Reading Diary Week 3

5 Retinal Vessel Segmentation terms

[1] P. Yin, Y. Fang, and Q. Wan, “Dual attention multiscale network for vessel segmentation in fundus photography,” Mathematics, vol. 10, no. 19, p.3687, 2022.

Global features
Which describes the visual content of the entire image by a single vector, this represent the texture, color, and shape information.
Example: The encoder–decoder structure of UNet combines low-level local features with high-level global features to produce high-resolution prediction.
Local features
Refers to a pattern or distinct structure found in an image, such as point, edge, or small image patch.
Example: The encoder–decoder structure of UNet combines low-level local features with high-level global features to produce high-resolution prediction.
Multiscale networks
This term is associated with feature extraction of an image by using convolutional blocks.
Example: we introduce the popular attention mechanisms and multiscale networks for vessel segmentation.
Receptive field
The region of the input data which is provided to the neural network.
Example: The deformable convolution block adaptively adjusts the receptive fields to capture vessels with variance shape and scale.
Residual module
A stack of layers set in such a way that the output of a layer is taken and added to another layer deeper in the block and mainly used to avoid vanishing gradient.
Example: This research introduce residual modules to the generator for better representation learning ability.