Objective To automatically segment diabetic retinal exudation features from deep learning color fundus images. Methods An applied study. The method of this study is based on the U-shaped network model of the Indian Diabetic Retinopathy Image Dataset (IDRID) dataset, introduces deep residual convolution into the encoding and decoding stages, which can effectively extract seepage depth features, solve overfitting and feature interference problems, and improve the model's feature expression ability and lightweight performance. In addition, by introducing an improved context extraction module, the model can capture a wider range of feature information, enhance the perception ability of retinal lesions, and perform excellently in capturing small details and blurred edges. Finally, the introduction of convolutional triple attention mechanism allows the model to automatically learn feature weights, focus on important features, and extract useful information from multiple scales. Accuracy, recall, Dice coefficient, accuracy and sensitivity were used to evaluate the ability of the model to detect and segment the automatic retinal exudation features of diabetic patients in color fundus images. Results After applying this method, the accuracy, recall, dice coefficient, accuracy and sensitivity of the improved model on the IDRID dataset reached 81.56%, 99.54%, 69.32%, 65.36% and 78.33%, respectively. Compared with the original model, the accuracy and Dice index of the improved model are increased by 2.35% , 3.35% respectively. Conclusion The segmentation method based on U-shaped network can automatically detect and segment the retinal exudation features of fundus images of diabetic patients, which is of great significance for assisting doctors to diagnose diseases more accurately.
ObjectiveTo investigate the prevalence and risk factors of tessellation fundus (TF) among Tianjin Medical University students with different refractive statuses. MethodsA cross-sectional study. From September to December 2019, 346 students from Tianjin Medical University were randomly selected and underwent slit-lamp examination, non-cycloplegic auto-refraction, subjective refraction, best-corrected visual acuity, ocular biometric measurement, and non-dilation fundus photography. The differences in the prevalence of TF in basic characteristics and ocular biometric parameters were compared. Based on the equivalent spherical (SE), refractive status was divided into the non-myopia group (SE＞－0.50 D) and the myopia group (SE≤－0.50 D). The myopia group was further divided into mild myopia group (－3.00 D＜SE≤－0.50 D), moderate myopia group (－6.00 D＜SE≤－3.00 D), and high myopia group (SE≤－6.00 D). According to the axis length (AL), the subjects were divided into AL＜24 mm group, 24-26 mm group, and ＞26 mm group. The logistic regression was used to analyze the risk factors affecting TF. Trend tests were performed for each risk factor and TF. ResultsOf the 346 subjects, 324 (93.6%, 324/346) were myopia, of whom 73 (21.1%, 73/346), 167 (48.3%, 167/346), and 84 (24.3%, 84/346) were mild myopia, moderate myopia, and high myopia, respectively; 22 (6.4%, 22/346) were non-myopia. There were 294 (85.0%, 294/346) students with TF in the macula, including 9 (40.91%, 9/22), 58 (79.45%, 58/73), 145 (86.83%, 145/167), and 82 (97.62%, 82/84) in non-myopia, low myopia, moderate myopia, and high myopia group, respectively; 52 (15.0%, 52/346) students were without TF in the macula. There were statistically significant gender differences (χ2=4.47), SE (t=6.29), AL (t=－8.29), anterior chamber depth (Z=－2.62), lens thickness (Z=－2.23), and average corneal radius (Z=－3.58) between students with and without TF in the macula (P＜0.05). Spherical equivalent and axial length were independent risk factors for TF and its severity (P≤0.001). With an increasing degree of myopia, and increasing axial length, the risk of TF increased (P for trend＜0.001). ConclusionsThe prevalence of TF is 85.0% among Tianjin Medical University students. TF is detected in the fundus of no myopia, mild myopia, moderate myopia and high myopia. The degree of myopia is higher, the AL is longer, the possibility of TF is higher.