Automated Ultrasonographic Measurement of Basic Fetal Growth Parameters

Abstract

The goal of our study was to compare the accuracy of automated ultrasonographic measurements of fetal biometric parameters with those achieved manually. Manual and automated fetal biometric measurements were performed in 83 patients between 11 + 0 and 40 + 6 weeks of gestation, regardless of BMI, fetal position, placental location and quantity of amniotic fluid. All patients were examined using Siemens Healthcare Ultrasound equipment, Syngo auto OB measurement, with a 6C2 probe (2.5 - 6 MHz) in a 2 D approach. The biometric measurement evaluation included the following parameters: crown-rump length (CRL), biparietal diameter (BPD), occipitofrontal diameter (OFD), head circumference (HC), abdominal circumference (AC), femur length (FL), and humerus length (HL). All patients were divided into four distinct groups according to the gestational age: Group 1: 11 + 0 to 14 + 6 weeks, Group 2: 15 + 0 to 23 + 6 weeks, Group 3: 24 + 0 to 34 + 6 weeks, and Group 4: 35 + 0 to 40 + 6 weeks. The results of the automated measurements were classified in four different groups: Score 1: the automatic measurements were within a range of ± 5 % of the manual measurements, Score 2: the automatic measurements varied within ± 10 % from those done manually, Score 3: the variability range was more than 10 % but less than 50 %, and Score 4: was related to an incorrect measurement. Comparing the automated with the manual measurements, Score 1 was found for 41.74 % of all biometric parameters. 24.37 % of the automated measurements were within Score 2, 11.65 % of the automated measurements showed Score 3 and 22.24 % had Score 4. Score 1 + 2 results were found in more than 62 % of cases for CRL, BPD, OFD, AC, and FL, followed by 55.91 % for HC measurements. The lowest result of 50.45 % was found for HL. Looking in detail at the four gestational age groups, a high range of variation was identified for the automatic biometric results for the different growth parameters: Group 1 had very good or good automatic results (Score 1 + 2) for CRL (76.93 %) and AC (76.47 %). Unacceptable measurements (= Score 3 + 4) were found for BPD (58.82 %), HC (64.71 %) and HL (82.35 %). Group 2 showed very good or good automatic results (Score 1 + 2) for BPD (90 %), OFD (83.34 %), HC (80.00 %), AC (80.00 %), FL (83.33) and HL (80.00 %). Score 3 + 4 results were found in less than 20 % of the cases in this group. Group 3 showed very good or good automatic results (Score 1 + 2) for BPD (87.50 %), OFD (79.17 %), and FL (87.50 %). Score 3 + 4 results were found high to be high for AC (41.66 %), and HL (37.50 %). Group 4 revealed very good or good automatic results (Score 1 + 2) only for OFD (83.34 %), while Score 3 + 4 were found high for BPD (41.67 %), HC (58.33 %), AC (50 %), FL (58.33 %) and HL (58.33 %). The comparison of all gestational age groups showed the best results for automatic measurements to be in the second group (15 + 0 to 23 + 6 weeks). Humerus length was the parameter with the highest unacceptable automated measurements in all 4 groups. Syngo auto OB measurement enables rapid automated measurement of fetal biometric parameters. Despite the fact that a number of problems remain to be solved, the described method may have a significant impact on daily workflow efficiency in fetal biometry.