Application of Improved U-Net Convolutional Neural Network for Automatic Quantification of the Foveal Avascular Zone in Diabetic Macular Ischemia

<div>Visualization results for FAZ segmentation. (a)–(c) are the original DR map of different severity. (d)–(f) are manually labeled FAZ area as ground truth. (g)–(i) are the prediction result of traditional U-Net. (j)–(l) are the prediction result of the proposed method. The green, red, and yellow regions represent the ground truth, the predicted area, and the overlap of the two areas, respectively.</div>

Journal of Diabetes Research

fig3

Figure 3

Figure 3: Application of Improved U-Net Convolutional Neural Network for Automatic Quantification of the Foveal Avascular Zone in Diabetic Macular Ischemia 

Figure 3 | Application of Improved U-Net Convolutional Neural Network for Automatic Quantification of the Foveal Avascular Zone in Diabetic Macular Ischemia 