Deep Learning for prediction of late recurrence of retinal detachment using preoperative and postoperative ultra-wide field imaging.

PURPOSE: To elaborate a deep learning (DL) model for automatic prediction of late recurrence (LR) of rhegmatogenous retinal detachment (RRD) using pseudocolor and fundus autofluorescence (AF) ultra-wide field (UWF) images obtained preoperatively and postoperatively. MATERIALS AND METHODS: We retrospectively included patients >18 years who underwent either scleral buckling (SB) or pars plana vitrectomy (PPV) for primary or recurrent RRD with a post-operative follow-up >2 years. Records of RRD recurrence between 6 weeks and 2 years after surgery served as a ground truth for the training of the deep learning (DL) models. Four separate DL models were trained to predict LR within the 2 postoperative years (binary outputs) using, respectively, UWF preoperative and postoperative pseudocolor images and UWF preoperative and postoperative AF images. RESULTS: A total of 412 eyes were included in the study (332 eyes treated with PPV and 80 eyes with SB). The mean follow-up was 4.0 ± 2.1 years. The DL models based on preoperative and postoperative pseudocolor UWF imaging predicted recurrence with 85.6% (sensitivity 86.7%, specificity 85.4%) and 90.2% accuracy (sensitivity 87.0%, specificity 90.8%) in PPV-treated eyes, and 87.0% (sensitivity 86.7%, specificity 87.0%) and 91.1% (sensitivity 88.2%, specificity 91.9%) in SB-treated eyes, respectively. The DL models using preoperative and postoperative AF-UWF imaging predicted recurrence with 87.6% (sensitivity 84.0% and specificity 88.3%) and 91.0% (sensitivity 88.9%, specificity 91.5%) accuracy in PPV eyes, and 86.5% (sensitivity 87.5%; specificity 86.2%) and 90.6% (sensitivity 90.0%, specificity 90.7%) in SB eyes, respectively. Among the risk factors detected with visualisation methods, potential novel ones were extensive laser retinopexy and asymmetric staphyloma. CONCLUSIONS: DL can accurately predict the LR of RRD based on UWF images (especially postoperative ones), which can help refine follow-up strategies. Saliency maps might provide further insight into the dynamics of RRD recurrence.

Deep learning for automatic prediction of early activation of treatment naïve non-exudative MNVs in AMD.

BACKGROUND: Around 30% of non-exudative macular neovascularizations(NE-MNVs) exudate within 2 years from diagnosis in patients with age-related macular degeneration(AMD).The aim of the study is to develop a deep learning classifier based on optical coherence tomography(OCT) and OCT angiography(OCTA) to identify NE-MNVs at risk of exudation. METHODS: AMD patients showing OCTA and fluorescein angiography (FA) documented NE-MNV with a 2-years minimum imaging follow-up were retrospectively selected. Patients showing OCT B-scan-documented MNV exudation within the first 2 years formed the EX-GROUP while the others formed QU-GROUP.ResNet-101, Inception-ResNet-v2 and DenseNet-201 were independently trained on OCTA and OCT B-scan images. Combinations of the 6 models were evaluated with major and soft voting techniques. RESULTS: Eighty-nine (89) eyes of 89 patients with a follow-up of 5.7 ± 1.5 years were recruited(35 EX GROUP and 54 QU GROUP). Inception-ResNet-v2 was the best performing among the 3 single convolutional neural networks(CNNs).The major voting model resulting from the association of the 3 different CNNs resulted in improvement of performance both for OCTA and OCT B-scan (both significantly higher than human graders' performance). Soft voting model resulting from the combination of OCTA and OCT B-scan based major voting models showed a testing accuracy of 94.4%. Peripheral arcades and large vessels on OCTA enface imaging were more prevalent in QU GROUP. CONCLUSIONS: Artificial intelligence shows high performances in identifications of NE-MNVs at risk for exudation within the first 2 years of follow up, allowing better customization of follow up timing and avoiding treatment delay. Better results are obtained with the combination of OCTA and OCT B-scan image analysis.

Artificial intelligence in age-related macular degeneration: state of the art and recent updates.

