A machine learning approach to explore predictors of graft detachment following posterior lamellar keratoplasty: a nationwide registry study

Muijzer, Marc B.; Hoven, C. M. W.; Frank, Laurence E.; Vink, G.; Bartels, Marjolijn C; Cheng, Yanny Y. Y.; Dhooge, Mario R. P.; Dickman, Mor M.; Dooren, Bart T. H. van; Eggink, C.A.; Geerards, Annette J M; Goor, Tom A. van; Lapid-Gortzak, Ruth; Luijk, Chantal M. van; Meulen, Ivanka J. van der; Nieuwendaal, Carla P.; Nuijts, Rudy M.M.A.; Nobacht, Siamak; Oahalou, Abdulkarim; Oosterhout, Emile C. A. A. van; Remeijer, Lies; Rooij, Jeroen van; Santana, Nathalie T.Y.; Stoutenbeek, Remco; Tang, Mei L.; Vaessen, Thijs; Visser, Nienke; Wijdh, Robert H J; Wisse, Robert P. L.

doi:10.1038/s41598-022-22223-y

Cited by 11 publications

(11 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…It has been reported that patients with prior failed transplants are at higher risk of requiring regrafting with subsequent corneal transplants. 12 , 13 Although not statistically significant, our findings are in line with this finding; patients for whom a regraft was necessary were more likely to have already had a history of prior transplant when compared to those patients who did not need a regraft (50% versus 36.9%, respectively; p = 0.264). Time in preservation and DTC seemed to be most related to the need to regraft, though neither achieved statistical significance.…”

Section: Resultssupporting

confidence: 83%

Impact of Donor, Recipient, and Graft Characteristics on Corneal Transplantation Outcomes

Shah¹,

Hasan²,

Winokur³

et al. 2023

OPTH

View full text Add to dashboard Cite

Purpose Although several donor, recipient, and graft characteristics have been studied in relation to corneal transplantation outcomes, no study to our knowledge has assessed the impact of donor cooling times on postoperative outcomes longitudinally. With only one corneal graft available for every 70 needed worldwide, this study seeks to identify any factors that could alleviate this shortage. Methods Patients undergoing corneal transplantation at the Manhattan Eye, Ear & Throat Hospital over a 2-year period were retrospectively studied. Study metrics included age, diabetic history, hypertensive history, endothelial cell density, death-to-preservation time (DTP), death-to-cooling time (DTC), and time-in-preservation (TIP). Postoperative transplantation outcomes, including best corrected visual acuity (BCVA) at 6- and 12-month follow-up visits, need for re-bubbling, and need for re-grafting, were assessed. Unadjusted univariate and adjusted multivariate binary logistic regressions were performed to determine the association of cooling and preservation parameters with corneal transplantation outcomes. Results Among 111 transplants, our adjusted model found that DTC ≥4 hours was associated with significantly worse BCVA, but only at 6-month postoperative follow-up (odds ratio [OR]: 0.234; 95% confidence interval [CI]: 0.073–0.747; p = 0.014). By 12-month follow-up, DTC > 4 hours was no longer associated with BCVA in a statistically significant manner (OR: 0.472; 95% CI: 0.135–1.653; p = 0.240). A similar trend was found at a DTC cutoff of ≥3 hours. None of the other studied parameters, including DTP, TIP, donor age, or medical history were significantly correlated with transplantation outcomes. Conclusion Longer DTC or DTP did not have a statistically significant effect on corneal graft outcomes after one year, though short-term outcomes were improved in donor tissues with DTC below four hours. None of the other studied variables correlated with transplantation outcomes. Given the global shortage of corneal tissue, these findings should be considered when determining suitability for transplantation.

show abstract

Section: Resultssupporting

confidence: 83%

Impact of Donor, Recipient, and Graft Characteristics on Corneal Transplantation Outcomes

Shah¹,

Hasan²,

Winokur³

et al. 2023

OPTH

View full text Add to dashboard Cite

show abstract

“…These differences will, however, be recorded during the data entry on the day of surgery into the eCRF. We standardise using an air tamponade because gas is, according to a data registry from the Netherlands, associated with poorer outcomes 24…”

Section: Methods and Analysismentioning

confidence: 99%

Belgian Endothelial Surgical Transplant of the Cornea (BEST cornea) protocol: clinical and patient-reported outcomes of Ultra-Thin Descemet Stripping Automated Endothelial Keratoplasty (UT-DSAEK) versus Descemet Membrane Endothelial Keratoplasty (DMEK) – a multicentric, randomised, parallel group pragmatic trial in corneal endothelial decompensation

