sphere and cylinder of each eye. As this treatment removes more stromal tissue than treating topography alone (Athen’s protocol), it was critical to obtain excellent crosslinking. A novel silicone suction ring device was developed to allow more complete and even saturation of the stroma with riboflavin. UVA was applied in a pulsed accelerated manner. UDVA, CDVA, keratometric astigmatism, spherical equivalent, maximum and mean keratometry (K) readings, and central corneal thickness were collected baseline, 1, 3, 6 and 12 months after treatment.

Results
The study comprised 37 patients (50 eyes) with a mean age of 37.32 years ± 12.56 (SD). There was a significant improvement of UDVA starting at 1 month and continued to improve to 0.28 ± 0.32 at 12 months postoperatively from 0.89 ± 0.48 at baseline (P < 0.01). At 12 months, 24.2% of eyes had 20/20 or better UDVA, and 48.5% had 20/25 or better UDVA; 75.8% gained over 2 lines in UDVA; 94% of eyes had 20/40 or better CDVA. Maximum K decreased from 51.63 ± 6.76 at baseline to 48.61 ± 5.08 at 12 months (P < 0.01). Mean K had a similar change postoperatively (P < 0.01). There were no significant adverse events in any case.

Conclusion
Studies show improved visual results treating topography and sphere/cylinder in sequential CXL. Simultaneous topography-guided PRK with CXL has been shown to yield superior results to sequential treatments. This protocol yields better results by using topography-guided PRK along with simultaneous improved technique CXL.
Two-Year Outcome of Topography-Guided Photorefractive Keratectomy with Collagen Cross-Linking for Keratoconus. (No Audio)
Authors
Simon P. Holland, MD, FRCS
David TC. Lin, MD, FRCS
Mukhtar Bizrah, MBBS, FRCOphth
Shwetabh Verma, MSc

Methods
A retrospective consecutive series of keratoconic eyes were studied to evaluate the outcomes of Topographic Guided Photorefractive Keratectomy (TG-PRK) with the Schwind Amaris 1050 Excimer laser with simultaneous corneal collagen cross-linking (CXL). Image capture with Sirius and CXL with the Athens protocol. Pre-operative and post-operative uncorrected distance visual acuity (UDVA), corrected distance visual acuity (CDVA), manifest refraction (MR) and topographic data were analyzed. Cases with sufficient data at 2 year follow-up were included.

Results
88 eyes had sufficient data at 2 year for analysis. 50 of 88 (57%) showed UDVA ≥20/40 post-operatively. 36 eyes (41%) had improved CDVA and 20 (23%) gained two or more lines while 22 eyes (25%) had loss CDVA with 8 (9%) lost 2 lines or more. Mean astigmatism was reduced from 2.83±1.81D to 1.82±1.64D. Mean spherical equivalent was improved from -3.46±3.65D to -0.52±2.14D. 4 eyes showed progression and 5 with haze judged sufficient to reduce CDVA.

Conclusion
Two year results of topography-guided PRK (Schwind Amaris) treatment with CXL for keratoconus show efficacy and safety, with more than half achieving 20/40 UDVA or better. 41% had improved CDVA and almost a quarter had CDVA improved two or more lines. It provides an alternative to contact lens intolerant keratoconus patients.
Accuracy of an Artificial Intelligence Algorithm in the Autonomous Diagnosis of Keratoconus
Authors
Eyüp Özcan, MD
Amr Elsawy, MSc
Taher K. Eleiwa, MD, MSc
Collin B. Chase
Mohamed Tolba, MD, MSc
Sonia H. Yoo, MD, ABO
Mohamed Abou Shousha, MD

Methods
In this prospective case control study, 43 eyes of 23 patients with keratoconus and 28 eyes of 14 healthy subjects were imaged using an ASOCT device (Envisu 2210, Bioptigen, Buffalo, IL). Thirty-six frames radial scan pattern images were evaluated using an AI algorithm (Bascom Palmer AI, version 1.0, Miami, FL). Receiver operating curves (ROC) were generated at patient and eye levels.

Results
The AI algorithm was able to correctly differentiate all patients with keratoconus from healthy subjects and thus achieving accuracy, sensitivity and specificity of 100%. It was able to diagnose 42 out of the 43 keratoconic eyes, achieving accuracy of 97.9%, sensitivity of 95.3% and specificity of 94.6%.

