Usher syndrome is an autosomal recessive disorder characterized by retinitis pigmentosa (RP), sensorineural hearing loss, and vestibular dysfunction. Polypoidal choroidal vasculopathy (PCV) is characterized by branching neovascular networks and polypoidal lesions. While typically classified within the pachychoroid spectrum, PCV can also manifest in eyes without choroidal thickening. Conversely, RP is usually associated with choroidal thinning, which may lead to underdiagnosis of PCV. To the best of our knowledge, PCV has not previously been reported in patients with Usher syndrome. Here, we present the first genetically confirmed case of Usher syndrome type 2 complicated by PCV that was successfully managed with intravitreal faricimab. A 63-year-old man presented with night blindness. The diagnosis of Usher syndrome type 2 was confirmed based on bilateral RP features, moderate sensorineural hearing loss, and two pathogenic USH2A variants. Six years after the initial diagnosis, the patient developed visual disturbance in the left eye, with the best-corrected visual acuity (BCVA) declining to 20/80. Multimodal imaging, including spectral-domain optical coherence tomography, fluorescein angiography, indocyanine green angiography, and swept-source optical coherence tomography angiography, confirmed the diagnosis of PCV. An amelanotic choroidal nevus was also identified on SD-OCT. After five intravitreal faricimab injections, the BCVA improved to 20/30, with a 75% reduction in pigment epithelial detachment height and complete subretinal fluid resolution without adverse effects. To our knowledge, this is the first reported case of PCV associated with a choroidal nevus in a patient with genetically confirmed Usher syndrome type 2. PCV should not be excluded in patients with RP despite characteristic choroidal thinning. Intravitreal faricimab was effective and well-tolerated, suggesting its therapeutic potential in such cases.

To characterize the clinical spectrum and age-distribution of pseudomelanomas in patients referred for suspected uveal melanoma at the largest ocular-oncology center in South Korea, where uveal melanoma incidence is the lowest worldwide. A retrospective review was conducted of consecutive patients referred for suspected uveal melanoma between January 2020 and August 2025. Diagnoses were confirmed using multimodal imaging. Patients were stratified into five groups (0-20, 21-40, 41-60, 61-80, and 81-100 years) to identify age-specific diagnostic patterns. Of 687 patients, 564 (82.1%) had pseudomelanomas and 123 (17.9%) had clinically confirmed melanoma, 81 of which had cytologic/pathologic confirmation. Common pseudomelanomas were choroidal nevus (N=120, 21.3%), congenital hypertrophy of the retinal pigment epithelium (CHRPE, N=70, 12.4%), choroidal hemangioma (N=53, 9.4%), choroidal metastasis (N=39, 6.9%), optic disc melanocytoma (N=35, 6.2%), and retinal capillary hemangioblastoma (N=32, 5.7%). Rare entities included vortex vein varix (N=7, 1.2%), posterior nodular scleritis (N=6, 1.1%), choroidal lymphoma (N=3, 0.5%), degenerative retinoschisis (N=2, 0.4%), RPE adenoma, uveal effusion, ciliary body schwannoma, ciliary body leiomyoma (N=1, 0.2%, respectively), etc. Uveal melanoma was absent in those aged 0-20 years and predominated in adults aged 41-80 years, along with nevus and metastasis, while CHRPE, retinal capillary hemangioblastoma, and choroidal osteoma were more common in ≤40 years. Peripheral exudative hemorrhagic chorioretinopathy was predominant in ≥81 years, while primary vitreoretinal lymphoma occurred in ≥61 years. Pseudomelanomas accounted for over 80% of suspected uveal melanoma cases in this Korean cohort, with age-specific distribution patterns. Recognizing pseudomelanomas features and these patterns may enhance diagnostic accuracy.

