Voluson e8

The ultrasound examinations were performed with the use of the GE Voluson E8, GE Voluson E10, and Philips iU22 ultrasound machines with convex transabdominal transducers. Fetal ear auricle length was measured from the tip of the helix to the end of the lobe in the longitudinal view. This measurement was performed during a detailed fetal ultrasound evaluation and the value was added to 80% of the unit’s database by one physician (MRL).

Our dataset consisted of 2694 2D ultrasound examinations of volunteers with gestational ages between 18–22 weeks which have been acquired and labelled during routine screenings by a team of 45 expert sonographers according to the guidelines set out in the UK FASP handbook [22] . Those guidelines only define the planes which need to be visualised, but not the sequence in which they should be acquired. The large number of sonographers involved means that the dataset contains a large number of different operator-dependent examination “styles” and is therefore a good approximation of the normal variability observed between different sonographers. In order to reflect the distribution of real data, no selection of the cases was made based on normality or abnormality. Eight different ultrasound systems of identical make and model (GE Voluson E8) were used for the acquisitions. For each scan we had access to freeze-frame images saved by the sonographers during the exam. For a majority of cases we also had access to screen capture videos of the entire fetal exam.

Fetal muscle ultrasound recordings were performed using a General Electric Healthcare Voluson E8 ultrasound machine. To avoid a bias by pregnancy-related signal disturbances (such as the maternal subcutaneous fat layer and/or the intra-uterine fetal conditions), we expressed fetal-MUD as a ratio between muscle- and bone- density: fetal-MUD-ratio = [mean muscle pixel value] / [mean bone pixel value] [16 (link)]. For analysis of the fetal muscle, we selected the whole muscle in a longitudinal section as the region of interest, see S1A Fig. The fetal-MUD-ratio was assessed for the biceps (reference bone: humerus), quadriceps (reference bone: femur), tibialis anterior (reference bone: tibia), hamstrings (reference bone: femur), gluteus (reference bone: hip) and calf muscles (reference bone: tibia or fibula). From each set of five images per muscle per fetus, we derived one data point by excluding the highest and lowest value and calculating the mean of the remaining three MUD values.

The transabdominal ultrasound was performed using three different ultrasound devices: Voluson S8 (GE Healthcare, Chicago, Illinois), Voluson E8 (GE Healthcare, Chicago, Illinois) and Phillips iU22 (Phillips, Seattle, Washington). The ultrasound was carried out by qualified observers upon routine conditions and guidelines. Second and third trimester biometry was performed by measuring the abdominal circumference (AC), the biparietal diameter (BPD), head circumference (HC) and the femur length (FL). The fetal weight and the weight percentile were calculated using the Hadlock curves [16 (link)]. The following Doppler parameters were measured: PI of the uterine arteries (UtA), PI of the umbilical artery (UA), PI of the middle cerebral artery (MCA), PI of the ductus venosus (DV) and the cerebroplacental ratio (CPR) as the ratio between MCA PI and UA PI.

The type of endometrial preparation was determined according to the experience of the physician, based on patients’ characteristics, including natural cycle (NC), hormone replacement therapy (HRT) and GnRH agonist combined with HRT (GnRH agonist-HRT) protocol. In short, women with regular menstrual cycles were allocated to NC, while patients having irregular cycles were offered either HRT or GnRH agonist-HRT protocol. The detailed protocols for endometrial preparation were described in our previous studies (23 (link)). To guarantee the accuracy and reliability as possible, EMT was measured by highly trained and experienced sonographers of the same team via Voluson E8 (GE Healthcare, Australia) with intracavity probes. We identified EMT by the largest diameter from one endometrial–myometrial interface to the other in the midsagittal plane. In NC cycles, EMT was measured on the day of hCG administration, while in women with HRT or GnRH agonist-HRT protocol, EMT was recorded from the last ultrasound prior progesterone initiation. Patients were categorized into three groups according to EMT: ≤7.5 mm, 7.5-12 mm and >12 mm, and 7.5-12 mm served as a reference group. These thresholds were selected based on the previous studies (12 (link), 14 (link), 24 (link)).

Gestational weeks (11-40 weeks’ gestational age) were corrected for the last menstrual period and for the results of the ultrasonographic examination. According to the median distribution of measured gestational weeks, we divided pregnancy into 3 periods with 24 and 34 weeks’ gestational age as the nodes (ie, <24, 24-34, and >34 weeks’ gestational age). Data on fetal biometrics (abdominal circumference [AC], head circumference [HC], humerus length, femur length [FL], and biparietal diameter [BPD]) (eTable 1 in the Supplement) and birth anthropometrics (length and weight) were retrieved from the medical records. Fetal biometrics had multiple ultrasonographic values (median, 4 [IQR, 1-6]) that were measured (using Voluson E8; GE Healthcare) by trained sonographers.^{25 (link)} The estimated fetal weight (EFW) was calculated using the formula of Hadlock et al^{26 (link)} based on AC, FL, and HC. Small size for gestational age was defined as a birth weight in the 10th percentile or lower, and large size for gestational age (LGA) was defined as a birth weight in the 90th percentile or higher according to birth weight reference percentiles for Chinese citizens.^{27 (link)} Macrosomia implied growth beyond an absolute birth weight that is historically 4000 g, whereas a birth weight lower than 2500 g was defined as low birth weight.