Age related macular degeneration (AMD) represents a leading cause of vision loss and it is expected to affect 288 million people by 2040. During the last decade, machine learning technologies have shown great potential to revolutionize clinical management of AMD and support research for a better understanding of the disease. The aim of this review is to provide a panoramic description of all the applications of AI to AMD management and screening that have been analyzed in recent past literature. Deep learning (DL) can be effectively used to diagnose AMD, to predict short term risk of exudation and need for injections within the next 2 years. Moreover, DL technology has the potential to customize anti-VEGF treatment choice with a higher accuracy than expert human experts. In addition, accurate prediction of VA response to treatment can be provided to the patients with the use of ML models, which could considerably increase patients' compliance to treatment in favorable cases. Lastly, AI, especially in the form of DL, can effectively predict conversion to GA in 12 months and also suggest new biomarkers of conversion with an innovative reverse engineering approach.

The impact of short-term postoperative face-up position on unintentional retinal displacement after pars plana vitrectomy for rhegmatogenous retinal detachment.

OBJECTIVE: To assess the effectiveness and safety of a rigorous short-term supine position in preventing inadvertent retinal displacement after pars plana vitrectomy (PPV) with gas tamponade for rhegmatogenous retinal detachment (RRD). METHODS: We conducted a retrospective observational analysis of a case series at two ophthalmological surgical centers. We included eyes diagnosed with macula-off RRD that maintained a strict face-up position for three hours immediately after PPV with intraoperative perfluorocarbon liquid (PFCL) and 20% sulfur hexafluoride (SF6) tamponade. Fundus autofluorescence (FAF) imaging was performed at one month post-operatively to identify unintentional retinal displacement through the detection of retinal vessel prints (RVPs) on FAF imaging using an ultrawide-field (UWF) imaging system. RESULTS: A total of 29 eyes with macula-off RRD were included in the study. The average age of the participants was 59.62 years. RRD involved one quadrant in two eyes, two quadrants in fourteen eyes, three quadrants in seven eyes, and four quadrants in six eyes. UWF-FAF imaging at one month follow-up after complete reattachment of the retina revealed RVPs in seven out of the 29 eyes (24.13%), with a mean displacement of 0.22 mm. In every case the displacement occurred downward. CONCLUSION: Our results suggest that adhering to a strict face-up position for three hours after PPV with PFCL and gas tamponade for macula-off RRD may lead to a low frequency and severity of inadvertent post-operative retinal displacement.

Ocular Involvement in Systemic Sclerosis: Updated Review and New Insights on Microvascular Impairment.

Systemic sclerosis (SSc) is a chronic multisystemic disease characterized by immunological activation, diffuse vasculopathy, and generalized fibrosis exhibiting a variety of symptoms. A recognized precursor of SSc is Raynaud's phenomenon, which is part of the very early disease of systemic sclerosis (VEDOSS) in combination with nailfold videocapillaroscopy (NVC) impairment. The pathophysiology of ocular involvement, alterations in internal organs, and body integumentary system involvement in SSc patients are complicated and poorly understood, with multiple mechanisms presumptively working together. The most prevalent ocular symptoms of SSc are abnormalities of the eyelids and conjunctiva as well as dry eye syndrome, due to fibroblasts' dysfunction and inflammation of the ocular surface. In particular, lagophthalmos, blepharophimosis limitation of eyelid motion, eyelid telangiectasia, and rigidity or tightening of the lids may affect up to two-third of the patients. In addition, reduction in central corneal thickness, iris defects and higher rates of glaucoma were reported. In the first reports based on retinography or fluorescein angiography, about 50% of SSc patients showed signs of vascular disease: peripheral artery occlusion, thinning of retinal pigment epithelium and choroidal capillaries, ischemic areas surrounded by intraretinal extravasation and microaneurysms, and peripheral capillary non-perfusion. Successively, thanks to the advent of optical coherence tomography angiography (OCTA), several studies highlighted significant impairment of either the choriocapillaris and retinal vascular plexuses, also correlating with NVC involvement and skin disease, even in VEDOSS disease. Given the sensitivity of this technique, ocular micro-vasculopathy may act as a tool for early SSc identification and discriminate between disease stages.