PurposeEarly detection of uveal melanoma is important for improving patient survival. We developed a model for automated early detection of choroidal melanocytic lesions from fundus images.MethodsPatch embeddings were created for 15 million ophthalmic images of various modalities using a self-supervised model, resized, and stored as 384-dimensional image vectors. Latent embeddings were applied to all color and pseudocolor fundus images from patients with confirmed choroidal nevus or melanoma from the Prospective Ocular Tumor Study (POTS) dataset between July 2019 and July 2024. K-means clustering was applied to random color (1000) and pseudocolor (400) images. Melanocytic lesions were identified and patch-level embeddings extracted. Images were image input at 592 × 592 pixels, overlayed on the patch grid, and binary labels applied to each lesion-containing patch. A supervised classifier was trained to detect melanocytic lesions. The model was validated on images from patients with (250) and without (250) “nevus,” “nevi,” or “melanoma” referenced in their medical record. Model generalizability was assessed on image types (CLARUS) not in the original training dataset. Area under the receiver operating characteristic curve (AUC-ROC), sensitivity, and specificity were calculated.ResultsMelanocytic lesions were detected with high performance: AUC-ROC was 0.9856 for color fundus, 0.9040 for pseudocolor, and 0.9544 for CLARUS.ConclusionsWe created a reliable melanocytic lesion detection model without task-specific fine-tuning, achieving high accuracy, sensitivity, and specificity.Translational RelevanceBy combining self-supervised representation learning with lightweight classifiers, it is possible to create robust diagnostic tools for melanocytic lesion detection with minimal annotation requirements.

Objectives:Uveal melanoma (UM) is the most common primary intraocular malignancy in adults and carries significant metastatic risk. Early and accurate diagnosis is essential, but challenging due to overlapping clinical features with benign choroidal nevi. Deep learning (DL) offers potential to support early and accessible detection, but model performance is limited by dataset size and quality. This study evaluates data-centric optimisation strategies for DL classification of UM using ultra-widefield (UWF) fundus photography.Methods:This retrospective study analysed UWF fundus photographs from 784 patients (864 images) seen at the University of Illinois Chicago eye clinic. A baseline binary classification model (UM vs. choroidal nevus) was compared with seven models incorporating optimisation strategies across three categories: class addition, dataset augmentation, and enhanced feature selection. Performance was assessed using AUC, F1 score, precision, and recall, with calibration evaluated via expected calibration error.Results:The baseline model achieved an AUC of 0.906 ± 0.032. The top-performing model incorporated healthy retinal controls as an additional class, achieving an AUC of 0.987 ± 0.011 and F1 scores of 0.966 (nevus) and 0.941 (UM). Dataset augmentation approaches yielded minimal performance gain, and Multi-strategy models showed no additive benefit.Conclusions:Data-centric optimisation significantly influences DL performance for UM detection. Three principles emerge: healthy class addition improves specificity and anatomical feature learning; data quality may outweigh quantity; and contextual input tuning is a key model parameter. These findings offer a practical framework for developing clinically robust, physician-supervised AI tools to support early UM triage and reduce diagnostic variability.

Eventual Atrophy-Associated Vision Loss from a Choroidal Nevus: 13 Years of Follow-Up.

Introduction: Choroidal melanoma is the most common primary intraocular malignancy in adults. Early detection through ophthalmic screening increases the likelihood of globe-preserving treatment and may reduce metastasis while maximizing survival. Racial and ethnic disparities in access to care may influence outcomes, metastasis, and mortality. Methods: This retrospective cohort study used data from 2004 to 2025 within a federated electronic health record database. Adult patients (≥18 years) with choroidal melanoma were identified. Race/ethnicity was categorized as non-Hispanic white (NHW) or Black/Hispanic. The primary outcomes were primary enucleation, development of liver metastasis, and all-cause mortality within 1, 3, or 5 years from the time of choroidal melanoma diagnosis. Propensity score matching (PSM) was performed to balance cohorts. Multivariate Cox proportional hazards models were used to identify factors associated with enucleation, metastasis, and mortality. Results: A total of 17,436 patients with choroidal melanoma were identified, of whom 659 (6.3%) were Black/Hispanic and 9,883 (93.7%) were NHW. After PSM, primary enucleation within 90 days of diagnosis occurred in 5.2% of Black/Hispanic patients versus 3.8% of NHW patients (risk ratio [RR] 1.36, 95% confidence interval [CI] 0.821–2.253, p = 0.2305). Cumulative rates of liver metastasis in the Black/Hispanic group were significantly higher than NHW at 1 year (RR 2.42, CI: 1.336–4.387), 3 years (RR 1.96, CI: 1.246–3.083), and 5 years (RR 1.77, CI: 1.195–2.611) (p < 0.01 for each). All-cause mortality was higher in the Black/Hispanic cohort at 1 year (RR 1.96, CI: 1.212–3.163), 3 years (RR 1.75, CI: 1.249–2.453), and 5 years (RR 1.48, CI: 1.105–1.976) (p < 0.01 for each). On multivariate Cox analysis, male sex (hazard ratio [HR] 1.49, CI: 1.275–1.736, p < 0.0001) was associated with higher risk of enucleation, while a history of choroidal nevus was protective (HR 0.59, CI: 0.424–0.831, p = 0.0024). For metastatic risk, Black/Hispanic race/ethnicity (HR 1.54, CI: 1.242–1.904) and enucleation (HR 3.27, CI: 1.845–5.780) were associated with an increased risk, whereas prior nevus was protective (HR 0.43, CI: 0.324–0.577). All-cause mortality was elevated with older age (HR 1.03, CI: 1.021–1.031), male sex (HR 1.22, CI: 1.084–1.373), Black/Hispanic race/ethnicity (HR 1.42, CI: 1.135–1.769), metastatic liver disease (HR 7.49, CI: 6.051–9.269), among other comorbidities. Prior nevus history conferred a survival benefit (HR 0.59, CI: 0.458–0.762). Conclusion: In patients with newly diagnosed choroidal melanoma, Black/Hispanic patients had higher rates of metastatic disease and mortality. Previous history of choroidal nevus was associated with a lower risk of primary enucleation, metastatic disease, and mortality.

