Time quantified detection of fetal movements using a new fetal movement algorithm

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Obstetrics

Background:

• Accurate fetal movement characterization is crucial for assessing fetal well-being.
• Existing automated fetal movement detection systems have limitations in performance and quantification.
• Real-time ultrasound imaging provides a valuable tool for monitoring fetal activity.

Purpose of the Study:

• To develop an automated ultrasound system for enhanced fetal movement detection and characterization.
• To create an improved methodology for quantifying the performance of fetal movement detection algorithms.
• To compare the efficacy of a novel algorithm (Russell algorithm) against a commercial system (HP M-1350-A).

Main Methods:

• Recorded 20-minute fetal movement segments from 101 patients using ultrasound.
• Utilized a UAMS-developed algorithm (Russell algorithm) and a Hewlett-Packard (HP) M-1350-A system.
• Expert examiners reviewed videotaped ultrasound images, scoring movements on a second-by-second basis for comparison.

Main Results:

• The Russell algorithm detected 95.53% of discrete movements >5 seconds, significantly outperforming the HP system (86.08%, p=0.012).
• Both systems showed lower sensitivity for short discrete movements (<5 seconds), with the Russell algorithm at 57.39% and the HP system at 35.22%.
• Neither system fully identified all fetal movements based on the second-per-second scoring system.

Conclusions:

• The Russell algorithm offers improved performance for detecting and characterizing fetal movements compared to the HP M-1350-A.
• Enhanced quantification methods confirm the superior performance of the Russell algorithm.
• Further development is needed to achieve complete identification of all fetal movements on a second-by-second basis.