Evaluation of macular atrophy in patients treated with macular buckle for myopic traction maculopathy (MTM): mid and long-term follow-up.

PURPOSE: To evaluate the incidence, rate and pattern of progression of myopic maculopathy in eyes operated with macular buckle (MB) for myopic traction maculopathy (MTM) vs a control group without surgery to find out if the progression varies due to the macular buckle's indentation, and to evaluate possible MB-related pigmentary changes or atrophy. METHODS: Eyes operated with macular buckle with two good quality fundus images: one preoperative or early postoperative image and the second image of at least 12 months apart, the control group was comprised by the contralateral eyes. Demographics, axial length, follow-up, stage of MTM and myopic maculopathy were reported. Groups and subgroups (mid and long-term follow-up) progression results were reported and compared. RESULTS: We included 116 eyes of 66 patients. Progression was found in 29 eyes (41.4%) and 23 eyes (50%) in the macular buckle group and control group, respectively. The progression rate was 73 per 1000 eye-years and 88.9 per 1000 eye-years in the macular buckle group and the control group, respectively. Axial length was found a predictor of progression (OR,2.59; p=0.02). CONCLUSION: Progression of myopic maculopathy was similar in both groups, even mildly grater in the control group. We did not find any MB-related pigmentary changes or atrophy.

Exploring AI-chatbots' capability to suggest surgical planning in ophthalmology: ChatGPT versus Google Gemini analysis of retinal detachment cases.

BACKGROUND: We aimed to define the capability of three different publicly available large language models, Chat Generative Pretrained Transformer (ChatGPT-3.5), ChatGPT-4 and Google Gemini in analysing retinal detachment cases and suggesting the best possible surgical planning. METHODS: Analysis of 54 retinal detachments records entered into ChatGPT and Gemini's interfaces. After asking 'Specify what kind of surgical planning you would suggest and the eventual intraocular tamponade.' and collecting the given answers, we assessed the level of agreement with the common opinion of three expert vitreoretinal surgeons. Moreover, ChatGPT and Gemini answers were graded 1-5 (from poor to excellent quality), according to the Global Quality Score (GQS). RESULTS: After excluding 4 controversial cases, 50 cases were included. Overall, ChatGPT-3.5, ChatGPT-4 and Google Gemini surgical choices agreed with those of vitreoretinal surgeons in 40/50 (80%), 42/50 (84%) and 35/50 (70%) of cases. Google Gemini was not able to respond in five cases. Contingency analysis showed significant differences between ChatGPT-4 and Gemini (p=0.03). ChatGPT's GQS were 3.9±0.8 and 4.2±0.7 for versions 3.5 and 4, while Gemini scored 3.5±1.1. There was no statistical difference between the two ChatGPTs (p=0.22), while both outperformed Gemini scores (p=0.03 and p=0.002, respectively). The main source of error was endotamponade choice (14% for ChatGPT-3.5 and 4, and 12% for Google Gemini). Only ChatGPT-4 was able to suggest a combined phacovitrectomy approach. CONCLUSION: In conclusion, Google Gemini and ChatGPT evaluated vitreoretinal patients' records in a coherent manner, showing a good level of agreement with expert surgeons. According to the GQS, ChatGPT's recommendations were much more accurate and precise.

OCT angiography 2023 update: focus on diabetic retinopathy.

Optical coherence tomography angiography (OCTA) has become part of the clinical practice and its growing applications are in continuous development. Coherently with the growing concern about the human and economic cost of diabetes, diabetic retinopathy (DR) was the most popular topic for OCTA studies in the past year. The analysis of the literature reveals that applications of OCTA in DR are in continuous growth. In particular, ultrawide field (UWF) OCTA and artificial intelligence (AI) based on OCTA images are affirming as the new frontiers of scientific research in the field. Diagnostic accuracy of AI methods based on OCTA is equal or superior to the one based on OCT methods and also bears potential to detect systemic associations. UWF OCTA is noninvasive method that is reaching similar accuracy of FA in detection of neovascularization and intraretinal microvascular abnormalities (IRMAs) and has allowed better characterization of microvascular peripherical changes in DR. Lastly, deep capillary plexus (DCP) characteristics seem to play a pivotal role in the development of diabetic macular edema (DME) and refinement of biomarkers for different phenotypes of DME and diabetic macular ischemia (DMI) is currently on its way.

