Volume 20, Number 9Monday, February 24, 2020FEBRUARY IS AGE-RELATED MACULAR DEGENERATION AWARENESS MONTH In this issue: (click heading to view article)
Deep Learning Model for GA Segmentation to Study Long-term Natural HistoryResearchers developed and validated a deep learning model for the automatic segmentation of geographic atrophy in color fundus images (CFIs) and its application to study growth rate of GA, as part of a prospective, multicenter, natural history study with up to 15 years of follow up.A total of 409 CFIs of 238 eyes with GA from the Rotterdam Study (RS) and the Blue Mountain Eye Study (BMES) for model development, and 3,589 CFIs of 376 eyes from the Age-Related Eye Disease Study (AREDS) were analyzed for GA growth rate. Researchers implemented and optimized a deep learning model based on an ensemble of encoder-decoder architectures for the segmentation of GA in CFIs. Four experienced graders delineated, in consensus, GA in CFIs from RS and BMES. These manual delineations helped evaluate the segmentation model using five-fold cross-validation. The model was further applied to CFIs from AREDS to study the growth rate of GA. Researchers used linear regression analysis to study associations between baseline structural biomarkers and the GA growth rate. They made a general estimate of the progression of GA area over time by combining growth rates of all eyes with GA from the AREDS set. Main outcome measures included the automatically segmented GA and GA growth rate. Here were some of the findings: • The model obtained an average Dice coefficient of 0.72 ±0.26 on the BMES and RS set while comparing the automatically segmented GA area to the graders’ manual delineations. • An intraclass correlation coefficient of 0.83 was reached between the automatically estimated GA area and the graders' consensus measures. • Nine automatically calculated structural biomarkers (area, filled area, convex area, convex solidity, eccentricity, roundness, foveal involvement, perimeter and circularity) were significantly associated with growth rate. • Combining all growth rates indicated that the GA area grew quadratically up to an area of around 12 mm2, after which the growth rate stabilized or decreased. Researchers determined that their deep learning model allowed for fully automatic and robust segmentation of GA in CFIs. They added that the segmentations could be used to extract structural characteristics of GA that predict its growth rate. SOURCE: Liefers B, Colijn JM, González-Gonzalo C, et al. A deep learning model for segmentation of geographic atrophy to study its long-term natural history. Ophthalmology 2020; Feb 14. [Epub ahead of print]. Qualitative & Quantitative Analysis of the Corneal Endothelium with Smartphone Specular MicroscopyInvestigators sought to demonstrate the feasibility of a lower-cost, portable method for qualitative and quantitative analysis of the corneal endothelium using a smartphone and a slit-lamp. Source: Fliotsos MJ, Deljookorani S, Dzhaber D, et al. Qualitative and quantitative analysis of the corneal endothelium with smartphone specular microscopy. Cornea 2020; Feb 14. [Epub ahead of print.] Intra- & Interocular Differences in Parafoveal Vascular Density in Diabetic Patients Without DRScientists evaluated the associations between optical coherence tomography angiography-measured vascular density (VD), intraocular and interocular VD differences, and clinical factors in diabetic patients without diabetic retinopathy. One of the researchers received financial support from Canon (Tokyo, Japan).They retrospectively reviewed 94 Type 2 diabetic patients without diabetic retinopathy who had undergone OCTA. They measured vascular density and vessel skeleton density in a 3-mm central zone in the total capillary plexus, superficial capillary plexus, deep capillary plexus (DCP) and choriocapillaris layers. And they determined the intraocular VD difference between the superior and inferior zones, while figuring the interocular VD difference between both eyes. Scientists evaluated associations between OCTA parameters and clinical factors. Here were some of the findings: • Vascular density and intraocular and interocular VD differences were significantly associated with signal strength of the image, which was related with age and lens opacity. • In multivariate analysis, diabetes duration was negatively associated with skeleton density in total capillary plexus and superficial capillary plexus layers, and positively associated with intraocular VD difference in the superficial capillary plexus layer. • The estimated glomerular filtration rate was negatively associated with the intraocular skeleton density difference in the total capillary plexus layer, interocular VD, and skeleton density differences in the total capillary plexus layer. Scientists reported that intraocular and interocular VD difference may be an easy and sensitive way to detect subtle early microvascular changes in diabetic patients. SOURCE: Kim YK, An Y, Park SP. Intraocular and interocular differences in parafoveal vascular density in diabetic patients without diabetic retinopathy. Retina 2020; Feb 11. [Epub ahead of print]. Micropulse Transscleral Cyclophotocoagulation for Refractory GlaucomaThis study evaluated the six-month safety and efficacy of micropulse transscleral cyclophotocoagulation (MP-TSCPC) in cases of uncontrolled glaucoma/ocular hypertension using a reduced energy protocol.Researchers initiated a retrospective analysis of individuals undergoing MP-TSCPC from January to April 2018. Individuals received up to 90 seconds of laser with settings of 2,000 mW/Cm2 and a duty cycle of 31.3 percent. A total of 29 individuals were included, with a mean age of 64.7 ±15.1 years. The most common diagnosis was primary open-angle glaucoma (41.4 percent) with a mean logMAR visual acuity of 1.5 ±1.2. All subjects had either undergone intraocular surgery (58.6 percent filtration surgery) or continuous wave diode laser prior to micropulse treatment. Mean pre-laser IOP was 26.2 ±11.1 mmHg. Here were some of the findings: • A significant IOP reduction (p<0.05) was found: It decreased to 15.8 ±5.4 mmHg (39.7 percent reduction) at one month; 15.04 ±5.25 mmHg (42.6 percent reduction) at three months; and 18.19 ±7.47 mmHg (30.6 percent reduction) at six months. • A corresponding reduction (p<0.05) was found in the number of topical agents required to control pressure from a baseline of 3.31 ±0.97: to 2.72 ±0.88 at one month; 2.76 ±0.91 at three months; and 2.90 ±1.08 at six months. • Requirements for oral acetazolamide reduced from 41.3 percent at baseline to 3.4 percent at six months. • Success rates were 75.9 percent at one month, 79.3 percent at three months, and 58.6 percent at six months. • Researchers found no drop in visual acuity, no change in central retinal thickness and no intraocular inflammation. Researchers wrote that MP-TSCPC at a decreased duration was effective at reducing intraocular pressure in ethnically diverse glaucoma patients refractory to previous glaucoma laser or surgeries at six months follow-up, and revealed no significant complications. They added that further work would be needed to confirm long-term efficacy and determine optimal settings. SOURCE: Vig N, Ameen S, Bloom P, et al. Micropulse transscleral cyclophotocoagulation: Initial results using a reduced energy protocol in refractory glaucoma. Graefes Arch Clin Exp Ophthalmol 2020; Feb 8. [Epub ahead of print]. BRIEFLY Acucela Provides Update on Emixustat Phase III Clinical Trial LumiThera to Expand into DR Research for Valeda Light Delivery System Alcon Switches Pataday to OTC Review of Ophthalmology® Online is published by the Review Group, a Division of Jobson Medical Information LLC (JMI), 11 Campus Boulevard, Newtown Square, PA 19073. |