Skip to main content
. 2022 Aug 8;11(8):7. doi: 10.1167/tvst.11.8.7

Figure 5.

Figure 5.

Typical examples of AOSLO-net outperforming nnUNet on detecting the parenting vessels of MAs when they are trained using (a) short mask set and (b) thick mask set. First column: enhanced AOSLO images. Second column: normal masks used to train the models. Third column: segmentation results of nnUNet. Fourth column: segmentation results of AOSLO-net. Numbers in images are the Dice scores. (a) AOSLO-net are capable of detecting the parenting vessel of MAs in images with ID 055, 082, 088 and 094, whereas nnUNet misses one of the parenting vessels(marked in yellow circles). Particularly, nnUNet again mistakenly predicts the vessel on the upper left for the case of ID 055. (b) AOSLO-net are capable of detecting the parenting vessel of MAs in images with ID 038, 046, 055 and 102, whereas nnUNet misses one of the parenting vessels(marked in yellow circles). In case of ID 055, implementation of thick mask set improves the segmentation results of AOSLO-net, but not for nnUNet.