In vivo cone-photoreceptor density comparison between eyes with subretinal drusenoid deposits and healthy eyes using high magnification imaging.

PURPOSE: To compare photoreceptor density automated quantification in eyes with subretinal drusenoid deposits (SDD) and healthy controls using Heidelberg Spectralis High Magnification Module (HMM) imaging. METHODS: Twelve eyes of 6 patients with intermediate AMD, presenting with SDD were included, as well as twelve eyes of healthy controls. Individual dot SDD within the central 30° retina were examined with infrared confocal laser ophthalmoscopy, HMM, and spectral-domain optical coherence tomography (SD-OCT). Photoreceptor density analysis was performed on the best-quality image using the ImageJ Foci Picker plugin, after the removal of SDD from the HMM image. Correlations were made between the HMM quantified photoreceptor density, SD-OCT characteristics, stage, and number of SDD. RESULTS: Mean age was 75.17 ± 2.51 years in the SDD group (3 males, 3 females) versus 73.17 ± 3.15 years in the healthy control group (p = 0.2). Defects in the overlying ellipsoid zone were present on SD-OCT in 8/12 (66.66%) eyes. The mean ± standard deviation foci detected (i.e., cone photoreceptors) was 7123.75 ± 3683.32 foci/mm2 in the SDD group versus 13,253 ± 3331.00 foci/mm2 in the healthy control group (p = 0.0003). The number of SDD was associated with a reduction in foci density, p = 0.0055, r = - 0.7622. CONCLUSION: The decreased cone density in eyes with SDD may correlate with a decrease in retinal function in intermediate AMD eyes independent of neovascular complications or outer retinal pigment epithelial atrophy.

Predictive factors for re-bubbling after DMEK: focus on the posterior corneal surface.

PURPOSE: To understand whether the preoperative morphology of the posterior corneal surface influences the rate of re-bubbling after Descemet membrane endothelial keratoplasty (DMEK). METHODS: After retrospectively analyzing the medical records of patients undergoing DMEK, in this multicentric cross-sectional study, we performed a binomial logistic regression analysis to assess significant predictors of re-bubbling and re-transplantation after surgery. Analyzed parameters included the preoperative diagnosis, anterior and posterior surface K1/K2, central corneal thickness, posterior Q value, and other posterior corneal surface parameters evaluated on the elevation maps produced by anterior segment optical coherence tomography. Results were stratified based on the surgeons' experience. RESULTS: We included 202 eyes of 202 patients with a mean age of 69.5 ± 12.4 years; 154 eyes were operated by a high-volume surgeon and 48 by one with less experience; 48 eyes (23.8%) underwent [Formula: see text] 1 re-bubbling and 14(6.9%) [Formula: see text] 1 re-transplantation. The presence of positive/less-negative posterior corneal irregularities and irregularities with greater absolute height had a significantly higher risk of re-bubbling in both the expert and less expert group (OR = 2.85 and 1.42, OR = 3.22 and 3.01, respectively, p < 0.05), whereas more negative posterior K1 and K2 were significant risk factors only in the former group (OR = 0.67 and 0.55, respectively, p < 0.05). Endothelial decompensation other than Fuchs and pseudophakic bullous keratopathy, more negative posterior Q values and smaller distances between center, and the highest/lowest posterior corneal surface irregularity correlated with an increased risk of graft failure (OR 1.23, 1.21, and 1.29, respectively, p < 0.05). CONCLUSION: Posterior corneal surface morphology significantly influences the risk of re-bubbling after DMEK.

Optical Coherence Tomography Angiography: A 2023 Focused Update on Age-Related Macular Degeneration.

Optical coherence tomography angiography (OCTA) has extensively enhanced our comprehension of eye microcirculation and of its associated diseases. In this narrative review, we explored the key concepts behind OCTA, as well as the most recent evidence in the pathophysiology of age-related macular degeneration (AMD) made possible by OCTA. These recommendations were updated since the publication in 2020, and are targeted for 2023. Importantly, as a future perspective in OCTA technology, we will discuss how artificial intelligence has been applied to OCTA, with a particular emphasis on its application to AMD study.

