Quantitative assessment of cervical softening during pregnancy with shear wave elasticity imaging: an in vivo longitudinal study

Abstract

We report here the results of a longitudinal study of cervix stiffness during pregnancy. Thirty women, ages ranging from 19 to 37 years, were scanned with ultrasound at five time points beginning at their normal first-trimester screening (8-13 weeks) through term pregnancy (nominally 40 week) using a clinical ultrasound imaging system modified with a special ultrasound transducer and system software. The system estimated the shear wave speed (its square proportional to the shear modulus under idealized conditions) in the cervix. We found a constant fractional reduction (about 4% per week) in shear wave speed with increasing gestational age. We also demonstrated a spatial gradient in shear wave speed along the length of the cervix (softest at the distal end). Results were consistent with our previous ex vivo and in vivo work in women. Shear wave elasticity imaging may be a potentially useful clinical tool for objective assessment of cervical softening in pregnancy.

Keywords: cervix; preterm birth; shear wave elastography imaging; shear wave speed.

Figure 1.

Elevation profiles of normalized pulse intensity resulting from an ARFI excitation for the prototype linear array used in this study (left) and an endovaginal curvilinear array (right) simulated using FIELD II. There are significant side-lobes in the elevation profile of the endovaginal probe that creates complicated shear wave wavefronts and difficulty in wave tracking. (Online version in colour.)

Figure 2.

B-mode images of the cervix using a standard endovaginal probe (a–c) and the corresponding sector images of the same cervix using a small linear array probe (d–f) at the first study visit at 8–14 weeks (a,d), third study visit at 22–24 weeks (b,e), and at the last study visit at 37–39 weeks (c,f). The white arrows at the top of each image indicate the location of the bladder reflection seen in each image. The internal os is on the left side of each image and on the right side is the external os. (Online version in colour.)

Figure 3.

(a) Rectilinear B-mode image showing the original ROI selected while scanning in the solid green lines and the dotted white lines showing the manually selected ROI divided into four locations during post-processing. (b) The same rectilinear B-mode image with the SWS estimates superimposed where the values are represented by a colour scale. The white arrows indicate the location of the bladder reflection. (Online version in colour.)

Figure 4.

(a) A B-mode image of a first-trimester cervix where the cervical canal is easily visible and curves vertically toward the transducer face. (b) The curvature of the canal is outlined in white line segments and the corresponding CA measurement is shown by the straight white lines extending from the internal os. This creates unreliable shear wave behaviour. Datasets where the cervical canal was at extreme angles produced very unreliable SWS estimates and were excluded from analysis. (Online version in colour.)

Figure 5.

Box-and-whisker plots of median SWS for all subjects and measurements at each location (distance in centimetres from the internal os) versus gestational age. The width of the box indicates the range of GA at the corresponding visit.

Figure 6.

SWS estimates from each repeat SWEI measurement at each gestational age for patient 18 (a) and patient 21 (b). The circles represent the median and the whiskers are the 25th and 75th percentiles of SWS estimates in the mid-proximal (0.75 cm) region. Although these data points were outliers in the cumulative dataset (each woman had a very stiff cervix at the initial visit), the decrease in SWS follows the same trend as the other datasets. (Online version in colour.)

Figure 7.

Box-and-whisker plots of median ln(SWS) for all subjects and measurements at each location (distance in centimetres from the internal os) versus gestational age. The location specific LME model fit is shown by the solid line and the final joint LME model by the dashed line.