Analysis of measurement process of placental volume in early pregnancy: an interobserver reliability study

To examine concordance among results obtained in measurement process of first-trimester placental volume using 3D ultrasound and eXtended Imaging Virtual Organ Computed-aided AnaLysis (XI-VOCAL®, 3DXITM, Medison, Seoul, Korea) image analysis by three different operators.

Twenty first-trimester normal pregnancy cases were selected for placental volume measuring using a Medison SA 8000 Live Prime® (Medison, Seoul, Korea) with a convex volumetric multifrequency abdominal probe. Images were processed and studied independently by three operators with different grade of training. Each operator obtained 50 slices per case. Thus, this study is based on 1000 measurements that have generated four different sets of placental volume determinations based on 5, 10, 15, and 20 slices, respectively. Results of measurement process were analyzed using reliability coefficients.

There was a good degree of concordance in the placental length values obtained by all operators and not depend from the number of cuts measured [intraclass correlation coefficient (ICC)=0.734]. However, the number of cuts is important to obtain a placental volume with more accuracy. Reliability coefficients were low when determining placental volume adjusted to placental length (ICC=0.293), but combined results of the two operators that were trained in the same way showed higher coefficients of reliability (ICC=0.682), and therefore, greater concordance when comparing with the operator that was not trained in the same way.

Higher coefficients of reliability guarantee high grades of concordance in the results among operators when measuring placental volumes independently, however, contouring process introduces high variability. Training in how to best use the image analysis software effectively assists in getting higher coefficients of reliability.