Determining Fetal Orientations From Blind Sweep Ultrasound Video

TL;DR

This work addresses the cognitive burden of fetal ultrasound by predicting coarse fetal orientations from blind-sweep videos to assist sonographers. It introduces an end-to-end pipeline that leverages a pre-trained PCBM for head detection/segmentation, template-based fetal presentation classification, and a geometry-based fetal lie predictor derived from head anatomy, with a thalamus-CSP-based method and a thalamus-only fallback. Evaluations on a 13-subject third-trimester dataset show promising results for presentation and lie predictions in evaluable cases, highlighting the influence of head visibility on performance. The proposed approach aims to augment clinical workflow with continuous orientation cues and points toward adaptive acquisition and real-time integration to enhance efficiency and decision support in obstetric care.

Abstract

Cognitive demands of fetal ultrasound examinations pose unique challenges among clinicians. With the goal of providing an assistive tool, we developed an automated pipeline for predicting fetal orientation from ultrasound videos acquired following a simple blind sweep protocol. Leveraging on a pre-trained head detection and segmentation model, this is achieved by first determining the fetal presentation (cephalic or breech) with a template matching approach, followed by the fetal lie (facing left or right) by analyzing the spatial distribution of segmented brain anatomies. Evaluation on a dataset of third-trimester ultrasound scans demonstrated the promising accuracy of our pipeline. This work distinguishes itself by introducing automated fetal lie prediction and by proposing an assistive paradigm that augments sonographer expertise rather than replacing it. Future research will focus on enhancing acquisition efficiency, and exploring real-time clinical integration to improve workflow and support for obstetric clinicians.

Figures (4)

• Figure 1: (a-d) The four fetal orientations. (e) Blind sweep protocol across the pregnant abdomen. The sweeps start from the center marked with V1, followed by two on the mother's left (V2, V3) and finally two on the right (V4, V5).
• Figure 2: The process of estimating fetal lie depicted from left to right. The top row illustrates the main approach of utilizing both thalamus and CSP, while the bottom row illustrates the thalamus-only fallback approach.
• Figure 3: Presentation prediction visualized with sweeps from 2 subjects. The temporal softmax probability for head from PCBM is indicated in blue, while the cephalic and breech template functions (see \ref{['eq:presentation_pattern_template']}) are indicated in gray.
• Figure 4: Lie prediction visualized for 9 subjects, with ground-truth in the captions.