Conclusion
The AI algorithm is a novel technique with excellent accuracy, sensitivity, and specificity in the diagnosis of keratoconus.
Transepithelial Topography Guided PRK for Keratoconus and Ectasia after Refractive Surgery – 6 Month Results
Authors
Steven A. Greenstein, MD
Peter S. Hersh, MD

Purpose
To evaluate the visual and topographic outcomes 6 months after transepithelial topography guided PRK (TETGPRK) in patients with keratoconus and post refractive ectasia.

Methods
In this retrospective analysis, 13 eyes underwent TETGPRK. Preoperatively, a Topolyzer (WaveLight®, Alcon, USA) and an OCT (Optovue, Inc, USA) epithelial thickness map were obtained to perform a transepithelial excimer (WaveLight® Allegretto, Alcon, USA) ablation. 5 patients received simultaneous corneal collagen crosslinking (CXL), 3 patients had a history of previous CXL, and 5 patients did not have CXL performed. Outcomes analyzed included, maximum keratometry (Kmax), thinnest pachymetry (Pthin) (both measured on the Pentacam topographer), uncorrected (UDVA) and corrected distance visual acuity (CDVA), and manifest refraction spherical equivalent (MRSE), at 6 months postoperatively.

Results
UDVA and CDVA improved by 0.49±0.32 logmar lines (p<0.001) and 0.13±0.18 logmar lines (p=0.002), at 6 months postoperatively. MRSE changed by 0.7±3.7D, but this change failed to reach statistical significance (p=0.5). With regards to tomography, Kmax flattened by 3.6±3.2D (P<0.001), and Pthin decreased by 34.4±22.5µm (p<0.001). In general, there was a trend toward more improvement of visual acuity in the eyes that underwent TETGPRK alone rather than TETGPRK and CXL simultaneously (TETGPRKCXL). However, there was more flattening and corneal thinning in the eyes that underwent TETGPRKCXL than TETGPRK alone.

Conclusion
Visual acuity (UDVA and CDVA) and topography (kmax) improved 6 months after TETGPRK. Further prospective and retrospective analysis is underway to determine if there is a statistical difference between TETGPRK alone and TETGPRKCXL, and to evaluate long term outcomes.
Partial-Refraction Topography-Guided Ablation Combined with Additional CXL (Athens Protocol) in Eyes Following Previous Failed CXL
Authors
Gregory J. Pamel, MD
A. John Kanellopoulos, MD, ABO

Purpose
To evaluate the efficacy and safety of applying the Athens Protocol (AP) after failed CXL in keratoconus

Methods
In this prospective, non-comparative, consecutive case series, 17 keratoconic eyes underwent the Athens Protocol after failed Dresden-protocol or epithelium-on CXL. Perioperative evaluation included UDVA, CDVA, refraction, keratometry, topography and tomography by Scheimpflug and anterior segment OCT

Results
Mean values at 2 years: UDVA improved 4 lines, CDVA 5 lines, steepest keratometry improved from 53.5 to 46.2 Diopters, but with a very wide range of 4.5-14 Diopters of flattening achieved. Re-epithelialization achieved at an average of day 5. Effective CXL line visualized in OCT cross-sections at an average 70% stromal depth. Epithelial remodeling resulted in an overall thickness reduction, while great increase over the cone area. None of the cases showed signs of ectasia progression.

Conclusion
CXL with the Athens Protocol offers a safe and what appears to be more effective option when conventional or epithelium-on CXL fails. The amount of corneal flattening achieved appears to be far greater than that expected from the ablation and the CXL-related added.
Corneal Crosslinking, Topographic PRK, and Cataract Surgery to Improve Cataract Outcomes
Author
Eric D. Donnenfeld, MD

Methods
Retrospective chart review of 17 eyes post-modified epithelium-on CXL followed by topography guided PRK, and then cataract surgery. Change in uncorrected visual acuity (UCVA) and best-corrected visual acuity (BCVA) was measured at baseline, post-CXL, post-PRK and post cataract surgery. Change in keratometry, refractive error, astigmatism and IOL utilized was analyzed.

