Purpose: To develop an annotated fundus angiographic dataset, including fundus fluorescein angiography (FFA) and indocyanine green angiography (ICGA), and establish baseline methods for automatic report generation.
Methods: This retrospective study reviewed patients aged ≥18 years who underwent FFA or ICGA at Rajavithi Hospital, Thailand, between 1 January and 31 December 2019. A total of 55 361 de-identified images from 1691 patients (3179 eyes) were annotated by retinal specialists with detailed descriptions of the type, location, shape, size and pattern of abnormal fluorescence. Two baseline methods were developed: (1) a classification-based approach using ResNet101 with class-specific residual attention for multi-label lesion recognition and (2) a language-generation approach using the Bootstrapping Language-Image Pre-training framework, fine-tuned on angiographic images and structured reports. Model performances were evaluated using F1 score and BERTScore.
Results: The dataset includes 24 diagnostic conditions, with macular neovascularisation (32.5%) being the most prevalent, followed by unremarkable findings (21.8%) and dry age-related macular degeneration (10.2%). Most eyes (81.8%) underwent both FFA and ICGA. Hyperfluorescence was observed in 75.6% of cases, predominantly due to leakage, while hypofluorescence was present in 28.1%. The classification-based method achieved an average score of 7.966, demonstrating superior performance in recognising choroidal neovascularisation, hyperfluorescent and hypofluorescent areas. The language-generation method achieved a comparable average score of 7.947, excelling in impression recognition and the hyperfluorescence identification.
Conclusion: We present the largest annotated fundus angiographic dataset to date, along with two effective baseline methods for automatic report generation, offering a valuable foundation for advancing artificial intelligence applications in ophthalmology.
Objectives: This study aimed to compare the efficacy of different interventions for myopia control in children, including 0.01% atropine (AP), orthokeratology (Ortho-k), and repeated low-level red-light therapy (RLRL), and their combinations by conducting a network meta-analysis.
Methods: We searched for randomised controlled trials (RCTs) in PubMed, Web of Science and Embase. The primary outcomes were the mean changes in the cycloplegic spherical equivalent (SE) and axial length (AL) at the 12-month follow-up. A Bayesian random-effects network meta-analysis was performed to estimate pooled weighted mean differences and 95% credible intervals.
Results: The analysis included 41 RCTs with 6434 eyes. Compared with the control group, all interventions were found to be effective at slowing myopia progression, combining direct and indirect evidence at the 12-month follow-up: RLRL therapy (AL -0.31 (0.39, 0.24), p<0.05; SE 0.76 (0.54, 0.98), p<0.05), 0.01% atropine (AL -0.13 (-0.20, 0.07), p<0.05; SE 0.25 (0.08, 0.42), p<0.05), Ortho-k therapy (AL -0.16 (-0.26, 0.06), p<0.05; SE 0.58 (0.05, 1.13), p<0.05) and 0.01% atropine+Ortho-k therapy (AL -0.27 (-0.38, 0.16), p<0.05; SE 0.76 (0.23, 1.31), p<0.05). The cumulative probability ranking suggested that RLRL therapy was the most effective intervention in slowing AL, followed by 0.01% atropine+Ortho-k, Ortho-k and 0.01% atropine.
Conclusions: This network meta-analysis provides evidence that RLRL, 0.01% atropine, Ortho-k and 0.01% atropine+Ortho-k are all effective in suppressing myopia progress. In terms of long-term treatment efficacy in slowing AL and SE procession, RLRL was the most effective intervention.
Background/aims: Circumscribed choroidal haemangioma (CCH) can cause visual impairment, primarily managed with photodynamic therapy (PDT). This study aimed to identify predictive factors of therapeutic response in patients with CCH treated with PDT and to analyse serial anatomical outcomes, focusing on visual acuity, tumour characteristics, and inner retinal microstructures.
Methods: This retrospective study included patients diagnosed with CCH at Severance Eye Hospital, Yonsei University, from January 2005 to December 2022. We reviewed patient records and multimodal imaging, assessing demographics, including post-PDT visual acuity changes, age, laterality and sex, and multimodal imaging features such as tumour location and height, presence of subfoveal subretinal fluid (SRF), enhanced reflectivity of the retinal pigment epithelium and external limiting membrane (ELM) disruption.
Results: Among 114 patients with CCH (82 men, 65.6%; mean age 52.1±14.0 years), 45 (39.5%) had asymptomatic CCH, while 69 (60.5%) presented with symptomatic CCH. PDT was administered to 41 symptomatic patients (59.4%). The pretreatment LogMAR visual acuity was 0.50±0.63, which remained stable 1 year after PDT (0.53±0.64, p=0.666). However, it worsened 3 years after PDT (0.81±0.89, p=0.024), around the time tumour regrowth was observed. At 1-year post-PDT, tumour volume and height decreased by an average of 34.6% and 32.4%, respectively, accompanied by a 75.0% reduction in SRF. Patients with >75% vol reduction showed greater SRF persistence, while a higher initial SRF height was associated with complete SRF absorption following PDT. ELM disruption was identified as a significant predictor of poorer visual acuity, reduced tumour volume response and limited SRF absorption.
Conclusions: The presence of SRF predicts favourable tumour size reduction and SRF absorption post-PDT, while ELM disruption serves as an important prognostic factor for visual acuity and structural outcomes, aiding in pre-PDT treatment planning.
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：