Predicting Visual Improvement After Macular Hole Surgery: A Combined Model Using Deep Learning and Clinical Features

Abstract

Purpose The purpose of this study was to assess the feasibility of deep learning (DL) methods to enhance the prediction of visual acuity (VA) improvement after macular hole (MH) surgery from a combined model using DL on high-definition optical coherence tomography (HD-OCT) B-scans and clinical features. Methods We trained a DL convolutional neural network (CNN) using pre-operative HD-OCT B-scans of the macula and combined with a logistic regression model of pre-operative clinical features to predict VA increase ≥15 Early Treatment Diabetic Retinopathy Study (ETDRS) letters at 6 months post-vitrectomy in closed MHs. A total of 121 MHs with 242 HD-OCT B-scans and 484 clinical data points were used to train, validate, and test the model. Prediction of VA increase was evaluated using the area under the receiver operating characteristic curve (AUROC) and F1 scores. We also extracted the weight of each input feature in the hybrid model. Results All performances are reported on the held-out test set, matching results obtained with cross-validation. Using a regression on clinical features, the AUROC was 80.6, with an F1 score of 79.7. For the CNN, relying solely on the HD-OCT B-scans, the AUROC was 72.8 ± 14.6, with a F1 score of 61.5 ± 23.7. For our hybrid regression model using clinical features and CNN prediction, the AUROC was 81.9 ± 5.2, with an F1 score of 80.4 ± 7.7. In the hybrid model, the baseline VA was the most important feature (weight = 59.1 ± 6.9%), while the weight of HD-OCT prediction was 9.6 ± 4.2%. Conclusions Both the clinical data and HD-OCT models can predict postoperative VA improvement in patients undergoing vitrectomy for a MH with good discriminative performances. Combining them into a hybrid model did not significantly improve performance. Translational Relevance OCT-based DL models can predict postoperative VA improvement following vitrectomy for MH but fusing those models with clinical data might not provide improved predictive performance.

Self-supervised contrastive learning improves machine learning discrimination of full thickness macular holes from epiretinal membranes in retinal OCT scans

“…The CNN models in this study also present a good starting point for fine-tuning other OCT-related diagnostic and prognostic questions. Prior works addressed whether the outcome of FTMH-corrective surgery can be predicted, with only moderate success [18, 32, 33]. A natural step would be to use the RaSCL approach to develop models for accurate surgical prognosis, which could have clinical benefit by helping to avoid unnecessary surgeries.…”

“…Deep learning based OCT image analysis for FTMH has also received attention lately, with models for classification [16,17], segmentation [23][24][25][26], and prognosis of success for FTMH corrective surgery [18,32,33]. A review is also available [34].…”

Application of artificial intelligence system for screening multiple fundus diseases in Chinese primary healthcare settings: a real-world, multicentre and cross-sectional study of 4795 cases

“… 19 Moreover, AI revealed high performances in predicting postoperative visual prognosis after ERM 20 and MH surgeries. 21 Furthermore, as an accessible extension of the brain and the only window for observing small vessels in vivo , 22 the retina can detect Alzheimer’s disease 23 and cardiovascular diseases 24 in addition to fundus diseases. However, significant gaps exist between developing and applying AI systems in clinical practice, 25 and only a few reports have validated AI systems in real-world scenarios.…”

Deep Learning Using Preoperative Optical Coherence Tomography Images to Predict Visual Acuity Following Surgery for Idiopathic Full-Thickness Macular Holes

“…More recently, some DL approaches have improved the prediction of VA outcomes [23] using OCT data [24]- [26]. In particular, convolutional neural network (CNN) models have achieved high performance in OCT image analysis studies; however, there have only been a limited number of studies investigating VA measurements [23], [24], [27]. Considering the success of prominent CNN-based networks in medicine [28]- [30], they used a ResNet [31] in the [23], VGG [32] in the [33], and CBR-Tiny models [34] in the [27] as a backbone.…”