Clinical Assessment of Breast Volume

Can 3D Imaging Be the Gold Standard?

Renee C. Killaars, MD; Myriam L. G. Preuβ, MD; Nathalie J. P. de Vos, MD; Camille C. J. L. Y. van Berlo, MD; Marc B. I. Lobbes, MD, PhD; René R. W. J. van der Hulst, MD, PhD; Andrzej A. Piatkowski, MD, PhD

Disclosures

Plast Reconstr Surg Glob Open. 2020;8(11):e3236 

In This Article

Discussion

This study investigated the Vectra XT 3D imaging system for determination of breast volume in comparison with the gold standard in literature (MRI) and the clinical standard (assessment by the plastic surgeon). Our study suggests that the use of 3D measured breast volumes is reliable. Both 3D breast volume and MRI breast volume showed excellent reliability. In Bland–Altman analysis, 3D measurements had even smaller limits of agreement compared with MRI, emphasizing a good reliability for 3D for breast volume measurement. However, the exactness of breast volume measured by the two methods differed significantly.

Three-dimensional breast measurements were lower than MRI breast measurements. Volume outcomes and differences depended on breast size. This was further analyzed using Bland–Altman analysis (Figure 3A). A fixed and proportional difference from the Bland–Altman plot could be seen, meaning that breast volume differences measured by MRI and 3D imaging are dependent on size, showing smaller differences for small breast volumes and larger differences for high breast volumes.

This was earlier demonstrated in a study of Yang et al, where proportional errors were found with respect to breast volume.[16] An increase of measurement error is seen with an increasing breast size, which is regarded as proportional difference. Secondly, the increasing slope of the plot suggests that the two can be related using a linear regression model. This was also demonstrated in Figure 4, in which a linear regression could be found between MRI and 3D. A similar linear association between 3D and MRI was elaborated earlier in a study by Yang et al. As results change in parallel, 3D measured volume is capable of predicting the MRI breast volume.[16] In 2011, Koch et al. also found significantly smaller volumes in 3D compared with MRI, but found 3D volume applicable for predicting MRI-measured volume using a linear regression.[14]

An possible explanation for the difference in measurement methods was suggested in the definition of dorsal boundary of the breast. The internal boundary can only be guided by the shape of the skin surface around the breast. This defines the closed object of a breast and can influence the dorsal limit used for volume calculation. The amount of subcutaneous fatty tissue or even bad posture can have a relevant influence on the interpolation of the breast.

This also means that this tool is not suitable for patients who have pectus excavatum deformities or scoliosis. A prone breast MRI, where the chest wall can exactly be defined (as the anterior border of the pectoralis muscle), is well suited for diagnosis, but it is less suitable for use in clinical applications of defining breast parameters.

The disadvantage of the clinical application of MRI is its measuring position. During the MRI examination, a patient is in a prone position so that the axillary tissue is shifted to the front and potentially added to the breast. In 3D imaging, the patient is in a standing position. This might minimalize the amount of lateral subcutaneous tissue that is included in the breast volume measurement,[14] and might also be influenced by the definition of chest wall, which is further to the front compared with posterior delineation of the manually segmented breasts on MRI.[17]

Some have tried to introduce a special MRI apparatus for unilateral breast MRI in a supine position, showing promising results with breast images comparable to prone position.[9,18,19] These have been suggesting a better clinical breast representation. However, supine positioning in MRI is not part of standard care and needs further development and use in clinical practice. Moreover, Khatam et al. demonstrated that a change in the subject's position from supine to upright can result in significant stretches in some parts of the breast skin, especially above the nipple, resulting in different volumes for a different patient positioning.[20]

All this implies that measuring position might have consequences for the breast volume that is measured. When a patient looks at herself in the mirror, she does not see the mammary gland but the external shape of the breast. As we are interested in quantifying the breast size in the way the patient perceives it in the mirror, the most adequate measuring position for measuring breast volume seems to be in standing position, instead of supine or prone.

