Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Pose-GuideNet: Automatic Scanning Guidance for Fetal Head Ultrasound from Pose Estimation

Qianhui Men, Xiaoqing Guo, Aris T. Papageorghiou, J. Alison Noble

TL;DR

Pose-GuideNet addresses the challenge of guiding fetal head ultrasound to standard planes by estimating the 3D pose of freehand 2D US frames relative to a 3D atlas. It employs a two-stage learning framework with a ResNeXt50-based PoseE backbone, atlas-based Dice and pose losses, and a cross-modality scheme that combines geometry-guided in-plane alignment with semantic-aware out-of-plane alignment to propagate pose across frames. The approach demonstrates sensor-free, pose-aware guidance for TVP and TCP biometry in the fetal head, with quantitative improvements in both motion-level correspondence and image-level similarity over baselines. The work enables automatic SP guidance and training support for sonographers without external tracking, with future work to convert 3D pose estimates into actionable probe movements for end-to-end SP acquisition.

Abstract

3D pose estimation from a 2D cross-sectional view enables healthcare professionals to navigate through the 3D space, and such techniques initiate automatic guidance in many image-guided radiology applications. In this work, we investigate how estimating 3D fetal pose from freehand 2D ultrasound scanning can guide a sonographer to locate a head standard plane. Fetal head pose is estimated by the proposed Pose-GuideNet, a novel 2D/3D registration approach to align freehand 2D ultrasound to a 3D anatomical atlas without the acquisition of 3D ultrasound. To facilitate the 2D to 3D cross-dimensional projection, we exploit the prior knowledge in the atlas to align the standard plane frame in a freehand scan. A semantic-aware contrastive-based approach is further proposed to align the frames that are off standard planes based on their anatomical similarity. In the experiment, we enhance the existing assessment of freehand image localization by comparing the transformation of its estimated pose towards standard plane with the corresponding probe motion, which reflects the actual view change in 3D anatomy. Extensive results on two clinical head biometry tasks show that Pose-GuideNet not only accurately predicts pose but also successfully predicts the direction of the fetal head. Evaluations with probe motions further demonstrate the feasibility of adopting Pose-GuideNet for freehand ultrasound-assisted navigation in a sensor-free environment.

Pose-GuideNet: Automatic Scanning Guidance for Fetal Head Ultrasound from Pose Estimation

TL;DR

Pose-GuideNet addresses the challenge of guiding fetal head ultrasound to standard planes by estimating the 3D pose of freehand 2D US frames relative to a 3D atlas. It employs a two-stage learning framework with a ResNeXt50-based PoseE backbone, atlas-based Dice and pose losses, and a cross-modality scheme that combines geometry-guided in-plane alignment with semantic-aware out-of-plane alignment to propagate pose across frames. The approach demonstrates sensor-free, pose-aware guidance for TVP and TCP biometry in the fetal head, with quantitative improvements in both motion-level correspondence and image-level similarity over baselines. The work enables automatic SP guidance and training support for sonographers without external tracking, with future work to convert 3D pose estimates into actionable probe movements for end-to-end SP acquisition.

Abstract

3D pose estimation from a 2D cross-sectional view enables healthcare professionals to navigate through the 3D space, and such techniques initiate automatic guidance in many image-guided radiology applications. In this work, we investigate how estimating 3D fetal pose from freehand 2D ultrasound scanning can guide a sonographer to locate a head standard plane. Fetal head pose is estimated by the proposed Pose-GuideNet, a novel 2D/3D registration approach to align freehand 2D ultrasound to a 3D anatomical atlas without the acquisition of 3D ultrasound. To facilitate the 2D to 3D cross-dimensional projection, we exploit the prior knowledge in the atlas to align the standard plane frame in a freehand scan. A semantic-aware contrastive-based approach is further proposed to align the frames that are off standard planes based on their anatomical similarity. In the experiment, we enhance the existing assessment of freehand image localization by comparing the transformation of its estimated pose towards standard plane with the corresponding probe motion, which reflects the actual view change in 3D anatomy. Extensive results on two clinical head biometry tasks show that Pose-GuideNet not only accurately predicts pose but also successfully predicts the direction of the fetal head. Evaluations with probe motions further demonstrate the feasibility of adopting Pose-GuideNet for freehand ultrasound-assisted navigation in a sensor-free environment.
Paper Structure (11 sections, 3 equations, 3 figures, 1 table)

This paper contains 11 sections, 3 equations, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Methods
  3. Pose Localization on 3D Atlas
  4. Cross-Dimension Cross-Modality Alignment on 2D US
  5. Experiments
  6. Data and Experimental Settings
  7. Evaluations with Motion-level Correspondence
  8. Evaluations with Image-level Correspondence
  9. Conclusion
  10. Acknowledgments.
  11. Disclosure of Interests.

Figures (3)

  • Figure 1: Principle of Pose-GuideNet. The system provides guidance on how to move to a SP of 3D anatomy, and the corresponding probe movement would be inferred through geometric mapping from 3D anatomy to the probe coordinate.
  • Figure 2: Overview of Pose-GuideNet for pose localization of fetal head.
  • Figure 3: The retrieved atlas planes of two example 2D head biometry acquisitions.