Results
17 eyes underwent CXL followed by topography guided PRK and then cataract surgery. Mean UCVA improved from 20/400 preop to 20/88 postop (P=0.01) and BCVA improved from 20/87 preop to 20/34 postop (P=0.005). Mean refractive sphere preop was -8.84 and post op was -1.02. Mean refractive cylinder preop was -7.12 and postop -1.45. All patients had at least a 2-line improvement in BCVA and there were no significant complications noted. Mean time from CXL to cataract surgery was 6 months

Conclusion
Sequential CXL, excimer laser topographic PRK, and cataract surgery is a successful procedure to improve both UCVA and BCVA in patients with keratoconus and cataracts.
Combined Topo-Guided PRK and CXL with Oxygen Supplementation in Keratoconus: 3 Year Follow up (No Audio)
Author
Christoph F. Kranemann, MD

Purpose
The study was to determine the effectiveness and safety of topography guided PRK combined with topography guided CXL and Oxygen supplementation compared to standard CXL alone.

Methods
Consecutive patients were randomized to CXL alone versus topography guided PRK with topography guided CXL combined with supplemental Oxygen and Mitomycin. Topo-guided PRK and CXL were centred on the cone and had a central treatment area of 4.5-5 mm and a 1.5 mm transition zone. Total energy for central zone was 10 Joules and 7.2 Joules for transition zone. For regular CXL a standard 9 mm zone and 7.2 Joules were used. They were followed for a minimum of 3 years with sequential topography, best corrected vision and corneal endothelial cell count. Any complications from haze to infection were recorded. Dry eye evaluations including tear osmolality were performed.

Results
44 eyes of 29 patients were enrolled. 23 were in the topo-guided group (group 1) and 21 in the standard CXL (group 2). At month 1 mean corrected vision was 20/80 and mean change ink's 3.5 Diopters in Group 1 and 20/60 and 2.0 diopters in Group 2. Month 24 the mean corrected VA was 20/20 and K change 5.4 diopters in Group 1 and 20/50 and 3.5 diopters in Group 2 (P<.01) and month 36 the mean corrected VA was 20/20 and K change 5.6 in Group 1 and 20/60 and 3.6 diopters in Group 2 (P<.01). Mild corneal haze occurred in 4/23 in Group 1 resolved at 3/12 and 6/21 in Group 2 with 2/21 having some haze still at months 36 in Group 2 (P<.1).

Conclusion
In a pilot study topography guided PRK with topo guided CXL using a smaller zone appears to be safe and effective. It might have the potential to yield more effective results and greater topographic stability then regular CXL.
Machine Learning Optimization to Integrate Scheimpflug Tomography and Biomechanics for Enhancing Ectasia Detection
Author
Renato Ambrósio, MD, PhD

Methods
A multicentric retrospective study including twenty-five international centers, comprehensively analyzed tomographic and biomechanical data from the Pentacam and Corvis ST (Oculus; Wetzlar, Germany). One eye randomly selected from 1,653 patients with normal corneas, and from 943 patients with clinical mild to moderate keratoconus were included. The study also included the eyes with normal topography from 479 patients with very asymmetric ectasia (VAE-NT) and the 372 unoperated ectatic (VAE-E) fellow eyes. The current TBI was tested and further optimizated using novel artificial intelligence machine learning for augmenting accuracy.

Results
For clinical ectasia (KC and VAE-E), the TBI had AUC of 0.999, with 98.3% sensitivity and 99% specificity (cut off 0.52). Considering the VAE-NT, the TBI had AUC 0.907 with 81.4% sensitivity and 84.6% specificity (cut off 0.25). The AUCs of the TBI were statistically that the AUC of CBI (Corvis Biomechanical Index; 0.956 and 0.78) and BAD-D (Belin/Ambrosio Deviation; 0.994 and 0.824; De Long; p<0.001). A new machine-learning algorithm, the BrAIN-TBI was developed with significantly higher AUC (0.978) for detecting VAE-NT, having 90.6% sensitivity and 94.5% specificity (cut of 0.37). Considering all cases, the BrAIN-TBI had higher AUC (0.991) than the TBI (0.974; De Long, p<0.001).

Conclusion
The current TBI had excellent accuracy for clinical ectasia, but there was a reduction in sensitivity for the VAE-NT cases. Improvements in artificial intelligence algorithms to integrate tomography and biomechanics data augments the accuracy to detect ectasia, possibly epitomizing the characterization of ectasia susceptibility.

We use cookies to measure site performance and improve your experience. By continuing to use this site, you agree to our Privacy Policy and Legal Notice.