Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Pubic Symphysis-Fetal Head Segmentation Network Using BiFormer Attention Mechanism and Multipath Dilated Convolution

Pengzhou Cai, Lu Jiang, Yanxin Li, Xiaojuan Liu, Libin Lan

TL;DR

A novel method, named BRAU-Net is proposed to solve the pubic symphysis-fetal head segmentation task, which adopts a U-Net-like encoder-decoder architecture with bi-level routing attention and skip connections, which effectively learns local-global semantic information.

Abstract

Pubic symphysis-fetal head segmentation in transperineal ultrasound images plays a critical role for the assessment of fetal head descent and progression. Existing transformer segmentation methods based on sparse attention mechanism use handcrafted static patterns, which leads to great differences in terms of segmentation performance on specific datasets. To address this issue, we introduce a dynamic, query-aware sparse attention mechanism for ultrasound image segmentation. Specifically, we propose a novel method, named BRAU-Net to solve the pubic symphysis-fetal head segmentation task in this paper. The method adopts a U-Net-like encoder-decoder architecture with bi-level routing attention and skip connections, which effectively learns local-global semantic information. In addition, we propose an inverted bottleneck patch expanding (IBPE) module to reduce information loss while performing up-sampling operations. The proposed BRAU-Net is evaluated on FH-PS-AoP and HC18 datasets. The results demonstrate that our method could achieve excellent segmentation results. The code is available on GitHub.

Pubic Symphysis-Fetal Head Segmentation Network Using BiFormer Attention Mechanism and Multipath Dilated Convolution

TL;DR

A novel method, named BRAU-Net is proposed to solve the pubic symphysis-fetal head segmentation task, which adopts a U-Net-like encoder-decoder architecture with bi-level routing attention and skip connections, which effectively learns local-global semantic information.

Abstract

Pubic symphysis-fetal head segmentation in transperineal ultrasound images plays a critical role for the assessment of fetal head descent and progression. Existing transformer segmentation methods based on sparse attention mechanism use handcrafted static patterns, which leads to great differences in terms of segmentation performance on specific datasets. To address this issue, we introduce a dynamic, query-aware sparse attention mechanism for ultrasound image segmentation. Specifically, we propose a novel method, named BRAU-Net to solve the pubic symphysis-fetal head segmentation task in this paper. The method adopts a U-Net-like encoder-decoder architecture with bi-level routing attention and skip connections, which effectively learns local-global semantic information. In addition, we propose an inverted bottleneck patch expanding (IBPE) module to reduce information loss while performing up-sampling operations. The proposed BRAU-Net is evaluated on FH-PS-AoP and HC18 datasets. The results demonstrate that our method could achieve excellent segmentation results. The code is available on GitHub.

Paper Structure

This paper contains 19 sections, 6 equations, 5 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Ultrasonic Image Segmentation
  4. Upsampling Methods
  5. Method
  6. BRAU-Net Architecture
  7. BiFormer Block
  8. Inverted Bottleneck Patch Expanding Module (IBPE)
  9. Loss Function
  10. Experiments
  11. Datasets
  12. Evaluation Metrics
  13. Implementation Details
  14. Comparison with Other Methods
  15. Ablation Study
  16. ...and 4 more sections

Figures (5)

  • Figure 1: (A) The fetal head-pubic symphysis segmentation challenge. (B) A transperineal ultrasound image exhibits the pubic symphysis (PS) and fetal head (FH). (C) The mask of the PS (in grey) and FH (in white).
  • Figure 2: The architecture of our proposed BRAU-Net. IBPE, LN and DW represent inverted bottleneck patch expanding, layer normalization and depth-wise convolution, respectively.
  • Figure 3: The visual segmentation results of our method and others on the FH-PS-AoP and HC18 datasets. The results of our method are closer to ground truth.
  • Figure 4: The visualization of back-propagated gradient activation maps using Grad-CAM for the final convolution layer on the PS-FH-AoP dataset.
  • Figure 5: Comparisons of efficiency vs performance of BRAU-Net against other methods.