Multi-level Attention Network for Segmenting Retinal Vessels

PROJECT TITLE :

Multi-level Attention Network for Retinal Vessel Segmentation

ABSTRACT:

In the screening, diagnosis, treatment, and evaluation of a variety of cardiovascular and ophthalmologic diseases, automatic vessel segmentation in the fundus images plays an important role. However, retinal vessel segmentation has been a problem for a long time because there is a lack of data that has been adequately annotated, the size of the vessels varies, and the structures of the vessels are intricate. In this paper, a novel Deep Learning model known as AACA-MLA-D-.Net is proposed. Its purpose is to fully utilize the low-level detailed information and the complementary information encoded in different layers in order to accurately distinguish the vessels from the background while maintaining a low level of model complexity. The architecture of the proposed model is based on U.Net, and the dropout dense block is proposed as a means to both mitigate the overfitting problem and preserve the maximum amount of vessel information that exists between convolution layers. The module for adaptive atrous channel attention is incorporated into the contracting path so that the relative importance of each feature channel can be determined automatically. After that, the multi-level attention module is proposed to integrate the multi-level features extracted from the expanding path and use them to refine the features at each individual layer using attention mechanism. This module is intended to integrate the multi-level features extracted from the expanding path. The proposed method has been validated on the three databases that are readily available to the general public; these are the DRIVE, STARE, and CHASE DB1 databases. The findings of the experiments show that the proposed method is capable of achieving better or comparable performance on retinal vessel segmentation while simultaneously reducing the complexity of the model. In addition, the method that has been suggested is capable of dealing with some difficult cases and has strong generalization ability.

Did you like this research project?

To get this research project Guidelines, Training and Code... Click Here