TL;DR
This paper introduces a novel, safe, and operator-independent quantitative cervical elastography system that combines transvaginal ultrasound imaging with stress measurement. This system accurately quantifies cervical softening during pregnancy, offering a promising tool for predicting preterm birth and understanding cervical remodeling.
What Problem Does This Paper Address?
Preterm birth is a major cause of neonatal morbidity and mortality, and accurately predicting it remains a significant clinical challenge. While cervical ripening (softening) is a key indicator, current methods like manual palpation are subjective and have limited predictive power. Existing ultrasound elastography techniques, such as strain elastography and shear wave elastography, have limitations: strain elastography lacks stress information for session-to-session comparisons, and shear wave elastography struggles with the heterogeneous nature of cervical tissue and can be operator-dependent. This paper aims to overcome these limitations by developing a quantitative, operator-independent, and safe-for-pregnancy system that measures both stress and strain in the cervix.
Methods
The core of this work is a quantitative cervical elastography system built upon a transvaginal ultrasound imaging setup. The system integrates a custom-designed, waterproofed stress sensor attached to the ultrasound probe to simultaneously record probe-surface stress. A correlation-based automatic feature tracking algorithm is employed to quantify cervical deformation from the ultrasound B-mode videos, from which strain is derived. After each imaging session, the stress sensor is calibrated against a load cell to convert its measurements into true stress. Finally, a linear regression is applied to the stress-strain data to estimate the cervical Young's modulus, a measure of tissue stiffness. The system includes a graphical user interface (GUI) to provide real-time feedback on measurement quality to the sonographer.
Datasets
The study utilized data from pregnant participants. Specifically, 22 participants were included, with three or more imaging visits each, and their last visit occurring after 34 weeks of gestation. Phantom experiments using four gelatin phantoms of varying concentrations (70, 90, 110, and 130 g/L) were also conducted to validate the system's accuracy, repeatability, and reproducibility.
Key Insights and Ideas
- Bridging the Gap in Elastography: The paper successfully integrates stress measurement with ultrasound imaging to overcome the limitations of existing elastography techniques, enabling quantitative and reproducible assessments of cervical elasticity.
- Safety and Practicality: The system is designed to be safe for use during pregnancy, utilizing a minimally modified transvaginal ultrasound probe that is already standard in obstetric care, making it familiar and accessible.
- Operator Independence: The development of an automatic feature tracking algorithm for strain quantification and the simultaneous stress measurement aim to reduce operator dependency, a significant advantage for clinical applications.
- Longitudinal Tracking of Cervical Softening: The system's ability to perform longitudinal measurements allows for the tracking of cervical softening throughout pregnancy, providing valuable insights into normal physiological changes and potential deviations.
Key Results
- Phantom Validation: The system demonstrated high accuracy and robustness in phantom experiments, showing good agreement with standard quasi-static compression methods. Repeatability and reproducibility were also confirmed.
- Cervical Softening During Pregnancy: In a pilot study of pregnant participants, the system detected a significant decrease in cervical stiffness (Young's modulus) as pregnancy progressed. The cervical Young's modulus was found to decrease by approximately 3.95% per week.
- Trimester-Specific Moduli: Geometric mean values of cervical Young's moduli were estimated for different trimesters: 13.07 kPa (1st), 7.59 kPa (2nd), and 4.40 kPa (3rd).
- High Measurement Quality: The system achieved a measurement quality requirement of PCC ≥ 0.8 for both stress-strain correlation (rss) and sensor calibration (rcal ≥ 0.85), indicating reliable data.
Limitations and Future Work
The authors acknowledge that the overall success rate of measurements meeting quality requirements was 54.7%. Future work includes improving the precision of Young's modulus estimation for specific anatomical regions of the cervix to map localized changes. Further optimization of the stress calibration system, simplification of sensor application, exploration of multiple sensors, and the integration of finite element mechanical modeling are also suggested to enhance efficiency and scalability. The technology's applicability to research and clinical uses, such as quantifying risks for preterm birth and cervical insufficiency, is highlighted.
Summary updated