Reproducibility of 3D power Doppler placental vascular indices

Abstract

We read with great interest the article by Guimaraes Filho et al. [1], evaluating the intra and inter observer reproducibility of 3D power Doppler. Based on their results, the authors concluded that “3D PD placental vascular indices obtained using the automatic sphere mode had a good intra and inter observer reproducibility for pregnancies between 26 and 35 weeks”. However, in another recent work evaluating the same issue by Lai et al. [2], the conclusions were totally diVerent: “There is insuYcient evidence to support the meaning, reliability or reproducibility of VOCAL (VI, FI or VFI) as a tool to quantify placental perfusion despite its use in multiple publications and journal submissions. Poor reproducibility is present at the most fundamental level. Further investigation into reproducibility of placental perfusion and quantiWcation using VOCAL is required before development and application as a clinically useful tool”. Reading more carefully the papers, we observed some methodological issues in both of them that should be considered. First, one should consider that reproducibility studies evaluate diVerences observed by one observer in two or more diVerent occasions (intra observer reliability) or by two or more diVerent observers (inter observer reliability). Additionally, when 3D ultrasound is considered, one should consider diVerent levels of reproducibility: data analysis (a single 3D data-set per subject is acquired) and acquisition (the 3D data-set is acquired by the observer who will perform the analysis). The acquisition of data is fraught with much more biological and physical variations. When reading the paper by Guimaraes Filho et al. [1], one may observe that only the reproducibility of the analysis was evaluated: the 3D datasets were acquired by a single observer in only one occasion per subject. The intra and inter observer reproducibility of the analysis evaluated by intraclass correlation coeYcients was good for VI (ICC = 0.92 and 0.75; intra and inter observer), still better for VFI (ICC = 0.86 and 0.90), and excellent for VFI (ICC = 0.98 and 0.99). These results are, however, expected as any single dataset contains stable information that can be quantiWed. Even though, the results presented by Guimareas Filho et al. [1] are not meaningless and the great majority of the published papers regarding the reproducibility of 3D power Doppler evaluated only the reproducibility of the analysis of pre acquired 3D data-sets; e.g., ovarian tumors [3], thyroid gland [4], and even a previous publication about intra observer reproducibility of placenta [5]. Unfortunately, only a few studies properly evaluated reproducibility of both acquisition and analysis of 3D ultrasound measurements [6] and 3D power Doppler ultrasonography [7, 8]. In the study by Raine-Fenning et al. [8], two observers performed their own acquisitions from ovaries and uterus using Wxed Doppler settings, and a good reproducibility was found for all 3D power Doppler indices (ICC > 0.92). In the study by Rovas et al. [7], two observers acquired and evaluated their own datasets from uterine cervix of pregnant women using Wxed Doppler settings. A good intraand interobserver reproducibility was observed for both VI and VFI (intraand interobserver ICCs between 0.89 and 0.98) whilst the reproducibility found for FI was W. P. Martins (&) · C. O. Nastri Departamento de Ginecologia e Obstetricia da Faculdade de Medicina de Ribeirao Preto da, Universidade de Sao Paulo, Av. Bandeirantes 3900, 8° andar, HCRP, Campus Universitario, Ribeirao Preto, Sao Paulo 14049-900, Brazil e-mail: wpmartins@gmail.com