Automated retinal vessel segmentation is among the most significant application and research topics in ophthalmologic image analysis. Deep learning based retinal vessel segmentation models have attracted much attention in the recent years. However, current deep network designs tend to predominantly focus on vessels which are easy to segment, while overlooking vessels which are more difficult to segment, such as thin vessels or those with uncertain boundaries. To address this critical gap, we propose a new end-to-end deep learning architecture for retinal vessel segmentation: hard attention net (HAnet). Our design is composed of three decoder networks: the first of which dynamically locates which image regions are “hard” or “easy” to analyze, while the other two aim to segment retinal vessels in these “hard” and “easy” regions independently. We introduce attention mechanisms in the network to reinforce focus on image features in the “hard” regions. Finally, a final vessel segmentation map is generated by fusing all decoder outputs. To quantify the network's performance, we evaluate our model on four public fundus photography datasets (DRIVE, STARE, CHASE_DB1, HRF), two recent published color scanning laser ophthalmoscopy image datasets (IOSTAR, RC-SLO), and a self-collected indocyanine green angiography dataset. Compared to existing state-of-the-art models, the proposed architecture achieves better/comparable performances in segmentation accuracy, area under the receiver operating characteristic curve (AUC), and f1-score. To further gauge the ability to generalize our model, cross-dataset and cross-modality evaluations are conducted, and demonstrate promising extendibility of our proposed network architecture.

Purpose: To determine the risk factors for low intraocular pressure (IOP) and its detrimental consequences after trabeculectomy. Patients and Methods: We performed a retrospective chart review of consecutive patients aged 12 years and above undergoing trabeculectomy alone by 1 of 2 surgeons between May 2000 and October 2008 at the Wilmer Institute. Results: Among 753 eyes of 596 patients, 112 eyes (14.9%) of 103 patients had an IOP of ≤5 mm Hg at ≥3 months postoperatively (late low IOP). Physical signs related to low IOP occurred in 61 eyes of 58 patients, and 40 eyes of 37 patients had revision surgery for low IOP. Physical signs of low IOP included 34 eyes with choroidal detachment or shallow anterior chamber, 10 with hypotony maculopathy, and 7 with both. Compared with 187 control eyes of 165 patients, risk factors for late low IOP included: surgeon 2 (P=0.0003), left eyes (P=0.03), and secondary glaucoma (P=0.05). Physical signs of low IOP were more common in phakic eyes (P=0.03), whereas need for revision surgery was associated with younger age (P=0.01). The presence of hypotony maculopathy significantly decreased the risk of choroidal detachment/shallow anterior chamber (P=0.003). Hypotony maculopathy was associated with younger age (P=0.02). Conclusions: Late low IOP was more common with techniques used by 1 of 2 surgeons and in eyes with secondary glaucoma. Clinical signs of low IOP were more common in phakic eyes. Younger age was a risk factor for hypotony maculopathy. Hypotony maculopathy and choroidal detachment tend to occur in different eyes.

Purpose: Choroidal thickness increases linearly with intraocular pressure (IOP) lowering. We studied the relationship between the change in size of the choroidal vasculature and IOP lowering after glaucoma procedures. Methods: Thirty eyes of twenty-nine patients were examined pre- and postoperatively for up to 6 months with standard clinical assessment, enhanced depth imaging spectral-domain optical coherence tomography (OCT), and axial length measurement. Each enhanced depth imaging spectral-domain OCT image was analyzed using three separate methods to determine the choroidal thickness, choroidal vessel thickness, choroidal interstitial thickness, large choroidal vessel layer thickness, medium choroidal vessel layer thickness, and light-dark ratio. Bivariate linear regression analysis was completed with largest change in IOP as the independent variable. The dependent variables included choroidal thickness, choroidal vessel thickness, and choroidal interstitial thickness, at the largest change in IOP. Multivariable regression analysis using a generalized estimating equation to account for multiple measurements per eye was also completed. Results: Mean choroidal vessel thickness increases 1.5 μm for every 1 mm Hg decrease in IOP (P < 0.0001; 95% confidence interval [CI], 0.8, 2.1) and choroidal interstitial thickness increases 1.3 μm for every 1 mm Hg change in IOP (P < 0.0001; 95% CI, 0.8, 1.8). There was no significant association between change in IOP and change in large choroidal vessel layer temporally (P = 0.13), nasally (P = 0.20), or subfoveally (P = 0.18). There was also no association between IOP and the light-dark ratio (P = 0.16). Conclusions: The increase in choroidal thickness at lower IOP is associated with approximately equal increases in its intravascular and extravascular compartments.

Introduction: The advent of Microinvasive Glaucoma Surgery (MIGS) offers a novel approach in the treatment of glaucoma with the number of procedures developing at an exciting pace. MIGS procedures aim to lower intraocular pressure (IOP) via: (1) increasing trabecular outflow, (2) increasing outflow via suprachoroidal shunts, (3) reducing aqueous production, and (4) subconjunctival filtration. Areas covered: A comprehensive search for published studies for Microinvasive Glaucoma Surgery (MIGS) devices or procedure was undertaken using PubMed. Search terms included ‘minimally invasive glaucoma surgery’, ‘microincisional glaucoma surgery’, and ‘microinvasive glaucoma surgery’. A manual search for each device or procedure was also performed. After review, randomized control trials and prospective studies were preferentially included. Expert commentary: These procedures offer several benefits: an improved safety profile allowing for intervention in earlier stages of glaucoma, combination with cataract surgery, and decreased dependence on patient compliance with topical agents. Established MIGS procedures have proven efficacy and more recent devices and procedures show promising results. Despite this, further study is needed to assess the long term IOP-lowering effectiveness of these procedures. Particularly, rigorous study with more randomized control trials and head-to-head comparisons would allow for better informed clinical and surgical decision-making. MIGS offers new solutions for glaucoma treatment.

Purpose: We evaluated the patient-control differences and predictive value of the retina as potential biomarkers for schizophrenia. Methods: The institutional study included both eyes of 58 schizophrenia spectrum disorder (SSD) patients (age 37.2 ± 12.3 years) and 35 controls (age 41.1 ± 15.2 years). Retinal nerve fiber layer (RNFL), ganglion cell–inner plexiform layer, outer retinal photoreceptor complex, and total macula thicknesses were measured by optical coherence tomography (OCT). Anterior segment parameters including central corneal thickness, anterior chamber depth, and axial length were measured to rule out confounds on the retinal measures. Results: The peripapillary RNFL was overall significantly thinner in SSD relative to controls (F = 3.97, P = 0.049), most pronounced in the temporal (5.2 µm difference, F = 6.95, P = 0.010) and inferior quadrants (12.1 µm difference, F = 7.32, P = 0.009). There were no significant group differences in thickness for the macular RNFL, ganglion, or photoreceptor cell related measures (P > 0.05). Peripapillary RNFL, central macula, and outer photoreceptor complex thicknesses were together able to classify SSD patients with 80% sensitivity and 71% specificity; area under the curve = 0.82 (95% confidence interval, 0.75–0.88). Conclusions: SSD patients exhibited significant RNFL thinning relative to controls. Notably, retinal thickness measures including both peripapillary and macular data exhibited improved diagnostic accuracy for SSD as compared to these regions alone. Translational Relevance: This is the first study to evaluate the predictive value of both the inner and outer retina in SSD. OCT retinal thickness measures including peripapillary data in conjunction with macular data may provide an informative, noninvasive in vivo ocular biomarker for schizophrenia.