UNet and Transformers: Deep Learning Based Methods for Medical Image Segmentation

Zhirui Ren

2024

Abstract

As a vital sub-part of medical image analysis and processing, image segmentation is a time-consuming and heavily experienced task when performed manually. With the revolutionary development of artificial intelligence (AI), intended to utilize the high efficiency and reliability of computerized information processing to address the problem of increasingly large quantities of medical images waiting to be processed, deep learning-based methods for segmentation tasks have become popular. Convolutional Neural Networks (CNNs) are an old leader in the computer vision community, but as transformer models have obtained excellent results in the field of natural language processing (NLP), increasing researchers have begun to explore whether they can also bring significant breakthroughs for image processing. In this review, some evaluation metrics are first to be introduced. Subsequently, the introduction of the core ATTENTION mechanism of transformers and three selected models with their performance follows. Through the survey, using the mature UNet method alone, good accuracy can be achieved, and if combined with the superiority of transformers’ global context capture ability, even better results can be obtained. Dedicated to promoting the birth of a generalized image model with high accuracy, this article is provided for researchers’ reference.

Paper Citation

in Harvard Style

Ren Z. (2024). UNet and Transformers: Deep Learning Based Methods for Medical Image Segmentation. In Proceedings of the 1st International Conference on Data Science and Engineering - Volume 1: ICDSE; ISBN 978-989-758-690-3, SciTePress, pages 547-551. DOI: 10.5220/0012838300004547

in Bibtex Style

@conference{icdse24,
author={Zhirui Ren},
title={UNet and Transformers: Deep Learning Based Methods for Medical Image Segmentation},
booktitle={Proceedings of the 1st International Conference on Data Science and Engineering - Volume 1: ICDSE},
year={2024},
pages={547-551},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012838300004547},
isbn={978-989-758-690-3},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Data Science and Engineering - Volume 1: ICDSE
TI - UNet and Transformers: Deep Learning Based Methods for Medical Image Segmentation
SN - 978-989-758-690-3
AU - Ren Z.
PY - 2024
SP - 547
EP - 551
DO - 10.5220/0012838300004547
PB - SciTePress