Management of Patients With DME With Good Visual Acuity in Routine Clinical Practice.

BACKGROUND AND OBJECTIVE: To determine outcomes of eyes with diabetic macular edema (DME) and best visual acuity (BVA) of 20/25 or better in routine clinical practice. PATIENTS AND METHODS: Retrospective study of 72 patients with DME and BVA of 20/25 or better. Patients were divided by anti-vascular endothelial growth factor (VEGF) treatment regimen: early (Group A), delayed (Group B), and none (Group C). RESULTS: Group A had higher baseline central subfield thickness (CST) (325 ± 62 µm) compared to Groups B (292 ± 24 µm) and C (296 ± 35 µm) (P = .033). All groups had similar 24-month CST (299 ± 62 µm, 280 ± 64 µm, 296 ± 65 µm; P = .61). There was no difference in baseline BVA among groups (81.9 ± 2.4, 83.2 ± 2.4, 82.4 ± 2.5 Early Treatment Diabetic Retinopathy Study [ETDRS] letters, respectively; P = .290), but at 6 months, Group A had lower BVA (76.6 ± 9.6 ETDRS letters) than groups B (81.9 ±3.3 ETDRS letters) and C (82.4 ± 5.0 ETDRS letters) (P = .008). There was no difference among groups in 24-month BVA (78.9 ± 6.6, 78.4 ± 12.3, and 80.6 ± 6.9 ETDRS letters, respectively; P = .448). CONCLUSION: Although observation may be indicated in eyes with stable BVA and CST less than 300 µm, anti-VEGF stabilizes BVA in eyes with CST greater than 300 µm and eyes with declining BVA. [Ophthalmic Surg Lasers Imaging Retina. 2021;52:247-256.].

Student becomes teacher: training faster deep learning lightweight networks for automated identification of optical coherence tomography B-scans of interest using a student-teacher framework.

This work explores a student-teacher framework that leverages unlabeled images to train lightweight deep learning models with fewer parameters to perform fast automated detection of optical coherence tomography B-scans of interest. Twenty-seven lightweight models (LWMs) from four families of models were trained on expert-labeled B-scans (∼70 K) as either "abnormal" or "normal", which established a baseline performance for the models. Then the LWMs were trained from random initialization using a student-teacher framework to incorporate a large number of unlabeled B-scans (∼500 K). A pre-trained ResNet50 model served as the teacher network. The ResNet50 teacher model achieved 96.0% validation accuracy and the validation accuracy achieved by the LWMs ranged from 89.6% to 95.1%. The best performing LWMs were 2.53 to 4.13 times faster than ResNet50 (0.109s to 0.178s vs. 0.452s). All LWMs benefitted from increasing the training set by including unlabeled B-scans in the student-teacher framework, with several models achieving validation accuracy of 96.0% or higher. The three best-performing models achieved comparable sensitivity and specificity in two hold-out test sets to the teacher network. We demonstrated the effectiveness of a student-teacher framework for training fast LWMs for automated B-scan of interest detection leveraging unlabeled, routinely-available data.

Optimizing Visualization of Membranes in Macular Surgery With Heads-Up Display.

BACKGROUND AND OBJECTIVE: To determine which optical parameter profiles (OPPs) can be utilized to improve the visualization of epiretinal membranes (ERMs) and the internal limiting membrane (ILM) using a three-dimensional heads-up microscope during 25-gauge pars plana vitrectomy. PATIENTS AND METHODS: Fourteen independent graders were asked to complete a questionnaire comparing each of the OPPs against the unaltered control image for each given surgical case. RESULTS: Analysis of the graders' responses indicated that higher values of hue are correlated with better visualization of ERM/ILM before and after dye application. There was overall agreement that OPPs could be used to enhance the visualization of the ERM and ILM during surgery. CONCLUSIONS: The use of OPPs to improve the visualization of specific structures is still new and heavily dependent on surgeon preference. The authors' study shows that some OPPs may enhance the visualization of the ERM and ILM during macular surgery. [Ophthalmic Surg Lasers Imaging Retina. 2020;51:584-587.].