Phenomenology of spontaneous closure in degenerative and mixed type lamellar macular hole.

PURPOSE: To the describe OCT imaging characteristics of a cohort of patients showing spontaneously closing degenerative or mixed type lamellar macular holes (LMH) and to compare them to the ones of a sex and age matched group showing stable lesions. METHODS: Patients diagnosed with degenerative and mixed type LMHs showing OCT-documented spontaneous anatomical closure were retrospectively selected from 3 specialized retina centres. An equal number of age and sex matching subjects were randomly selected among patients with anatomically stable lesions. RESULTS: Eleven (11) spontaneously closing (SC group) and 11 stable (ST group) degenerative LMH with a mean follow up of 4 years were recruited. Hyperreflective inner border (HIB) and linear hyperreflectivity in the outer plexiform layer (LHOP) at baseline were significantly more prevalent in SC group in processed images (respectively p = 0.007 and p = 0.003). A borderline significance in lamellar hole associated epiretinal proliferation (LHEP) at last follow up was detected (p = 0.085). As for mixed type LMH, 10 patients for SC group and 10 for ST group were recruited. LHOP at baseline in processed images was significantly more prevalent in SC group (p = 0.005). CONCLUSIONS: Spontaneously closing LMHs show higher prevalence of HIB and LHOP at the beginning of the closing process, a difference which is enhanced by image processing. These signs might be a signal of microglial and Muller cells coordinated activation.

Refractive outcome in combined phacovitrectomy: Anterior segment changes and corrective factor for IOL power calculation improvement.

PURPOSE: To analyze differences in refractive outcome Δ (difference between postoperative and expected refractive error) and in anterior segment changes between cataract surgery patients and combined phacovitrectomy patients. We also aimed to provide a corrective formula allowing to minimise the refractive outcome Δ in combined surgery patients. METHODS: Candidates for phacoemulsification and combined phacovitrectomy (respectively PHACO and COMBINED groups) were prospectively enrolled in two specialised centres. Patients underwent best corrected visual acuity (BCVA) assessment, ultra-high speed anterior segment optical coherence tomography (OCT), gonioscopy, retinal OCT, slit lamp examination and biometry at baseline, 6 weeks postoperatively and 3 months postoperatively. RESULTS: No differences in refractive Δ, refractive error and anterior segment parameters were noted between PHACO and COMBINED group (109 and 110 patients respectively) at 6 weeks. At 3 months, COMBINED group showed a spherical equivalent of -0.29 ± 0.10 D versus -0.03 ± 0.15 D in PHACO group (p = 0.023). COMBINED group showed a significantly higher Crystalline Lens Rise (CLR), angle-to-angle (ATA) and anterior chamber width (ACW) and a significantly lower anterior chamber depth (ACD) and refractive Δ with all 4 considered formulas at 3 months. For IOL power lower than 15, a hyperopic shift was observed instead. CONCLUSIONS: Anterior segment OCT suggests anterior displacement of the effective lens position in patients undergoing phacovitrectomy. A corrective formula can be applied to IOL power calculation to minimize undesired refractive error.

'Structural OCT changes distinguishing between myopic macular haemorrhages due to choroidal neovascularization and spontaneous Bruch's membrane rupture: the "myopic 2 binary reflective sign".

OBJECTIVE: To evaluate the sensitivity and specificity of structural optical coherence tomography (OCT) in comparison to fluorescein angiography (FA) and OCT angiography (OCTA) in discerning between macular haemorrhages (MH) due to myopic choroidal neovascularization (m-CNV) and idiopathic macular haemorrhage (IMH) in myopic patients and to suggest a new OCT biomarker to discern these two entities. METHODS AND ANALYSIS: In this longitudinal retrospective study, patients affected by MH and pathological myopia were included. All patients underwent OCTA and FA to discern bleeding from m-CNV or IMH. Furthermore, all patients underwent a structural OCT and 2 expert graders evaluated the presence of the myopic 2 binary reflective sign as a biomarker to discern between IMH and bleeding from m-CNV. RESULTS: Forty-seven eyes of 47 patients were enrolled. By means of angiographic examinations, 34 out of 47 eyes with MH (57%) were diagnosed as m-CNV, whereas 13 eyes (43%) as IMH. Using structural OCT, the graders identified the presence of the myopic 2 binary reflective sign in 13 out of 13 eyes with IMH. In 33 out of 34 cases with m-CNV, the 2 graders established the absence of the sign. This accounted for 100% of sensibility and 97% of specificity of structural OCT in discerning between MH from m-CNV and IMH. CONCLUSION: Structural OCT can discern with good reliability between IMH and bleeding from m-CNV based on the presence/ absence of the myopic 2 binary reflective sign. This could be of paramount relevance in the clinical setting for the diagnosis and treatment of HM patients.

