We were interested to read the paper by Guiot et al. entitled ‘Is three-dimensional power Doppler ultrasound useful in the assessment of placental perfusion in normal and growth-restricted pregnancies?’ in the February edition of the White Journal1. The authors used ‘3D power Doppler ultrasound indices [for] the assessment of placental perfusion’ and studied the relationship of these indices to ‘gestational age (GA), placental position, and umbilical artery Doppler FVW [flow velocity waveform] patterns in normal and IUGR [intrauterine growth-restricted] pregnancies’. We have some methodological queries and would also like to comment on several statements made about the three-dimensional (3D) vascular indices and their purported relationship to true blood flow characteristics.

The authors make many assumptions and several factually incorrect statements about quantitative three-dimensional power Doppler angiography. They state that ‘even very small blood movements within the investigated volume can be detected by a combination of power and color Doppler sonography, and their impact in the given volume, representing the overall perfusion, is evaluated by indices computed by built-in algorithms’. Movement of blood, or any fluid for that matter, does not result in a Doppler signal. It is the erythocytes within the blood or scatterers within a fluid medium that are responsible for the Doppler shift that is represented by a color map or waveform. Blood flow and perfusion are distinct entities with differing definitions. Flow is the amount of blood passing per unit time (usually per minute) while perfusion is defined as the amount of flow in a volume of tissue per unit time and is usually measured as milliliters per minute per gram of tissue, although, since the density of most tissues is near to 1 g/mL, units of milliliters per minute per milliliter are often substituted2. The three-dimensional vascular indices reported in the study are not a function of time and their quantification cannot, therefore, equate to perfusion or flow. Whilst analysis of Doppler waveforms takes into consideration the cardiac cycle it does not account for volume flow as it is a two-dimensional assessment and is not a measure of perfusion either.

The authors' descriptions of the 3D vascular indices are also incorrect and therefore misleading, as they refer to the vascularization index (VI) as being representative of ‘overall perfusion’ and say that the flow index (FI) ‘evaluates the overall blood flow in the sample volume’, but both perfusion and blood flow are time-dependent parameters. The FI represents the average color value of all color voxels, so does not even remotely consider the overall blood flow. The vascularization flow index (VFI) is described as being a measure of ‘blood velocity in the same sample volume’, but power Doppler provides no velocity information and so this statement is illogical. Current opinions are that these indices are representative of the percentage of power Doppler data within the defined volume of interest (VI), the mean signal intensity of the power Doppler information (FI), and a combination of both factors, derived through their multiplication (VFI), and it has been suggested that they are representative of vascularity and flow intensity3, 4.

One of the aims of the study was to investigate the effect of attenuation on the 3D vascular indices, as the authors quite rightly state that ‘different placental positions require different insonation depths (and) measurements could also be dependent on this parameter’. This is not a matter for debate; power Doppler is depth-dependent and subject to attenuation. This does not require assessment as it has been shown in both in-vitro5 and in-vivo6 studies, but further research is required to develop methodologies to account for this confounding factor, which currently limits all studies other than certain gynecological ones in which the organ of interest is in close proximity to the transducer. The only way to assess vascularity and flow within different subjects is to standardize measurements by calculating an index known as the ‘fractional moving blood volume’ (FMBV), which has been shown to correlate well with true tissue perfusion7, 8. Of the 3D vascular indices, the VFI is most similar to the FMBV because it shows the moving blood volume. However, it is not truly fractional as it is not standardized.

The concept of comparing different sites within the same placenta is good and, to some degree, addresses the problem of attenuation, as the patient self-standardizes. However, the data presented are the overall median values for the placentae from 30 growth-restricted fetuses and 15 controls. These absolute values are meaningless without information on placental position, as the authors maintained their Doppler settings in all cases. The more posterior placentae will have had lower vascular indices as a function of increasing depth and if these were not evenly distributed amongst the two groups they should not have been compared as separate entities9. Indeed, after accounting for the distribution of the data, the mean sampling depth should have been stated and this should have been equal between the groups before any further comparisons were made.

Insufficient details were provided about the Doppler settings, which should have included the power, gain, rise and persistence and reject values. One reason why the ‘intervillous flow velocity [was] below the Doppler threshold’ could have been that the PRF was set too high. A PRF of 0.6 kHz may have been more appropriate as it provides more information about low flow10 and, in the absence of data with a lower PRF setting, the authors cannot conclude that their ‘results relate only to fetoplacental blood flow’.

There is no description of the technique used for data acquisition, arguably the hardest and yet most important component of 3D power Doppler angiography. We know a single observer undertook the vascular measures but was a single dataset acquired and how many times was this measured? This information should have been provided and the reliability of their measures presented. Poor intraobserver measurement reliability affects the discriminative power of a study and could explain the differences and the lack of differences between the groups. In the absence of these data, no meaningful conclusions can be made and the authors' findings of a ‘large interplacental variability in the VI and VFI for controls and IUGR placentae’ may simply reflect their measurement technique. Their conclusion that ‘the FI is the only reproducible 3D vascular index for assessment of the placenta’ cannot be made in the absence of measurement reliability data.

Their Table 2 shows the range of the non-normally distributed data and includes the extreme values of 0 and 99.9 for the VI within each placenta. These figures equate to the complete absence of vascular information or ‘blood flow’ and the presence of absolute vascularity with no gray-scale information, i.e. the presence of blood vessels only. These extremes are highly unexpected as the placenta is a network of vessels and villi. These findings could, however, be explained by an inappropriately small sampling volume, so it is important that the authors document the size of the spheres used for each ‘vascular biopsy’. The spherical models available through the VOCAL software program are not standardized so we would be interested to know how the sphere volume was maintained throughout the study to ensure that a repeatable sampling technique was used. The large coefficient of variation that they found within each placenta (85.4, 93.0 and 24.7 for VI, FI and VFI, respectively) may reflect variations in the sphere volume or user application of the sampling method. Glass body imaging was used to avoid sampling spiral arteries, but a limited description of their technique is provided. The fact that the sampling sites were chosen subjectively by such a technique will have introduced bias and random samples should have been performed, although we understand and accept the need to avoid sampling too near the chorionic and basal plates. A 25° sector angle was used throughout the study. This is unlikely to have been sufficiently large to sample the entire placenta three dimensionally at these gestational ages. Their two-dimensional image (Figure 1) with the five spherical sampling sites is greater than 25°, but the sampling was a 3D measurement. If the middle sample was midway between the central and peripheral sites, they randomly chose two lines of interest with the central site as the constant. This could not have examined the placenta thoroughly as it is a 3D structure and the study cannot truly have investigated the ‘intraplacental variability of the VI, FI and VFI’, which was one of their aims.

In our opinion, the unclear methodology and incorrect biophysical assumptions about power Doppler invalidate the study, and the authors' conclusions should be interpreted with caution. We feel the reader with limited knowledge of the biophysics of power Doppler and three-dimensional ultrasound may be misled by the information presented in the article. As a collaborative group we were disappointed, considering we have raised these issues previously11, 12, and we suggest that any further work in this field be conducted only after careful consideration of the limitations in power Doppler quantification. Future studies must address the issue of standardization if they are to have any scientific merit.


  1. Top of page

N. J. Raine-Fenning*, A. W. Welsh‡, N. W. Jones†, G. Bugg†, * University of Nottingham, Queen's Medical Centre, Nottingham, UK, † Nottingham University Hospitals NHS Trust, Queen's Medical Centre, Nottingham, UK, ‡ Royal Hospital for Women, University of New South Wales, Sydney, NSW, Australia