Ní Dhubhghaill,

de Bruyn,

Claerhout

et al. 2023

BMJ Open

View full text Add to dashboard Cite

ObjectivesCorneal blindness is the third most frequent cause of blindness globally. Damage to the corneal endothelium is a leading indication for corneal transplantation, which is typically performed by lamellar endothelial keratoplasty. There are two conventional surgical techniques: Ultra-Thin Descemet Stripping Automated Endothelial Keratoplasty (UT-DSAEK) and Descemet Membrane Endothelial Keratoplasty (DMEK). The purpose of this study is to compare both techniques.Methods and analysisThe trial compares UT-DSAEK and DMEK in terms of clinical and patient reported outcomes using a pragmatic, parallel, multicentric, randomised controlled trial with 1:1 allocation with a sample size of 220 participants across 11 surgical centres. The primary outcome is the change in best-corrected visual acuity at 12 months. Secondary outcomes include corrected and uncorrected vision, refraction, proportion of high vision, quality of life (EQ-5D-5L and VFQ25), endothelial cell counts and corneal thickness at 3, 6 and 12 months follow-up appointments. Adverse events will also be compared 12 months postoperatively.Ethics and disseminationThe protocol was reviewed by ethical committees of 11 participating centres with the sponsor centre issuing the final definitive approval. The results will be disseminated at clinical conferences, by patient partner groups and open access in peer-reviewed journals.Governance of the trialBoth, trial management group and trial steering committee, are installed with representatives of all stakeholders involved including surgeons, corneal bankers, patients and external experts.Trial registration numberNCT05436665.

show abstract

“…Overall, a growing body of evidence supports the use of AI-based pre- and postoperative screening as a tool that could help with predicting and diagnosing pathologic consequences of surgery. 14 , 19 , 30 , 32 , 33 Deep learning models have been employed to identify postoperative DMEK graft detachments, 19 to predict future need and suitability for keratoplasty, 30 to quantify DMEK graft detachment segmentation, 32 to identify the best predictors for graft detachment after endothelial keratoplasty, 33 and to predict the need for rebubbling after DMEK surgery. 34 Indeed, clinical trials are currently starting to incorporate AI into corneal pathology identification.…”

Section: Discussionmentioning

confidence: 99%

Deep Learning Using Preoperative AS-OCT Predicts Graft Detachment in DMEK

Patefield

Meng

Airaldi

et al. 2023

Trans. Vis. Sci. Tech.

View full text Add to dashboard Cite

Purpose To evaluate a novel deep learning algorithm to distinguish between eyes that may or may not have a graft detachment based on pre–Descemet membrane endothelial keratoplasty (DMEK) anterior segment optical coherence tomography (AS-OCT) images. Methods Retrospective cohort study. A multiple-instance learning artificial intelligence (MIL-AI) model using a ResNet-101 backbone was designed. AS-OCT images were split into training and testing sets. The MIL-AI model was trained and validated on the training set. Model performance and heatmaps were calculated from the testing set. Classification performance metrics included F1 score (harmonic mean of recall and precision), specificity, sensitivity, and area under curve (AUC). Finally, MIL-AI performance was compared to manual classification by an experienced ophthalmologist. Results In total, 9466 images of 74 eyes (128 images per eye) were included in the study. Images from 50 eyes were used to train and validate the MIL-AI system, while the remaining 24 eyes were used as the test set to determine its performance and generate heatmaps for visualization. The performance metrics on the test set (95% confidence interval) were as follows: F1 score, 0.77 (0.57–0.91); precision, 0.67 (0.44–0.88); specificity, 0.45 (0.15–0.75); sensitivity, 0.92 (0.73–1.00); and AUC, 0.63 (0.52–0.86). MIL-AI performance was more sensitive (92% vs. 31%) but less specific (45% vs. 64%) than the ophthalmologist's performance. Conclusions The MIL-AI predicts with high sensitivity the eyes that may have post-DMEK graft detachment requiring rebubbling. Larger-scale clinical trials are warranted to validate the model. Translational Relevance MIL-AI models represent an opportunity for implementation in routine DMEK suitability screening.

show abstract

A machine learning approach to explore predictors of graft detachment following posterior lamellar keratoplasty: a nationwide registry study

Cited by 11 publications

References 50 publications

Impact of Donor, Recipient, and Graft Characteristics on Corneal Transplantation Outcomes

Impact of Donor, Recipient, and Graft Characteristics on Corneal Transplantation Outcomes

Deep Learning Using Preoperative AS-OCT Predicts Graft Detachment in DMEK

Contact Info

Product

Resources

About