Refractive Outcomes With New Generation Formulas for Intraocular Lens Power Calculation in Radial Keratotomy Patients.

PURPOSE: Radial keratotomies (RKs) are responsible for corneal irregularities resulting in biometric errors and lower best-corrected visual acuity (BCVA) due to lower-order and higher-order optical aberrations. The aim of the study was to compare performances of new and old generation formulas in a population of RK patients. METHODS: RK patients who underwent phacoemulsification with intraocular lens (IOL) implantation were retrospectively recruited. Inclusion criteria were availability of preoperative and 6-month postoperative BCVA assessment, topography, and tomography. Documented refraction instability, corneal ectasia, and previous ocular surgery except for RK were exclusion criteria. Mean prediction error (ME), mean absolute prediction error (MAE), and incidence of MAE > 0.25D were calculated for SRK-T, Barrett True K, EVO 2.0, Kane, and PEARL-DGS. RESULTS: Twenty-seven patients with a mean baseline BCVA of 0.32 ± 0.18 logMAR and a mean corneal root mean square (RMS) value of 1.59 ± 0.91 µm were included. EVO 2.0, Kane, and PEARL-DGS showed a significantly lower MAE and lower ME compared with all other formulas ( P < 0.001 and P < 0.001) and a significant lower incidence of MAE >0.25D ( P < 0.001). Significant differences were still detected when using 3-mm mean keratometry for IOL calculation. CONCLUSIONS: PEARL-DGS, Kane, and EVO 2.0 formulas show superior accuracy in IOL power calculation compared with SRK-T and Barrett True K in RK patients, with no significant differences between the 3.

Smaller-Incision new-generation implantable miniature telescope: Three-months follow-up study.

PURPOSE: To evaluate three months follow-up of SING IMT implant in patients affected by late-stage AMD. DESIGN: Prospective cohort study. SUBJECTS: In a total of 80 eyes of 40 patients who underwent the enrollment tests, 11 patients' eyes affected by late-stage AMD matched the inclusion criteria and underwent SING IMT implant from February to June 2022. METHODS: Before surgery, each patient underwent the enrollment examination to verify inclusion and exclusion criteria. MAIN OUTCOME MEASURES: BCVA for distance and for near, IOP, ACD and ECD were evaluated at 1 and 3 months follow up. Also quality of life in doing the activities of daily life was evaluated. RESULTS: BCVA for distance and for near improved from baseline to 3 months follow up (23.91 ± 9.418 ETDRS letters and 59.09 ± 11.58 ETDRS letters respectively (p < 0.001). An endothelial cell loss was shown (p < 0.001), with a rate of cell density reduction around 8.3% (baseline vs 3 months). CONCLUSIONS: SING IMT could be a valid surgical device to improve patients' sight and quality of life which have been deteriorated by late-stage macular degeneration. Further studies with more patients and longer follow up are needed to confirm our results.

Deep learning model for automatic differentiation of EMAP from AMD in macular atrophy.

