Evaluation of clinical images is essential for diagnosis in many specialties. Therefore the development of computer vision algorithms to help analyze biomedical images will be important. In ophthalmology, optical coherence tomography (OCT) is critical for managing retinal conditions. We developed a convolutional neural network (CNN) that detects intraretinal fluid (IRF) on OCT in a manner indistinguishable from clinicians. Using 1,289 OCT images, the CNN segmented images with a 0.911 cross-validated Dice coefficient, compared with segmentations by experts. Additionally, the agreement between experts and between experts and CNN were similar. Our results reveal that CNN can be trained to perform automated segmentations of clinically relevant image features.

Despite advances in artificial intelligence (AI), its application in medical imaging has been burdened and limited by expert-generated labels. We used images from optical coherence tomography angiography (OCTA), a relatively new imaging modality that measures retinal blood flow, to train an AI algorithm to generate flow maps from standard optical coherence tomography (OCT) images, exceeding the ability and bypassing the need for expert labeling. Deep learning was able to infer flow from single structural OCT images with similar fidelity to OCTA and significantly better than expert clinicians (P < 0.00001). Our model allows generating flow maps from large volumes of previously collected OCT data in existing clinical trials and clinical practice. This finding demonstrates a novel application of AI to medical imaging, whereby subtle regularities between different modalities are used to image the same body part and AI is used to generate detailed inferences of tissue function from structure imaging.

Introduction:To investigate whether a systematic approach to subgrouping traumatic ptosis according to etiology can allow for better tailoring of prognosis and treatment.Methods:Retrospective chart review of patients with trauma-related blepharoptosis managed by Oculoplastic surgery specialists at an academic medical center from January 1995 to November 2015. Injury mechanism, eyelid position and function, interventions, and outcomes were reviewed.Results:Of 648 patients treated for blepharoptosis, 55 (8.5%) were traumatic. Careful review revealed 4 subcategories of traumatic ptosis cases: aponeurotic (n = 16), myogenic (n = 18), neurogenic (n = 7), and mechanical (n = 14). Margin reflex distance (MRD1) at presentation was significantly worse for the myogenic subtype (-0.59 mm, SD ±2.09, P = 0.046). The aponeurotic subtype had the best average levator function at presentation (14.29 mm, SD ±2.05), while myogenic had the worst (8.41 mm, SD ±4.94) (P = 0.004). Thirty-five (63.6%) patients were managed surgically. Final MRD1 was significantly different for each subtype (P = 0.163), with aponeurotic 2.63 mm (SD ±1.01), myogenic 1.29 mm (SD ±2.24), neurogenic 1.79 mm (SD ±2.48), and mechanical 2.31 mm (SD ±1.18). There was a significant increase in MRD1 from presentation to final follow up across all groups (P < 0.05).Conclusion:Traumatic ptosis is heterogenous. Systematically evaluating traumatic ptosis cases by trauma mechanism can guide decisions about prognosis and management. Two-thirds of cases were treated surgically, with most patients responding well to conjunctiva-Müller resection or external levator advancement. While all subgroups demonstrated improvement in MRD1 at final follow up, aponeurotic cases had the best prognosis, while myogenic fared the worst and required the longest for maximal recovery.

BACKGROUND/AIMS To examine the relationship between change in optic nerve head (ONH) morphology and retinal blood flow in patients with open-angle glaucoma (OAG) of African (AD) and European descent (ED) over three years. METHODS 112 patients with OAG (29 AD; 83 ED) underwent assessment of ONH morphology using Heidelberg retinal tomography (HRT-III) and retinal blood flow using confocal scanning laser Doppler. Repeated measures analysis of covariance was used to compare baseline and 3-year measurements and Pearson correlations were calculated to evaluate the relationships. RESULTS In OAG patients of AD, change in superior mean retinal blood flow was strongly, negatively correlated with change in cup/disc (C/D) area ratio (r=−0.78, p=0.020) and cup area (r=−0.75, p=0.0283) and strongly, positively correlated with change in rim area (r=0.74, p=0.0328) over three years. In OAG patients of AD, change in inferior mean retinal blood flow was strongly, negatively correlated with changes in C/D area ratio (r=−0.88, p=0.0156) and linear C/D ratio (r=−0.86, p=0.0265) over three years. In OAG patients of ED, these correlations were weak and did not reach statistical significance. DISCUSSION OAG patients of AD may have a stronger vascular component to their glaucoma pathophysiology than patients of ED.