Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

EViT-Unet: U-Net Like Efficient Vision Transformer for Medical Image Segmentation on Mobile and Edge Devices

Xin Li, Wenhui Zhu, Xuanzhao Dong, Oana M. Dumitrascu, Yalin Wang

TL;DR

EviT-UNet is proposed, an efficient ViT-based segmentation network that reduces computational complexity while maintaining accuracy, making it ideal for resource-constrained medical devices.

Abstract

With the rapid development of deep learning, CNN-based U-shaped networks have succeeded in medical image segmentation and are widely applied for various tasks. However, their limitations in capturing global features hinder their performance in complex segmentation tasks. The rise of Vision Transformer (ViT) has effectively compensated for this deficiency of CNNs and promoted the application of ViT-based U-networks in medical image segmentation. However, the high computational demands of ViT make it unsuitable for many medical devices and mobile platforms with limited resources, restricting its deployment on resource-constrained and edge devices. To address this, we propose EViT-UNet, an efficient ViT-based segmentation network that reduces computational complexity while maintaining accuracy, making it ideal for resource-constrained medical devices. EViT-UNet is built on a U-shaped architecture, comprising an encoder, decoder, bottleneck layer, and skip connections, combining convolutional operations with self-attention mechanisms to optimize efficiency. Experimental results demonstrate that EViT-UNet achieves high accuracy in medical image segmentation while significantly reducing computational complexity.

EViT-Unet: U-Net Like Efficient Vision Transformer for Medical Image Segmentation on Mobile and Edge Devices

TL;DR

EviT-UNet is proposed, an efficient ViT-based segmentation network that reduces computational complexity while maintaining accuracy, making it ideal for resource-constrained medical devices.

Abstract

With the rapid development of deep learning, CNN-based U-shaped networks have succeeded in medical image segmentation and are widely applied for various tasks. However, their limitations in capturing global features hinder their performance in complex segmentation tasks. The rise of Vision Transformer (ViT) has effectively compensated for this deficiency of CNNs and promoted the application of ViT-based U-networks in medical image segmentation. However, the high computational demands of ViT make it unsuitable for many medical devices and mobile platforms with limited resources, restricting its deployment on resource-constrained and edge devices. To address this, we propose EViT-UNet, an efficient ViT-based segmentation network that reduces computational complexity while maintaining accuracy, making it ideal for resource-constrained medical devices. EViT-UNet is built on a U-shaped architecture, comprising an encoder, decoder, bottleneck layer, and skip connections, combining convolutional operations with self-attention mechanisms to optimize efficiency. Experimental results demonstrate that EViT-UNet achieves high accuracy in medical image segmentation while significantly reducing computational complexity.

Paper Structure

This paper contains 10 sections, 6 equations, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Method
  3. Architecture overview
  4. Efficientformer block
  5. Downsampling and Upsampling
  6. Skip Connection
  7. Experiments And Results
  8. Conclusion And Discussion
  9. COMPLIANCE WITH ETHICAL STANDARDS
  10. ACKNOWLEDGMENTS

Figures (3)

  • Figure 1: Comparison of model size, FLOPs, and performance (Dice on Synapse dataset). The area of each circle is proportional to the number of parameters (model size).
  • Figure 2: (a) The architecture of EViT-Unet, which is composed of encoder, bottleneck, decoder, and skip connections. "Local" blocks use convolution, while "Global+Local" blocks use the combination of convolution and self-attention. (b) The details of "Local" blocks. "DW.Conv," is the depthwise convolution mobilenets. (c) The details of "Global+Local" blocks.
  • Figure 3: Comparison of segmentation results in Synapse(A), Glas(B) and MoNuSeg(C) dataset.