Since 3D measurements are captured in standing position, it makes the breast measurement outcome the most clinically applicable. The Vectra XT 3D software uses automated calculation of volume, which is programmed by the software. This automated process is time-reducing and efficient. Unlike some estimation errors, especially in ptotic breasts, where the submammarian fold is difficult to determine and hence the caudal limit of the breast,[14,21] it is an objective way of determining breast parameters. In this study, no detailed information was reported on ptosis grading of the breasts. However, a large variety of breast volumes were included in the study, representing a varied patient population. No errors were present in volume analysis with 3D imaging, and the extend of incorrect measurement in ptotic breast did not appear to be a relevant problem in this study.

Additionally, reliability of MRI and 3D imaging was compared. The results on reproducibility of both techniques showed excellent results. ICC of 3D breast measurements was 0.991, whereas the ICC of MRI measurements was 0.990.

Bland–Altman analysis showed only a good agreement between 2 measurements, which was even better than agreement between 2 MRI measurements. No proportional difference nor fixed difference was seen from this analysis. This means that reliability of 3D breast volume was high for small and for large breast volumes, and not depending on breast size. Earlier measurements with 3D imaging for other clinical applications than breast volumes have also shown high reproducibility.[21–25]

When looking at clinical aspects, 3D has several advantages over MRI for breast assessment. Three-dimensional measurements have the advantage over MRI to be non-invasive, and capturing images takes minimal time. No extra appointment needs to be planned for a breast image. Patients with claustrophobia, metal implants or cardiac pacemakers can benefit from a 3D image compared with MRI. This makes the camera suitable for the use in daily practice in clinic hours with patients. MRI avoids x-ray dosage, but the process is expensive and performed only for a small number of women in routine practice.[26] The costs for an MRI examination are estimated to be around 280–1400 USD per scan. Moreover, segmentation on imaging software is still mostly conducted manually, which is labor-intensive, and evidence supporting automatic segmentation tools are scarce.[27] Koch et al. have found significantly shorter time for 3D recording and volume assessment compared with MRI recording and volume assessment. For the Vectra XT 3D imaging system, the major limitation is its relatively high acquisition costs of 40,000 USD and its lack of portability. Many large academic centers as well as aesthetic plastic surgeons have already a 3D imaging system available. Until now, it does not seem to be cost-effective for smaller peripheral centers.[27]

Still, many plastic surgeons use only some standardized measurements to estimate volume. The difference between 3D imaging and the plastic surgeon's volume assessment seemed comparable: 17.7 cm3 (95% CI: –53.3 to 17.9; P = 0.323). However, in Bland–Altman analysis, limits of agreement ranged from –270.7 to 235.3 (Figure 5). Considering that the minimum volume difference detectable by the human eye is 50 cm3 by subjective judgment,[28] we suggest a low agreement between the plastic surgeon's estimation and the 3D measurement. From analyzing the plot, it seems that the scatter around the bias line gets larger as the average gets higher. This might suggest that for larger breasts, a larger variability of estimated breast volume is found.

This study did not aim to investigate the reproducibility of plastic surgeon's estimation. The answers to this question are limited by our current study design. For future research, the next step would be to compare two surgeon's estimations of breast.

We consider our current data useful for clinical practice since this study was, to our knowledge, the first which has compared 3D breast volume to a clinical assessment of breast volume with the patient in standing position. Secondly, we showed that volume measurement performed by MRI or 3D imaging is an objective parameter which provide reliable results for breast volume. We think that the breast volume can be different for both methods because of different measuring position (prone position versus standing position).

Future research should focus on reproducibility of plastic surgeon's estimation of breast parameters to see if 3D breast volumes are superior in the clinical assessment of breasts. This could increase the clinical utility of 3D imaging for breast assessment and could represent an important step toward a more standardized approach to breast surgery.

processing....