No abstract

Background Population-based studies have linked measures of sleep disordered breathing (SDB) to nocturnally occurring atrial fibrillation (AF) episodes. Whether measures of SDB and sleep quality are associated with prevalent AF has not been studied in an unselected population. We investigated the cross-sectional association with prevalent AF of objectively collected prespecified measures of overnight sleep breathing disturbances, sleep stage distributions, arousal, and sleep duration. Methods AF prevalence, defined by diagnosis codes, study electrocardiography and sleep study was examined among Multi-Ethnic Study of Atherosclerosis (MESA) participants who underwent polysomnography in the MESA Sleep Study (n=2048). Measurements and Main Results Higher apnea hypopnea index (AHI) was associated with increased odds of AF, although the significance was attenuated after full adjustment for covariates including prevalent cardiovascular disease (OR: 1.22 [0.99–1.49] per SD [17/hr], p=0.06). Analyses of sleep architecture measures and AF revealed significantly lower odds of AF associated with longer duration of slow wave sleep (SWS) (OR: 0.66 [0.5–0.89] per SD [34 min], p=0.01) which persisted after additionally adjusting for AHI (OR: 0.68 [0.51–0.92], p=0.01). Higher sleep efficiency was significantly associated with lower likelihood of AF but the significance was lost when adjusted for AHI. No significant association was present between sleep duration and AF. In a model including both AHI and arousal index, the association between AHI and AF was strengthened (AHI: OR 1.49 [1.15–1.91] per SD, p=0.002) and a significant inverse association between arousal index and AF was observed (OR 0.65 [0.50–0.86] per SD [12/hr], p=0.005). Conclusions In a study of a large multi-ethnic population, AF was not only associated with AHI severity, but was also more common in individuals with poor sleep quality as measured by reduced SWS time, a finding that was independent of AHI.

Entropy images, representing the complexity of original fundus photographs, may strengthen the contrast between diabetic retinopathy (DR) lesions and unaffected areas. The aim of this study is to compare the detection performance for severe DR between original fundus photographs and entropy images by deep learning. A sample of 21,123 interpretable fundus photographs obtained from a publicly available data set was expanded to 33,000 images by rotating and flipping. All photographs were transformed into entropy images using block size 9 and downsized to a standard resolution of 100 × 100 pixels. The stages of DR are classified into 5 grades based on the International Clinical Diabetic Retinopathy Disease Severity Scale: Grade 0 (no DR), Grade 1 (mild nonproliferative DR), Grade 2 (moderate nonproliferative DR), Grade 3 (severe nonproliferative DR), and Grade 4 (proliferative DR). Of these 33,000 photographs, 30,000 images were randomly selected as the training set, and the remaining 3,000 images were used as the testing set. Both the original fundus photographs and the entropy images were used as the inputs of convolutional neural network (CNN), and the results of detecting referable DR (Grades 2–4) as the outputs from the two data sets were compared. The detection accuracy, sensitivity, and specificity of using the original fundus photographs data set were 81.80%, 68.36%, 89.87%, respectively, for the entropy images data set, and the figures significantly increased to 86.10%, 73.24%, and 93.81%, respectively (all p values <0.001). The entropy image quantifies the amount of information in the fundus photograph and efficiently accelerates the generating of feature maps in the CNN. The research results draw the conclusion that transformed entropy imaging of fundus photographs can increase the machinery detection accuracy, sensitivity, and specificity of referable DR for the deep learning-based system.