Choroidal tumors comprise a heterogeneous group of intraocular lesions ranging from benign entities such as choroidal nevus and circumscribed choroidal hemangioma to malignant tumors including uveal melanoma and choroidal metastasis. Accurate differentiation is crucial, as management and prognosis exhibit significant variations.Choroidal nevi are the most prevalent benign intraocular tumors and are generally asymptomatic. However, a small subset of these nevi carries a risk of malignant transformation into melanoma. Risk stratification utilizing multimodal imaging criteria, such as the TFSOM-DIM and MOLES systems, is pivotal in clinical decision-making.Uveal melanoma represents the most common primary intraocular malignancy in adults and is associated with high metastasis rates, particularly to the liver. Prognosis depends on tumor size, location, histopathologic features, and molecular alterations such as monosomy 3 or BAP1 loss. Treatment options range from globe-preserving radiotherapy to adjuvant or neoadjuvant surgical resection and enucleation in advanced cases.Choroidal hemangioma is a benign vascular tumor that can lead to visual impairment due to exudative retinal detachment. Photodynamic therapy is currently considered the preferred treatment in symptomatic cases.Choroidal metastases represent the most prevalent intraocular malignancies, with a primary origin in breast or lung carcinoma. Multimodal imaging is essential for diagnosis and monitoring, and management requires interdisciplinary coordination.A structured multimodal diagnostic approach is crucial for accurate classification, risk assessment, and individualized therapeutic planning in patients with choroidal tumors.

Objective: To describe and compare the internal vascular features of choroidal melanoma (CM) and choroidal nevus on optical coherence tomography angiography (OCTA). Methods: This cross-sectional study included patients clinically diagnosed with CM or choroidal nevus at Beijing Tongren Hospital between July 2019 and March 2025. Tumor-centered OCTA images were acquired, and the intrinsic tumor vasculature was reconstructed on two en face planes, the choriocapillaris slab and the full-thickness choroid slab. The malignant transformation risk of choroidal nevus was assessed using the multimodal imaging-based TFSOM-DIM criteria. Nevi with two or more high-risk features were classified as high-risk nevi, and the OCTA features of high and low-risk nevi was analyzed. The OCTA features were compared between the CM and nevus groups using the χ² test or Fisher's exact test. Results: A total of 95 eyes (95 patients) were included, consisting of 49 right eyes and 46 left eyes. There were 45 males (47.4%) and 50 females (52.6%), with a mean age of (47.6±14.1) years. All patients had unilateral involvement. The CM group included 67 patients (67 eyes), with 35 right eyes and 32 left eyes, comprising 36 males (53.7%) and 31 females (46.3%), with a mean age of (47.1±14.3) years. The choroidal nevus group included 28 patients (28 eyes), with 14 right eyes and 14 left eyes, consisting of 9 males (32.1%) and 19 females (67.9%), with a mean age of 48.8±13.6 years. There were no significant differences between the two groups in eye laterality (χ²=0.04, P=0.842), age (t=0.53, P=0.299), or sex distribution (χ²=3.69, P=0.055). In the choroid en face slab, disorganized vascular networks and vascular loops were observed in 66 CM cases (98.5%) and 14 nevus cases (50.0%; χ²=34.95, P<0.001). In the choriocapillaris en face slab, patchy avascular areas were detected in 54 CM cases (80.6%) and 9 nevi (32.1%; χ²=20.79, P<0.001), while the intrinsic tumor vasculature was present in 52 CM cases (77.6%) and 8 nevi (28.6%; χ²=20.41, P<0.001). Two distinct vascular patterns were identified within choroidal nevi on the choroid slab: 14 nevi (50.0%) exhibited diffusely attenuated flow signals without identifiable tumor vasculature, whereas the other 14 nevi (50.0%) demonstrated melanoma-like disorganized vascular networks. Among 17 low-risk and 11 high-risk nevi, melanoma-like disorganized vascular networks were present in 9 low-risk nevi (29.4%) and 5 high-risk nevi (81.8%), with a statistically significant difference (χ²=7.34, P=0.018). Conclusion: In OCTA images, CMs and choroidal nevi exhibit different internal vascular networks. CMs typically exhibit disorganized vasculature with vascular networks and loops, whereas choroidal nevi more often show attenuated flow signals or finer, more orderly vascular structures. The presence of melanoma-like disorganized vascular networks in choroidal nevi may significantly predict malignant transformation and serve as an imaging indicator for identifying high-risk nevi.

