Objective:
Accurate assessment of breast volume is very important in preoperative planning of immediate breast reconstruction (IBR) in order to achieve breast symmetry and a satisfactory cosmetic outcome. In this study, we introduced an innovative breast volume measurement using the automatic segmentation of magnetic resonance imaging (MRI). In addition, we compared the accuracy and reliability of this innovative method with four other conventional modalities including manual segmentation of MRI, three-dimensional (3D) surface imaging, mammography, and anthropomorphic measurement (BREAST-V formula).
Material and Methods:
From 2016 to 2020, patients scheduled to mastectomy with IBR were enrolled in this study. Five different volume assessments including automatic segmentation of MRI, manual segmentation of MRI, 3D surface imaging, mammography and BREAST-V formula were performed to evaluate breast volume. The results were validated by the water displacement volume of the mastectomy specimen.
Results:
41 women who were scheduled to have 44 mastectomies were included. There was a strong linear association between the breast volume measured by MRI (either automatic or manual segmentation) and water displacement volume of mastectomy specimens (automatic segmentation: r = 0.942, p < 0.001; manual segmentation: r = 0.944, p < 0.001). Among the 44 breasts, the volume of nine breasts were evaluated by all five measurements and the intraclass correlation coefficient (ICC) was evaluated and MRI had significant association with volume of the mastectomy specimen (automatic segmentation: k = 0.913, p < 0.001; manual segmentation: k = 0.938, p < 0.001). In Bland-Altman plots, the relative error from automatic segmentation of MRI demonstrated a high level of agreement and no systemic or proportional bias was found.
Conclusions:
Breast volume assessment using either automatic segmentation or manual segmentation of MRI had the better accuracy and reliability among all five methods in our study. Most importantly, volume measurement using automatic segmentation of MRI was more easily-accessible and less time-consuming.