Pioneer Research Awards

VICTORIA (Validation of Interleukins and other Cytokines Together with OCT imaging for Rapid Infectious keratitis Assessment )

Dr. Haidar Khalil
Kepler University Clinic Linz

Prof. Dr. Matthias Bolz
Kepler University Clinic Linz

Dr. Nino Hirnschall, PhD
Kepler University Clinic Linz

In the VICTORIA project, we use OCT scans for rapid infectious keratitis assessment. Analyzing tear cytokines alongside OCT images helps distinguish between bacterial pathogens, advancing non-invasive diagnosis for better patient outcomes.

Development and Validation of a Novel Machine Learning-Based Intraocular Lens Power Formula for Cataract Surgery Combined with Endothelial Keratoplasty

Marco Pellegrini
Ospedali Privati Forlì “Villa Igea”
Viale Gramsci 42, Forlì (Italy)

This multicentric observational retrospective study evaluated distinct optimization strategies to improve the accuracy of the SRK/T formula for intraocular lens (IOL) power calculation in patients undergoing combined phacoemulsification and Descemet membrane endothelial keratoplasty (DMEK).

We analyzed 215 patients who underwent combined phaco-DMEK surgery at the tertiary referral centers of Ospedali Privati Forlì, Italy and Southend University Hospital, United Kingdom. The dataset was evaluated using ten different seeds, with each seed partitioning the data into 80% training and 20% independent test sets. For machine learning approaches, the training sets underwent internal 5-fold cross-validation for hyperparameter tuning and optimal parameter selection.

The baseline SRK/T formula demonstrated a mean absolute error (MAE)(defined as the average absolute difference between predicted and actual postoperative spherical equivalent in diopters) of 1.029±0.176 D, achieving 35.0% of eyes within ±0.5 D and 58.0% within ±1.0 D of predicted target refraction.

The segmented residual correction, a statistical stratified regression approach, partitioned eyes into five distinct axial length ranges and developed independent correction models for each segment. Ridge regression with L2 regularization (α=0.12) learned segment-specific coefficients incorporating linear, quadratic, cubic, and interaction terms. This method achieved MAE of 0.414±0.100 D, representing 59.8% error reduction, with 70.0% of eyes within ±0.5 D and 88.0% within ±1.0 D.

XGBoost gradient boosting, a machine learning algorithm employing sequential tree construction where each tree corrects residual errors of the preceding ensemble, achieved very similar performance (MAE 0.415±0.095 D, 59.6% error reduction, 67.0% within ±0.5 D, 82.5% within ±1.0 D). The algorithm incorporated 14 engineered features including biomechanical ratios and a K×CCT interaction term, optimizing through 5-fold cross-validation grid search with L1/L2 regularization.

In conclusion, both the statistical regression approach and the machine learning XGBoost method demonstrated approximately 60% reduction in MAE compared to the standard SRK/T formula, representing a significant advancement that could substantially improve refractive outcomes and reduce the need for postoperative corrections in the setting of combined phaco-DMEK surgery.

ANTISEPSIS (Pre-Operative Antisepsis and treatment of Blepharitis with Povidone-Iodine 1%)

Dr Evgeny Gelman

Elishai Assayag

David Teren

Shaare Zedek Medical Center, Jerusalem

Summary: 
In this 2-staged study, povidone-iodine (PVI) 1% will be used for 2 clinical indications. 1) The aim of the first stage is to investigate the effect of eyelids cleansing with 1% PVI on the clinical status of the eyelids and ocular surface of patients with anterior blepharitis. Positive results may suggest a safe and effective treatment option for this common diagnosis. We hypothesize that PVI 1% will be superior compared to standard eyelid hygiene wipes, in terms of objective and subjective outcomes of blepharitis. 100 blepharitis patients will be recruited. Each patient will treat one eye once daily with 1% PVI for 30 days by scrubbing the eyelid margin with the solution. The fellow eye will serve as the control and will be given the standard treatment (commercial eye wipes). Prior to treatment initiation, various ocular surface variables will be assessed, such as dry eye grading, subjective discomfort scales, ocular surface questionnaire and other clinical signs. After 30 days an identical evaluation will be performed. 2) The aim of the second stage is to evaluate the efficacy of povidone-iodine 1% in reducing bacterial flora load in the conjunctiva before cataract surgery and to show non-inferiority to povidone-iodine 5% antiseptic pre-operative protocol. The hypothesis is that povidone-iodine 1% will be less toxic to the corneal epithelium. Positive results may indicate an important change to be considered in the pre-operative antisepsis protocol. One hundred patients who were referred for cataract will be divided into two groups. In both groups, drops of local anesthetic will be instilled and then a culture will be taken from the conjunctiva before irrigation with povidone-iodine solution. In one group, the conjunctival sac will be irrigated with povidone-iodine 5%. In the second group, three consecutive ocular surface irrigations with povidone-iodine 1% will be performed with 30 seconds between each application. A sample will be taken from the lower fornix, and 100 microliters of 0.5% sterile sodium thiosulfate will be inserted into the flocked applicator test tube in order to neutralize the action of povidone-iodine. Then each patient will be evaluated on a slit lamp for superficial punctate keratopathy.