The purpose of the study is to report a case of a rhegmatogenous retinal detachment (RRD) associated with a retinal tear overlying a choroidal nevus. This study is a case report. A 66-year-old female patient has been referred to our department because of a retinal detachment in her left eye. Her ocular history was notable for a choroidal nevus in her left eye, which has been under observation for the past 11 years without any signs of malignancy or abnormal changes. Dilated funduscopic examination revealed a superior macula-off bullous RRD that originated from a retinal tear located at the previously observed choroidal nevus. The patient successfully underwent a 23-g vitrectomy and cryotherapy with SF6 gas tamponade, combined with cataract surgery and biopsy. While choroidal nevi are relatively prevalent, the specific progression to RRD due to a tear is a rare occurrence in clinical practice, and it presents with unique diagnostic and therapeutic challenges.

Choroidal neovascularisation associated with a choroidal nevus: Multimodal imaging.

IntroductionThis study aimed to evaluate choroidal nevi (CN) volume as assessed by widefield optical coherence tomography (WF-OCT) and by ultrasound (US)-based mathematical approximations.MethodsIn a prospective, exploratory study, CN of consecutive patients were imaged with WF-OCT (VG 200D, Intalight Inc., San Jose, California, USA) and US (Quantel Absolu, Cournon, D’Auvergne Cedex, France). Volume was assessed by CN annotation on 64 B-scans on WF-OCT and two approximations based on US measurements (thickness, largest and perpendicular basal diameter (LBD; PBD)): (a) π/6 × thickness × LBD × PBD and (b) 2π/3 × [(LBD + PBD)/4]2 × thickness.ResultsEighteen nevi of 18 patients were included in the study. Mean volume evaluated on WF-OCT was 15.1 ± 24.1 mm3. US approximations resulted in (a) 23.0 ± 35.0 mm3 and (b) 23.5 ± 35.5 mm3. Intraclass correlation coefficients showed very good agreement between WF-OCT and US measurements (a) of 0.947 (CI 0.860–0.980) and (b) of 0.945 (CI 0.853–0.979). Stepwise backward elimination multivariable linear regression showed an association between difference of volume measurements and lesion thickness and LBD for approximations (a) and (b).ConclusionWF-OCT and US-based approximations show good interdevice reliability for CN volume in this exploratory study. Thickness and LBD have significant impact on discrepancy of volume measurements with a tendency that small CN are underestimated and large CN are overestimated with US. This potentially limits interchangeable use of US and WF-OCT.Supplementary InformationThe online version contains supplementary material available at 10.1007/s40123-025-01271-3.

Assessment of choroidal melanoma and nevus lesions using ultrasound vibro-elastography and parametric imaging approach.

Uveal melanomas are almost always unilateral. Bilateral presentation is extremely rare, about 1 in 80 million/year. This report presents the case of a 40-year-old man who developed bilateral choroidal melanoma sequentially. The right eye with a de novo juxtapapillary choroidal melanoma was managed with 18-mm notched I125 plaque with full distribution. Eighteen months later, transformation of a choroidal nevus into a melanoma was detected in the fellow eye, and was treated with 12-mm round I125 plaque with full distribution. The patient tested positive for BAP1 germline mutation (c.659+1G >A p.?; rs2153227622). After three years of follow-up, both tumors were regressed and the patient was receiving immunotherapy for related liver metastasis.

