We read with interest the paper recently published by Jokubkiene et al.1 about the role of three-dimensional power Doppler ultrasound (3D-PDU) in discriminating benign from malignant ovarian tumors. First of all, we would like to congratulate the authors for performing this study. They used a similar approach to the one proposed by us2 and their findings—essentially that indices derived from 3D studies of vascularized ovarian tumors are higher in malignant than in benign lesions—are in close agreement with ours. However, we would like to comment on issues relating to their methodology and target population. Whole tumor assessment by 3D-PDU may be inadequate because, in large tumors for example, it is virtually impossible to include the whole tumor in just one volume box. In such cases, information can be lost and subsequent analysis can be subject to bias. Furthermore, in most tumors vascularization is located in a few areas and not evenly distributed throughout the tumor. Thus one might include within the volume box misleading information. Gray signals from stationary blood clots and mucous plugs, for example, will be processed by the Virtual Organ Computer-aided AnaLysis (VOCAL) software as normal, non-malignant tissue that would reduce the relative proportion of vascular data within the sampled volume. Furthermore, if the tumor contains cystic areas, these are irrelevant for analysis. If they contain color voxels resulting from artifacts or noise, 3D-PDU calculations may be erroneously affected. In our opinion, cystic areas should be excluded to avoid this potential source of error. The selection of a 5-cm3 spherical sample volume from the ‘part of the tumor that appeared to be the most vascularized on the basis of subjective evaluation’ was found by the authors to be reproducible, but it would be interesting to know more about the examiners' experience. It may well be that this approach is reproducible in experienced hands but not in those of less experienced examiners. We agree that 3D-PDU is likely to add little to the correct diagnosis of malignancy in a general population of ovarian tumors. However, we believe that the target population for this technique should be a selected one. We also endorse the authors' assertion that diagnostic performance based on the subjective impression of an experienced examiner is difficult to improve upon. Most ovarian tumors can be correctly classified by B-mode ultrasonography. However, there are some ovarian lesions that are very difficult to classify3. Two-dimensional PDU may be useful for predicting malignancy in these tumors, by reducing the false-positive rate, as demonstrated in a multicenter study that included more than 800 ‘complex’ adnexal masses4. However, there are still some benign tumors that can resemble malignant tumors, such as cystadenofibromas, low malignant potential tumors, fibromas, Brenner tumors and Granulosa cell tumors, and which are very difficult to discriminate, especially if there is no evidence of metastatic disease. Many of these tumors will exhibit vascularization and, in these cases, 3D-PDU may play a role by reducing the false-positive rate while not affecting substantially the sensitivity5. The authors report in their paper 31 benign ovarian lesions (39.2% of all benign lesions) that showed unilocular-solid, multilocular-solid and solid appearances (no malignant lesion, however, had a simple unilocular or multilocular appearance). For many examiners these unilocular-solid, multilocular-solid or solid benign tumors would be considered, at the very least, as questionable, and a challenge to accurately distinguish from their malignant counterparts in the absence of metastatic disease. It is not stated how many of these benign tumors had flow detectable within the solid components, but it can be inferred by just looking at the ‘color score’ (88% of benign tumors had a color score ≥ 2) that many of them did. How could these vascularized tumors with solid components, especially if flow was present in the solid component, be classified as benign with a high degree of certainty? In our opinion, it is precisely this population of ovarian tumors that should be assessed by 3D-PDU. Finally, we would have appreciated a more in-depth discussion from the authors about their findings. The papers referred to in the Discussion section are exclusively from the authors' group and other studies that have addressed the same question2, 6 have not been commented upon or compared with their data. J. L. Alcázar*, S. Guerriero , * Department of Obstetrics and Gynecology, Clínica Universitaria de Navarra, University of Navarra Pamplona, Spain, Department of Obstetrics and Gynecology, Ospedale San Giovanni di Dio University of Cagliari, Cagliari, Italy