Background/Objectives: Neovascular age-related macular degeneration (nAMD) is a retinal disorder leading to irreversible central vision loss. The pro-re-nata (PRN) treatment for nAMD involves frequent intravitreal injections of anti-VEGF medications, placing a burden on patients and healthcare systems. Predicting injections needs at each monitoring session could optimize treatment outcomes and reduce unnecessary interventions. Methods: To achieve these aims, machine learning (ML) models were evaluated using different combinations of clinical variables, including retinal thickness and volume, best-corrected visual acuity, and features derived from macular optical coherence tomography (OCT). A “Leave Some Subjects Out” (LSSO) nested cross-validation approach ensured robust evaluation. Moreover, the SHapley Additive exPlanations (SHAP) analysis was employed to quantify the contribution of each feature to model predictions. Results: Results demonstrated that models incorporating both structural and functional features achieved high classification accuracy in predicting injection necessity (AUC = 0.747 ± 0.046, MCC = 0.541 ± 0.073). Moreover, the explainability analysis identified as key predictors both subretinal and intraretinal fluid, alongside central retinal thickness. Conclusions: These findings suggest that session-by-session prediction of injection needs in nAMD patients is feasible, even without processing the entire OCT image. The proposed ML framework has the potential to be integrated into routine clinical workflows, thereby optimizing nAMD therapeutic management.

Neovascular age-related macular degeneration (nAMD) is one of the major causes of vision impairment that affect millions of people worldwide. Early detection of nAMD is crucial because, if untreated, it can lead to blindness. Software and algorithms that utilize artificial intelligence (AI) have become valuable tools for early detection, assisting doctors in diagnosing and facilitating differential diagnosis. AI is particularly important for remote or isolated communities, as it allows patients to endure tests and receive rapid initial diagnoses without the necessity of extensive travel and long wait times for medical consultations. Similarly, AI is notable also in big hubs because cutting-edge technologies and networking help and speed processes such as detection, diagnosis, and follow-up times. The automatic detection of retinal changes might be optimized by AI, allowing one to choose the most effective treatment for nAMD. The complex retinal tissue is well-suited for scanning and easily accessible by modern AI-assisted multi-imaging techniques. AI enables us to enhance patient management by effectively evaluating extensive data, facilitating timely diagnosis and long-term prognosis. Novel applications of AI to nAMD have focused on image analysis, specifically for the automated segmentation, extraction, and quantification of imaging-based features included within optical coherence tomography (OCT) pictures. To date, we cannot state that AI could accurately forecast the therapy that would be necessary for a single patient to achieve the best visual outcome. A small number of large datasets with high-quality OCT, lack of data about alternative treatment strategies, and absence of OCT standards are the challenges for the development of AI models for nAMD.

Background and Objectives: Here, we report the anatomical and functional outcomes of Pars Plana Vitrectomy (PPV) with feeder vessel ligation, with or without endoresection in cases of retinal detachment (RD) secondary to retinal capillary hemangioblastoma (RCH). Materials and Methods: This retrospective observational study included 12 eyes with RD secondary to RCH. Based on the location of the lesion and the features of the RD, eyes were divided into two groups. Seven eyes with RCH located in Zone 2 or Zone 3, associated with tractional retinal detachment (TRD), underwent PPV with feeder vessel ligation and tumor endoresection. Five eyes, either with RCH in Zone 2 or Zone 3 associated with exudative retinal detachment or with RCH in Zone 1 associated with RD, underwent PPV with feeder vessel ligation alone, without tumor endoresection. Outcome measures included local tumor control, best-corrected visual acuity (BCVA), anatomical success, and rates of complications. Results: RCH regressed completely in 100% of eyes with no evidence of recurrence. The mean follow-up was 4.6 years. In the endoresection group, the mean BCVA was 2.18 ± 0.3 logMAR at baseline and 0.95 ± 0.5 logMAR after surgery (p = 0.018), whereas in the second group, the baseline mean BCVA was 1.33 ± 0.2 logMAR and 1.52 ± 0.7 logMAR postoperatively. In the first group, retinal attachment was achieved in all eyes, whereas in the second group, two eyes presented with persistent RD and proliferative vitreoretinopathy (PVR). No cases of phthisis bulbi or neovascular glaucoma were observed. Conclusions: PPV combined with feeder vessel ligation and endoresection appears to be an effective treatment for TRD secondary to RCH located in Zones 2 and 3, providing satisfactory anatomical and visual outcomes considering the severity of the disease. In cases where tumor location precludes endoresection, PPV with feeder vessel ligation alone may still be a viable option, although the potential risk of PVR could persist.

Purpose: The purpose of this paper is to report the clinical manifestations, diagnostic evaluation, management and outcomes of microsporidial keratitis in post-keratoplasty eyes. Methods: This is a retrospective review of three patients diagnosed with microsporidial stromal keratitis in post-keratoplasty eyes between January 2012 and December 2021 at a tertiary referral center (Ospedali Privati Forlì “Villa Igea”, Forlì, Italy). Results: All patients presented with fine multifocal granular infiltrates following keratoplasty for a presumed herpetic keratitis. No microorganisms were isolated from the corneal scrapings and no clinical response was observed with broad-spectrum antimicrobial therapy. In all cases, confocal microscopy demonstrated spore-like structures. The histopathologic examination of the excised corneal buttons confirmed the diagnosis of microsporidial stromal keratitis. Following therapeutic keratoplasty and treatment with an initial high dose and extended taper of topical fumagillin, clinical resolution was achieved in all eyes. The Snellen visual acuities at the final follow-up were 20/50, 20/63 and 20/32. Conclusions: Prior to definitive surgery, confocal microscopy can be employed for the in vivo detection of pathogenic microorganisms such as Microsporidium. In post-keratoplasty eyes, therapeutic keratoplasty and an initial high dose of topical fumagillin with extended taper can allow the resolution of microsporidial stromal keratitis with a satisfactory visual prognosis.