PurposeUveal melanoma (UM) is the most common intraocular malignancy in adults, with high metastatic risk and poor prognosis. Current screening and triaging methods for melanocytic choroidal tumors face inherent limitations, particularly in regions with limited access to specialized ocular oncologists. This study proposes a reference framework for future research in UM detection. It highlights the trade-offs between different computer vision (CV) approaches based on data availability, computer resources, annotation effort, and clinical applicability.MethodsIn total, 864 Optos images of UM, choroidal nevi, and congenital hypertrophy of the retinal pigment epithelium were included in the study. Three CV models—classification, detection, and segmentation—were implemented using a shared ResNet-50 backbone. Performance metrics included area under the curve (AUC) for classification, F1 score for detection, and Dice score for segmentation. An ablation study evaluated robustness to data scarcity. Model interpretability was enhanced through Grad-CAM visualizations and confusion matrices.ResultsClassification, detection, and segmentation achieved AUC scores of 94%, 93%, and 95%, respectively. Segmentation showed the best F1 score (89%) and a Dice score of 75%. Classification performed the best with limited data.ConclusionsIn high-resource scenarios (100+ images), all models performed similarly, while in low-resource scenarios (<70 images), the classification model outperformed the others. This suggests that simpler models may offer better value for classification tasks with limited resources.Translational RelevanceGiven different CV algorithms’ different clinical use cases and development costs, this study takes a comparative look at these algorithms for the task of UM analysis on ultra-widefield fundus images.

Attention-Based Multimodal Deep Learning for Uveal Melanoma Classification Using Ultra-Widefield Fundus Images and Ocular Ultrasound

To evaluate the diagnostic and treatment recommendation performance of multimodal large language models (MLLMs) in identifying and classifying retinal lesions as choroidal nevus or melanoma, as well as compare their performance with expert human graders. This retrospective cross-sectional study included 48 eyes from 47 patients diagnosed with either choroidal nevus or melanoma. Patient demographics, including age, sex, ethnicity, best-corrected visual acuity (BCVA), and symptoms, were documented. Color fundus, autofluorescence, optical coherence tomography, and B-scan images were collected. The ocular images and patient characteristics were presented to ChatGPT 4.0, Gemini Advanced 1.5 Pro, and Perplexity Pro. Responses were recorded and compared with the clinical diagnoses and treatment recommendations made by two expert human graders. Diagnostic and treatment agreement, accuracy, sensitivity, and specificity were analyzed. Gemini consistently outperformed ChatGPT and Perplexity across diagnostic and treatment prompts. The highest model performance was observed for prompts requesting treatment recommendations with clinical information, where Gemini achieved the highest accuracy (0.725), followed by Perplexity (0.647) and ChatGPT (0.314). Performance was lowest for prompts requiring strict clinical criteria, with all models showing poor sensitivity. Both human graders outperformed all MLLMs in accuracy and sensitivity on most prompts (P < 0.005). Accuracy did not improve when provided demographic or clinical data, except for Gemini. Human graders outperform current MLLMs, which show only moderate ability to diagnose choroidal nevi or melanoma from imaging. This study highlights limitations and potential of MLLMs in aiding diagnosis and treatment of choroidal lesions.

PurposeTo evaluate the proportion of patients with choroidal nevus in an ocular oncology practice who would be suitable candidates for monitoring via a virtual photoscreening model, wherein imaging is asynchronously reviewed and longitudinally managed by an ocular oncologist.MethodsData were abstracted from the Prospective Ocular Tumor Study Database for all patients diagnosed with choroidal nevus from July 2019-October 2024. A stratification model for identification of virtual pathway candidates was defined and implemented as follows: good virtual pathway candidates included those with posteriorly located lesions for adequate photographic capture and no more than two high-risk features. Risk factors, outlined by the TFSOM-DIM acronym, included thickness >2 mm, presence of subretinal fluid, visual acuity 20/50 or worse, presence of orange pigment, low internal reflectivity, and basal diameter >5 mm. Of posteriorly located lesions, those with 0-2 risk factors were defined as good virtual pathway candidates, 3 risk factors as acceptable in select cases, and 4 or more as poor candidates.ResultsThere were 614 choroidal nevi in 586 patients diagnosed during the study period. Mean patient age at initial visit was 63.5 years old (median 66.1, range 5.8-94.8), with 59.4% female and 98.8% white. Analysis identified 436 (71%) nevi in 408 (69.6%) patients as good virtual pathway candidates, 53 (8.6%) nevi in 53 (9%) patients as potential candidates in select cases, and 125 (20.4%) nevi in 125 (21.3%) patients as poor candidates.ConclusionNearly 70% of patients with choroidal nevi followed by Mayo Clinic Ocular Oncology were identified as good candidates for follow-up via virtual photoscreening. This highlights the potential for improved clinical efficiency through virtual frameworks in ocular oncology.