To create a deep learning (DL) classifier pre-trained on fundus autofluorescence (FAF) images that can assist the clinician in distinguishing age-related geographic atrophy from extensive macular atrophy and pseudodrusen-like appearance (EMAP). Patients with complete outer retinal and retinal pigment epithelium atrophy secondary to either EMAP (EMAP Group) or to dry age related macular degeneration (AMD group) were retrospectively selected. Fovea-centered posterior pole (30° × 30°) and 55° × 55° degree-field-of-view FAF images of sufficiently high quality were collected and used to train two different deep learning (DL) classifiers based on ResNet-101 design. Testing was performed on a set of images coming from a different center. A total of 300 patients were recruited, 135 belonging to EMAP group and 165 belonging to AMD group. The 30° × 30° FAF based DL classifier showed a sensitivity of 84.6% and a specificity of 85.3% for the diagnosis of EMAP. The 55° × 55° FAF based DL classifier showed a sensitivity of 90% and a specificity of 84.6%, a performance that was significantly higher than that of the 30° × 30° classifer (p = 0.037). Artificial intelligence can accurately distinguish between atrophy caused by AMD or by EMAP on FAF images. Its performance are improved using wide field acquisitions.

Systematic ophthalmologic evaluation in cardio-facio-cutaneous syndrome: A genotype-endophenotype correlation.

Cardio-facio-cutaneous syndrome (CFCS) is a rare genetic disorder belonging to the RASopathies, a group of developmental syndromes caused by upregulated RAS/MAPK signaling. Pathogenic variants affecting four genes, KRAS, BRAF, MAP2K1 and MAP2K2, encoding core signal transducers of the pathway, underlie the condition. Major clinical features include a distinctive facies, ectodermal and cardiac anomalies, reduced postnatal growth, intellectual disability, and musculoskeletal abnormalities. Similar to other RASopathies, reports of visual impairment, high refractive error, optic nerve pallor, and other ocular abnormalities have been anecdotally reported in the literature. The aim of our study is to report the prevalence of ophthalmologic abnormalities in a large monocentric cohort of individuals affected by CFCS and explore the occurrence of genotype-endophenotype correlations in this series of patients. We observed that BRAF mutations are associated to a higher prevalence of anisometropia >3D (11.8% vs. 0%) and high astigmatism (29.4% vs. 0%; both p < 0.001) while patients with mutations in other genes had a significantly higher prevalence of myopia >6 D (60% vs. 5.9%; p = 0.012). Pale optic disc was associated with higher prevalence of inferior oblique muscle (IO) overaction (33.3% vs. 0%) and lower prevalence of ptosis (0% vs. 11.8%; both p < 0.001). Combined exotropia, IO overaction and nystagmus were frequent in patients with pale optic nerve. Our findings might suggest the need for earlier ophthalmologic referral for CFCS patients due to high risk of amblyopia, especially those expressing BRAF mutations.

ARTIFICIAL INTELLIGENCE'S ROLE IN DIFFERENTIATING THE ORIGIN FOR SUBRETINAL BLEEDING IN PATHOLOGIC MYOPIA.

PURPOSE: To identify salient imaging features to support human-based differential diagnosis between subretinal hemorrhage (SH) due to choroidal neovascularization (CNV) onset and SH without CNV (simple bleeding [SB]) in pathologic myopia eyes using a machine learning (ML)-based step-wise approach. METHODS: Four different methods for feature extraction were applied: GradCAM visualization, reverse engineering, image processing, and human graders' measurements. GradCAM was performed on a deep learning model derived from Inception-ResNet-v2 trained with OCT B-scan images. Reverse engineering consisted of merging U-Net architecture with a deconvolutional network. Image processing consisted of the application of a local adaptive threshold. Available OCT B-scan images were divided in two groups: the first group was classified by graders before knowing the results of feature extraction and the second (different images) was classified after familiarization with the results of feature extraction. RESULTS: Forty-seven and 37 eyes were included in the CNV group and the simple bleeding group, respectively. Choroidal neovascularization eyes showed higher baseline central macular thickness ( P = 0.036). Image processing evidenced in CNV eyes an inhomogeneity of the subretinal material and an interruption of the Bruch membrane at the margins of the SH area. Graders' classification performance improved from an accuracy of 76.9% without guidance to 83.3% with the guidance of the three methods ( P  = 0.02). Deep learning accuracy in the task was 86.0%. CONCLUSION: Artificial intelligence helps identifying imaging biomarkers suggestive of CNV in the context of SH in myopia, improving human ability to perform differential diagnosis on unprocessed baseline OCT B-scan images. Deep learning can accurately distinguish between the two causes of SH.