U-Net with Dilated Convolutions and Channel Attention for Cell Segmentation

Minz, Abhishek (2023) U-Net with Dilated Convolutions and Channel Attention for Cell Segmentation. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration manual) Download (784kB) | Preview

Abstract

The process of manual cell segmentation is a time-intensive and tedious task, frequently influenced by the subjective judgment of the operator. Consequently, the application of deep learning-based semantic segmentation for automating cell segmentation has proven highly beneficial in the analysis of microscopic images. The U-Net architecture, a widely used fully Convolutional Neural Network, achieves state-of-the-art results in biomedical image segmentation. However, despite its excellent performance, it does encounter certain limitations such as loss of spatial information and inconsistent segmentation of images with large variations in artifact size. To address these issues a novel convolutional neural network is proposed by modifying the backbone and skip connection of the U-Net. Non-linear filters have been used in place of simple concatenation of features to reduce the semantic gap and dilated convolutions have been used to extract multi-scale features with channel attention module to better segment images with a wider color spectrum. The proposed model outperforms the baseline U-Net by 1.77% with a mean IoU of 87.92 % and produces better segmentation masks of cellular images.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Milosavljevic, Vladimir UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science T Technology > Biomedical engineering Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence > Computer vision Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence > Computer vision Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Tamara Malone
Date Deposited:	29 Nov 2024 13:45
Last Modified:	29 Nov 2024 13:45
URI:	https://norma.ncirl.ie/id/eprint/7214

Actions (login required)

View Item