Improving the timing of intraocular lens exchange surgery: Clinical study on the functional and morphological course of patients with IOL opacification

Dr Timur Yildrim

Department of Ophthalmology, Heidelberg University Hospital, Germany

Late intraocular lens (IOL) opacification is the primary cause for IOL explantation today.(Goemaere et al. 2020) The condition usually takes several years to develop and results in a deteriorated visual quality.(Labuz et al. 2018) Clinically, optical parameters affected are the straylight level, contrast sensitivity and visual acuity.(Scherer et al. 2020) However, even patients with morphologically advanced IOL opacification can achieve an excellent visual acuity of 20/20 Snellen or better.(Yildirim et al. 2020) Defining the timing of IOL exchange can therefore be a challenge and has to be weighted against risks of a demanding surgery. Today, the decision and timing to perform IOL exchange is mainly based on the clinician's empirical judgment. A standardized assessment method allowing more solid advice for patients suffering from this condition is needed.

We hypothesize that IOL exchange in eyes with IOL calcification decreases intraocular straylight but has little effect on the visual acuity. Sample size calculation was based on the results of a previous laboratory study and indicated that a total number of 22 eyes are needed to achieve a 95% chance of detecting a difference at a 5% level of significance (2 tailed).(Yildirim et al. 2021)

In this single-centre, non-interventional clinical study, we aim to include 30 eyes of patients presenting with IOL opacification without additional ocular comorbidities. Prior to surgery, we measure various functional optical parameters including the corrected distance visual acuity, contrast sensitivity and intraocular straylight (C-quant, Oculus, Berlin, Germany). Additionally, we aim to assess the IOL morphology using a high-resolution optical coherence tomograph (Anterion, Heidelberg Engineering, Heidelberg, Germany), analyse the corneal topography and wavefront (Pentacam Wave, Oculus, Berlin, Germany), assess surgical complications and evaluate the course of functional optical parameters. Four to 12 months after surgery, patients will be invited for follow-up to re-evaluate the measurements and calculate pre- and postoperative differences. 

By collecting clinical data and creating a register of patients with IOL opacification that undergo IOL exchange, we aim to develop clinical guidelines to distinguish between different types of this condition and ultimately better diagnose, advise and treat such patients. Additionally, the economic burden of an IOL exchange surgery with its sequelae can then be better estimated and taken into consideration when treating patients.

CAMERA (Innovative Ocular Surface Grading Camera) 

Martina Angi

Fondazione IRCCS Istituto Nazionale dei Tumori

Bernhard Steger

Medical University of Innsbruck

Mario Romano

Humanitas University

Vito Romano

Spedali Civili di Brescia

To assess the ocular surface grading system using two imaging devices: cornea dome lens (CDL) and slit lamp camera. The CDL camera is a novel device able to acquire images of the entire ocular surface. It provides optimal details with accurate colour reproduction, with a resolution of 12 microns over a 21 mm field size with a curvature of 8 mm in an ex vivo model. CDL, not CE marked yet, will be compared with a state-of-the-art commercial slit lamp camera.

PCME is associated to genetic alterations of single nucleotide polymorphisms (SNP)” - (“PCME-SNP”)

PD Dr. Tobias Brockmann
Universitätsmedizin Rostock

Pseudophakic cystoid macular edema (PCME) is the most common complication following cataract surgery, which is usually accompanied with a considerable loss of visual acuity. The incidence of a clinically significant PCME was reported with 1-8% (Grzybowski A, et al. Clin Interv Aging. 2016). Especially patients with the presentation of a PCME in the first eye have a 20-fold higher risk to suffer a PCME in the fellow eye (McCafferty et al. BMC Ophthalmology 2017). These epidemiologic aspects draw the attention to genetic causes in the development of PCME. So far, genetic causes in the pathogenesis of PCME have not been revealed.

The research hypothesis of this study is that PCME is associated to genetic alterations of single nucleotide polymorphisms (SNP).

In the research project, Dr. Brockmann will identify single nucleotide polymorphisms (SNP) that are significantly correlated to the presentation of a clinically relevant PCME. To conduct the study, a review of retrospective clinical files is to be assessed to identify patients without particular risk factors, which suffered from a PCME after uneventful cataract surgery. Selected patients will be scheduled, within this study, for blood sampling and macular diagnostics. A matched patient collective, without the history of PCME, will serve as reference. DNA will be extracted from all samples and further processed for microarray-based genome-wide SNP genotyping.

As clinical and scientific impact collects data, the expected findings of this project will account an improved understanding of patho-mechanisms involved in PCME formation. In a translational approach, new therapeutic and/or prophylactic concepts may arise that might also be transferred to other indications, other than PCME. Furthermore, the data will provide a set of relevant SNPs that might be of diagnostic relevance to preoperatively identify patients with a high individual risk for PCME, beyond known risk factors. In this respect, both therapeutic and diagnostic aspects are of high clinical and economic impact due to the common occurrence and the unmet medical needs of PCME.

This study has begun in December 2020, supported by the ESCRS, and an initial report expects to provide the key outcomes by the end of October, 2022.