Patients' perspective on emergency treatment of ophthalmologic diseases during the first phase of SARS-CoV2 pandemic in a tertiary referral center in Germany - the COVID-DETOUR questionnaire study.

BACKGROUND: During the first wave of the COVID-19 pandemic, the need of treatment of urgent ophthalmological diseases and the possible risk of a SARS-CoV-2 infection had to be weighed against each other. In this questionnaire study, we aimed to analyze potential barriers and patients' health beliefs during and after the lockdown early 2020 in a tertiary referral center in Kiel, Germany. METHODS: Patients admitted for the treatment of urgent ophthalmic diseases between March 1st, 2020, and June 3rd, 2020, were asked to participate in a questionnaire study. After informed consent was obtained, patients were interviewed using a standardized questionnaire which addressed aspects of their medical history, their health beliefs concerning the COVID-19 pandemic and barriers on their way to the treatment center. The study group was subdivided into two subgroups, depending on the occurrence of their symptoms, before and after the lockdown was ended on April 20th, 2020. RESULTS: Ninety-three patients were included, 43 in subgroup A (before April 20th) and 50 in subgroup B (April 20th or later). Retinal disorders were the most common causes for admission (approximately 60%).. Only 8 patients (8.6%) experienced a delay between their decision to visit a doctor until the actual examination. Every fourth patient was afraid of a COVID-19 infection, and expected a higher likelihood for an infection at the hospital. Patients with comorbidities tended to be more likely to be afraid of an infection (correlation coefficient 0.183, p = 0.0785) and were significantly more likely to be concerned about problems with organizing follow-up care (corr. Coefficient 0.222, p = 0.0328). Higher age was negatively correlated with fear of infection (corr. Coefficient - 0.218, p-value 0.034). CONCLUSION: In this questionnaire study, only a minority of patients indicated a delay in treatment, regardless of whether symptoms occurred before or after the lockdown before April 20th, 2020. While patients with comorbidities were more concerned about infection and problems during follow-up care, patients of higher age - who have a higher mortality - were less afraid. Protection of high-risk groups should be prioritized during the SARS-CoV-2 pandemic. TRIAL REGISTRATION: The study was registered as DRKS00021630 at the DRKS (Deutsches Register Klinischer Studien) before the conduction of the study on May 5th, 2020.

When Two Eyes Don't Suffice-Learning Difficult Hyperfluorescence Segmentations in Retinal Fundus Autofluorescence Images via Ensemble Learning.

Hyperfluorescence (HF) and reduced autofluorescence (RA) are important biomarkers in fundus autofluorescence images (FAF) for the assessment of health of the retinal pigment epithelium (RPE), an important indicator of disease progression in geographic atrophy (GA) or central serous chorioretinopathy (CSCR). Autofluorescence images have been annotated by human raters, but distinguishing biomarkers (whether signals are increased or decreased) from the normal background proves challenging, with borders being particularly open to interpretation. Consequently, significant variations emerge among different graders, and even within the same grader during repeated annotations. Tests on in-house FAF data show that even highly skilled medical experts, despite previously discussing and settling on precise annotation guidelines, reach a pair-wise agreement measured in a Dice score of no more than 63-80% for HF segmentations and only 14-52% for RA. The data further show that the agreement of our primary annotation expert with herself is a 72% Dice score for HF and 51% for RA. Given these numbers, the task of automated HF and RA segmentation cannot simply be refined to the improvement in a segmentation score. Instead, we propose the use of a segmentation ensemble. Learning from images with a single annotation, the ensemble reaches expert-like performance with an agreement of a 64-81% Dice score for HF and 21-41% for RA with all our experts. In addition, utilizing the mean predictions of the ensemble networks and their variance, we devise ternary segmentations where FAF image areas are labeled either as confident background, confident HF, or potential HF, ensuring that predictions are reliable where they are confident (97% Precision), while detecting all instances of HF (99% Recall) annotated by all experts.