PurposeDifferentiation between choroidal melanoma (CM) and choroidal nevus (CN) is primarily made clinically with the aid of multimodal imaging. Both entities can be distinguished by their vascular architecture in histopathology. OCT angiography (OCT-A) is a noninvasive method to visualize retinal and choroidal vasculature and has rarely been used to differentiate CM and CN, and if so, exclusively descriptively. The purpose of this study was to differentiate both entities via OCT-A using an open-source vascular analysis software (OCTAVA), using quantitative choroidal vasculature analysis.DesignRetrospective case-control study.SubjectsTreatment-naïve patients with CM and CN who obtained OCT-A of the tumor at the University Hospital Essen between November 2022 and December 2024.MethodsChoroidal OCT-A sections were analyzed using an ImageJ-based open-source software (OCTAVA).Main Outcome MeasuresTo assess the comparison between CM and CN using vessel area density, vessel length density, total vessel length, mean vessel diameter, and tortuosity index.ResultsThirty patients with CM and 30 patients with CN could be included in the study. Choroidal melanoma showed significantly reduced vessel area density (30.46% vs. 37.10%; P < 0.001), vessel length density (1.83% vs. 2.40%; P < 0.001), total vessel length (11.8 vs. 19.7 mm; P < 0.001), and tortuosity index (0.140 vs. 0.155; P = 0.008) compared with CN, whereas vessel diameter showed no significant difference. Receiver operating curve analysis demonstrated good diagnostic performance for vessel area density, vessel length density, and total vessel length (area under the curve: 0.81–0.86).ConclusionsOCT angiography using open-source software quantitative choroidal vasculature analysis may play a future role in objective differentiation between CM and CN. Future role of OCT-A in multimodal imaging for melanocytic choroidal tumors must, however, be further validated in larger, prospective trials and across different, preferably commercially available OCT platforms.Financial Disclosure(s)The authors have no proprietary or commercial interest in any materials discussed in this article.

PurposeThis article explores the application of artificial intelligence (AI) in the differentiation of choroidal melanocytic lesions, specifically choroidal nevi and small melanomas, within the field of ocular oncology. The primary topic highlights the significance of accurately diagnosing these lesions to enhance patient outcomes and management strategies.DesignThe study reviews of the role of AI in differentiating choroidal melanocytic lesions, particularly choroidal nevi from small melanomas, examining clinical and imaging risk factors. It explores deep learning (DL) applications for image classification and assesses AI's potential impact on patient care, diagnostic accuracy, and regulatory concerns in ocular oncology.MethodsTo achieve this, the methods discussed in this paper revolve around employing DL techniques, which utilize artificial neural networks to analyze high-dimensional medical images. This approach enables automated classification and image analysis of ophthalmic data, allowing for the identification of intricate patterns and features that may be imperceptible to clinicians. Additionally, the text reviews existing clinical and imaging risk factors associated with the growth of choroidal nevi into melanoma, leveraging this information to inform and enhance AI algorithms.ResultsThe anticipated results of integrating AI into clinical practice include increased diagnostic accuracy, which can lead to earlier identification of high-risk lesions and, consequently, timely interventions. This proactive approach has the potential to improve patient care significantly by facilitating better management strategies, thus enhancing patient outcomes. Artificial intelligence may also uncover subtle imaging features that would otherwise be overlooked, providing a more comprehensive assessment of lesions.ConclusionIn conclusion, the paper emphasizes the transformative potential of AI in ocular oncology, advocating for its integration with existing imaging technologies. While AI offers promising advancements in diagnostic practices and patient care, the paper also acknowledges the necessity of addressing regulatory and implementation challenges to fully harness these benefits. Overall, the incorporation of AI technologies into the diagnostic workflow has the potential to not only save vision but also improve survival rates, marking a significant step forward in the management of choroidal melanocytic lesions.Financial Disclosure(s)Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Development of a deep learning model to classify choroidal melanoma risk factors based on color fundus photographs☆