Deep Learning for Localized Detection of Optic Disc Hemorrhages.

PURPOSE: To develop an automated deep learning system for detecting the presence and location of disc hemorrhages in optic disc photographs. DESIGN: Development and testing of a deep learning algorithm. METHODS: Optic disc photos (597 images with at least 1 disc hemorrhage and 1075 images without any disc hemorrhage from 1562 eyes) from 5 institutions were classified by expert graders based on the presence or absence of disc hemorrhage. The images were split into training (n = 1340), validation (n = 167), and test (n = 165) datasets. Two state-of-the-art deep learning algorithms based on either object-level detection or image-level classification were trained on the dataset. These models were compared to one another and against 2 independent glaucoma specialists. We evaluated model performance by the area under the receiver operating characteristic curve (AUC). AUCs were compared with the Hanley-McNeil method. RESULTS: The object detection model achieved an AUC of 0.936 (95% CI = 0.857-0.964) across all held-out images (n = 165 photographs), which was significantly superior to the image classification model (AUC = 0.845, 95% CI = 0.740-0.912; P = .006). At an operating point selected for high specificity, the model achieved a specificity of 94.3% and a sensitivity of 70.0%, which was statistically indistinguishable from an expert clinician (P = .7). At an operating point selected for high sensitivity, the model achieves a sensitivity of 96.7% and a specificity of 73.3%. CONCLUSIONS: An autonomous object detection model is superior to an image classification model for detecting disc hemorrhages, and performed comparably to 2 clinicians.

Rapidly progressive streptococcus dysgalactiae corneal ulceration associated with erlotinib use in stage IV lung cancer.

PURPOSE: To present a unique case of streptococcus dysgalactiae keratitis with progression to corneal perforation and endophthalmitis, in the setting of epidermal growth factor receptor inhibitor (erlotinib) therapy for advanced non-small cell lung cancer. OBSERVATIONS: An 89-year-old female with non-small cell lung cancer on erlotinib presented with corneal perforation due to infectious keratitis. Microbial cultures grew streptococcus dysgalactiae, a virulent pathogen known to affect immunocompromised patients that has not been previously described to cause infectious keratitis. Despite aggressive medical intervention, the clinical course was complicated by rapid progression to no light perception visual acuity in the setting of endophthalmitis with orbital cellulitis, necessitating evisceration. CONCLUSIONS AND IMPORTANCE: Epidermal growth factor receptor inhibitor therapy can result in significant ocular complications including dry eyes, epithelial keratopathy, non-healing abrasions, infectious keratitis, and rarely, corneal melting and perforation. These side effects can predispose patients to aggressive infections with rare organisms, highlighting the importance of understanding the ocular side effects of systemic chemotherapeutic agents.

Endothelial survival after Descemet-stripping automated endothelial keratoplasty in eyes with retained anterior chamber intraocular lenses: two-year follow-up.

PURPOSE: To evaluate endothelial cell survival 2 years after Descemet-stripping automated endothelial keratoplasty (DSAEK) for the treatment of endothelial dysfunction in the presence of an anterior chamber intraocular lens (AC IOL). SETTING: LSU Eye Center, New Orleans, Louisiana, and Florida Eye Microsurgical Institute, Boynton Beach, Florida, USA. DESIGN: Case series. METHODS: This study comprised eyes with endothelial failure that had DSAEK in the presence of an AC IOL. Donor central endothelial cell density (ECD) was recorded 6 months, 1 year, and 2 years postoperatively and compared with preoperative ECD eye-bank values. RESULTS: The study evaluated 25 eyes; data from 20 eyes were available up to 2 years postoperatively. The mean preoperative ECD was 2606 cells/mm(2). At 1 year, the mean ECD was 1943 cells/mm(2) ± 266 (SD), representing a mean cell loss from preoperative measurements of 24% ± 12%. At 2 years, the mean ECD was 1831 ± 291 cells/mm(2), representing a 28% ± 13% cell loss from preoperative values. The additional cell loss between the first and second postoperative years was not statistically significant (P=.265). CONCLUSIONS: Descemet-stripping automated endothelial keratoplasty grafts in the presence of a well-centered AC IOL with an AC IOL-to-endothelial depth greater than 3.0 mm had a mean postoperative donor endothelial cell loss of 24% at 1 year and 28% at 2 years. There was no significant difference in cell loss in this series compared with ECD loss in DSAEK surgeries in the presence of a posterior chamber IOL. FINANCIAL DISCLOSURE: No author has a financial or proprietary interest in any material or method mentioned.

Endothelial cell survival after descemet stripping with automated endothelial keratoplasty with retained anterior chamber intraocular lens.

PURPOSE: To evaluate the donor endothelial cell loss in the first year after Descemet stripping endothelial keratoplasy (DSAEK) for the treatment of endothelial dysfunction in the presence of an anterior chamber (AC) intraocular lens (IOL). METHODS: We performed a retrospective study of a surgical case series of 25 patients who underwent DSAEK in the presence of an AC IOL. The donor central endothelial cell density (ECD) was recorded at 6 months and 1 year postoperatively and then compared with the preoperative eye bank values. RESULTS: The average and SD ECD at 6 months was 1876 ± 286 cells per square millimeters representing a mean cell loss from preoperative ECD measurement of 37% ± 13%. At 1 year postoperatively, ECD was 1793 ± 311 cells per square millimeters that correspond to a 40% ± 11% cell loss from preoperative ECD measurement. The increased cell loss between 6 months and 1 year was not significant statistically (P = 0.365). CONCLUSIONS: DSAEK in the presence of a well-centered AC IOL and an AC depth greater than 3 mm has a mean donor endothelial cell loss of 37% at 6 months and 40% at 12 months postoperatively. We found no difference in cell loss in our series compared with previous ECD loss in DSAEK surgery alone. Emphasis in a surgical technique that minimizes tissue manipulation and reduces the risk of graft-AC IOL touch should be considered to maximize the endothelial graft survival in these cases.