Three-dimensional breast assessment by multiple stereophotogrammetry after breast reconstruction with latissimus dorsi flap

Introduction: Numerous methods exist for the assessment of the female breast. Traditionally, a subjective approach was taken for surgical planning and evaluation of the postoperative outcome. Several objective methods have been developed to support this procedure, among which are laser scanning, MRI, mammography, ultrasound and photography. Recently, 3D imaging technology has been developed. Material & Method: 3D breast assessment by multiple stereophotogrammetry was examined. A custom-made imaging system with eight digital cameras arranged in four camera pods was utilised. This system was used for breast capture, resulting in eight images obtained by the cameras. The merging of these images and 3D image construction was carried out by C3D software and the volume assessment of the 3D images was made using breast analysis tool (BAT) software, developed by Glasgow University. A validation study was conducted. Nine plaster models were investigated and their volume determined by 3D stereophotogrammetry and water displacement method. Water displacement was considered to be the gold standard for comparison. The plaster models were specially made in order to represent a variety of shapes and sizes of the female breast. Each plaster model was examined 10 times by each method. Further, the volumes of the breasts of six female volunteer live models were investigated by the same two methods and the results compared. A special focus was placed on the reproducibility of the assessment. Each live model was captured with the 3D capture system three times at two different time points after retaking a special pose in a custom-made positioning frame. Altogether, each live model was captured six times, resulting in six 3D images, each of which was measured three times with BAT software. A patient study was conducted in 44 patients after unilateral immediate breast reconstruction with Latissimus dorsi flap and no contra-lateral surgery. Each patient underwent 3D imaging with the multiple stereophotogrammetry system. During capture, the special pose in the custom-made positioning frame was taken by the patient’s leaning forward almost horizontally with the upper body for the breasts to rise off the chest wall to enable full breast coverage by the cameras. 3D images were constructed with C3D software and volumes measured with BAT. For each patient, one 3D image was constructed and measured four times with BAT software. In addition to the volume determination, a shape analysis was conducted. For this purpose, 10 landmarks were determined according to recommendations in the literature. Two landmarks, sternal notch and xiphoid, were marked, forming an imaginary midline between each other and four landmarks on each breast, i.e. the medial and lateral ends of the infra-mammary fold, and the most prominent and most inferior breast points were utilised for symmetry assessment between the right and left breasts. Each landmark was recorded four times by the operator on the 3D image and three-dimensional coordinates obtained. By assessment of the left and right breasts a breast asymmetry score was calculated. Firstly, breast asymmetry was assessed objectively on the 3D images through the centroid size, which was determined as the square root of the sum of squared Euclidian distances from each landmark to the centroid. The centroid was the geometric mean of the landmarks. Secondly, asymmetry was assessed through breast volume by application of BAT software. Thirdly, asymmetry was examined through the landmarks themselves by investigation of the mismatch of the landmark configuration of one breast and its relabelled and matched reflection. The non-operated and reconstructed sides were compared and landmarks were recorded by the operator in three dimensions in four repeated tests. A decomposition of the total landmark asymmetry into its factors was conducted by fixation of the surface of the non-operated side and translation, rotation and scaling of the surface of the reconstructed side. For comparison, a subjective breast assessment was conducted by six expert observers who rated the results after breast reconstruction by subjective qualitative assessment of the symmetry in 2D images of the same 44 patients in six poses. For this purpose the Harris scale was utilised, providing a score of 1 to 4 for poor to excellent symmetry. Results: The results revealed that differences in the obtained volumes in the plaster models were not significant. In contrast, differences in the breast volumes measured in the live models were significant. The examination of the reproducibility revealed that overall reproducibility obtained by stereophotogrammetry was better than that obtained by water displacement. No correlation between breast size and reproducibility of the measurements was found. The results of the patient study demonstrated that the reproducibility of the landmarks was within 5 mm. There was a non-significant difference of the centroid sizes between both breasts. There was a significant difference of the volumes between the two breasts, with the non-operated side being larger than the reconstructed side. Volume was considered to be a more accurate measure for comparison of both breasts than centroid size as it was based on thousands of data points for the calculation as opposed to only four points of the centroid size. The statistical analysis of the landmark data provided a mathematical formula for determination of the breast asymmetry score. The average asymmetry score, derived by landmark assessment as the degree of mismatch between both sides, was 0.052 with scores ranging from 0.019 (lowest score) to 0.136 (highest score). The decomposition of the landmark-based asymmetry revealed that location was the most important factor contributing to breast asymmetry, ahead of intrinsic breast asymmetry, orientation and scale. When investigating the subjective assessment, the inter-observer agreement was good or substantial. There was moderate agreement on the controls and fair to substantial intra-observer agreement. When comparing the objective and subjective assessments, it was found that the relationship between the two scores was highly significant. Conclusion: We concluded that 3D breast assessment by multiple stereophotogrammetry was reliable for a comparative analysis and provided objective data to breast volume, shape and symmetry. A breast asymmetry score was developed, enabling an objective measurement of breast asymmetry after breast reconstruction. 3D breast assessment served as an objective method for comparison to subjective breast assessment