Does the size of three-dimensional power Doppler spherical sampling affect the interobserver reproducibility of measurements of vascular indices in adnexal masses?

Past, present and future ultrasonographic techniques for analyzing ovarian masses

Ultrasonography is today the method of choice for distinguishing between benign and malignant adnexal pathologies. Using pattern recognition several types of tumors can be recognized according to their characteristic appearance on gray-scale imaging. Color Doppler imaging should be used only to perform a semiquantitative color score or evaluate the flow location. International Ovarian Tumor Analysis group had standardized definitions characterizing adnexal masses and suggested the use of 'simple rules' in premenopausal women. Recently, the use of 3D vascular indices has been proposed but its potential use in clinical practice is debated. Also computerized aided diagnosis algorithms showed encouraging results to be confirmed in the future.

Limitations of three-dimensional power Doppler angiography in preoperative evaluation of ovarian tumors

Background

This study describes the accuracy of three-dimensional power Doppler (3D-PD) angiography as secondary method for differential diagnosis of ovarian tumors.

Method

Seventy-five women scheduled for surgical removal of adnexal masses were assessed by transvaginal ultrasound. Ovarian tumors were classified by IOTA simple rules and two three-dimensional blocks were recorded. In a second step analyses, a 4 cm³ spherical sample was obtained from the highest vascularized solid area of each stored block. Vascularization index (VI), flow index (FI) and vascularization-flow index (VFI) were calculated. The repeatability was assessed by concordance correlation coefficient (CCC) and limits of agreement (LoA), and diagnostic accuracy by area under ROC curve.

Results

IOTA simple rules classified 26 cases as benign, nine as inconclusive and 40 as malignant. There were eight false positive and no false negative. Among the masses classified as inconclusive or malignant by IOTA simple rules, the CCCs were 0.91 for VI, 0.70 for FI, and 0.86 for VFI. The areas under ROC curve were 0.82 for VI, 0.67 for FI and 0.81 for VFI.

Conclusions

3D-PD angiography presented considerable intraobserver variability and low accuracy for identifying false positive results of IOTA simple rules.

Spatiotemporal image correlation with spherical sampling and high-definition flow: new 4-dimensional method for assessment of tissue vascularization changes during the cardiac cycle: reproducibility analysis

OBJECTIVES

To describe and assess the interobserver reproducibility of a new method for evaluation of ovarian vascularization using spatiotemporal image correlation-high definition flow (STIC-HDF).

METHODS

Stored 4-dimensional (4D) STIC-HDF volume data from 39 healthy pre-menopausal fertile women (aged <35 years) examined in the follicular part of the menstrual cycle by transvaginal sonography were assessed by two different examiners blinded from each other (one in Spain the other in Poland). Using 1-cm(3) spherical sampling, the vascularization index (VI) from the most vascularized part of the ovarian stroma was calculated at two different moments of the cardiac cycle (systole and diastole). System settings were kept constant for all patients (pulse repetition frequency, 0.6 kHz; gain, 0.2) with a depth of 40 mm. Analysis was performed offline using 4D software on a personal computer. On the basis of VI and vascularization-flow index (VFI) values during systole and diastole, 4 new 4D indices were defined: 4D systolic/diastolic volumetric index (4D-SDVI = VI(syst)/VI(diast)), 4D hemodynamic volumetric index (4D-HVI = [VI(syst) + VI(diast)]/[VI(syst) - VI(diast)]), 4D systolic/diastolic vascularization-flow index (4D-SDVFI = VFI(syst)/VFI(diast)), and 4D hemodynamic vascularization-flow index (4D-HVFI = [VFI(syst) + VFI(diast)]/[VFI(syst)- VFI(diast)]). Reproducibility of measurements was estimated by calculating the intraclass correlation coefficient (ICC).

RESULTS

The systolic VI, diastolic VI, 4D-SDVI, 4D-HVI, systolic VFI, diastolic VFI, and 4D-HVFI showed good reproducibility (ICC, 0.992, 0.994, 0.879, 0.915, 0.995, 0.995, and 0.893, respectively). The 4D-SDVFI showed moderate reproducibility (ICC, 0.797).

CONCLUSIONS

We describe 4 new 4D vascular indices for assessing tissue vascularization using STIC-HDF technology. Assessment of ovarian vascularization using this STIC-HDF spherical sampling is reliable. The calculation of these new indices is reproducible between two different examiners.

Does sphere volume affect the performance of three-dimensional power Doppler virtual vascular sampling for predicting malignancy in vascularized solid or cystic-solid adnexal masses?

OBJECTIVE

To assess whether, when using spherical sampling with Virtual Organ Computer-Aided Analysis (VOCAL) for calculating three-dimensional (3D) power Doppler angiography (PDA) indices, the sphere volume affects performance in the prediction of malignancy in vascularized cystic-solid or solid adnexal masses.

METHODS

One hundred and thirty-eight women (mean +/- SD age, 51.8 +/- 14.1 years) diagnosed as having vascularized cystic-solid or solid adnexal masses on B-mode and two-dimensional (2D) power Doppler ultrasound were evaluated by 3D-PDA prior to surgery. Five women had bilateral masses, giving a total number of 143 masses analyzed. Vascularization was assessed using VOCAL software. 3D-PDA vascular indices (vascularization index (VI), flow index (FI) and vascularization flow index (VFI)) from the most vascularized area within papillary projections and solid areas were calculated automatically using spherical sampling. Five different volumes of sphere were used (1 cm(3), 2 cm(3), 3 cm(3), 4 cm(3) and 5 cm(3)) in each case. A definitive histological diagnosis was obtained in each case after surgical tumor removal.

RESULTS

One hundred and seventeen (82%) masses were malignant and 26 (18%) were benign. Morphological evaluation revealed 34 (24%) unilocular solid masses, 49 (34%) multilocular solid masses and 60 (42%) mostly solid masses. The 1-cm(3) sphere could be used in 100% of the cases, the 2-cm(3) sphere could be used in 98.2% of the cases and the 3-5-cm(3) spheres could be used in 97.2% of the cases. The median VI, FI and VFI for all sphere volumes were significantly higher in malignant compared with non-malignant tumors. Receiver-operating characteristics curve analysis showed that VI and VFI, independently of sphere volume, were better predictors of malignancy than was FI. The best cut-off values for the 3D-PDA indices differed depending on sphere volume. VI was significantly more specific than were VFI and FI.

CONCLUSIONS

Sphere volume does not affect the performance of 3D-PDA. We recommend the use of different cut-off values for 3D-PDA indices for discriminating between benign and malignant adnexal masses, depending on the sphere volume used. Use of VI is preferable due to its higher specificity.