Structured-light surface scanning system to evaluate breast morphology in standing and supine positions

Breast shapes are affected by gravitational loads and deformities. Measurements obtained in the standing position may not correlate well with measurements in the supine position, which is more representative of patient position during breast surgery. A dual color 3D surface imaging system capable of scanning patients in both supine and standing positions was developed to evaluate the effect of changes in body posture on breast morphology. The system was evaluated with breast phantoms to assess accuracy, then tested on ten subjects in three body postures to assess its effectiveness as a clinical tool. The accuracy of the system was within 0.4 mm on average across the model. For the human study, there was no effect of body posture on breast volumes (p value > 0.05), but we observed an effect of completeness of breast scans on body posture (p value  < 0.05). Post-hoc tests showed that the supine position and the standing position with hands at the waist differed significantly (p value  < 0.05). This study shows that the system can quantitatively evaluate the effect of subject postures, and thereby has the potential to be used to investigate peri-operative changes in breast morphology.


S1.1 System setup
A schematic of the structured-light scanning (SLS) system is presented in Figure S1(a). The two scanners were mounted on an articulating arm with the ability to switch their orientation to scan both at the standing and supine positions [Figures S1(b) and S1(c)] from different angles. The articulating arm was attached to a wheeled medical cart (dimensions 52 cm x 92 cm x 64 cm) for mobility. Each SLS system consisted of two cameras (STC-MBS231U3V, Sentech America, USA) and a projector (K132, Acer Inc, Taiwan) as shown in Figure S1(a).
Each camera was fitted with a fixed focal lens (M1214-MP2, 12 mm focal length, Computar, USA) and connected to a laptop (ThinkPad W540, Lenovo Group Ltd., China) via a USB 3.0 port.
Each projector was connected to the laptop and projected over an area of 550 mm x 340 mm at a working distance of 60 cm. The original HP cameras that came with the commercial package had a field of view (FOV) of 240 mm x 180 mm at a working distance of 60 cm. These cameras were replaced with the Sentech cameras which provided a FOV of 534 mm x 400 mm at the same distance. System control and image acquisition was performed using the software provided by HP.

S1.2 Dual color 3D-SI
While two sets of SLS systems improved system coverage, intersystem cross talk prevented these systems from operating simultaneously on the same field of view. To overcome this limitation, two optical filter sets were installed on the cameras and projectors [ Figure S1(a)].
Blue and green filters were selected to minimize interference between scanners. The green filter set was comprised of a high performance OD4 long-pass filter at 500 nm (TECHSPEC, Edmund Optics, Ø25 mm and Ø50 mm). The blue filter set was comprised of a high performance OD4 short-pass filter at 450 nm (TECHSPEC, Edmund Optics, Ø25 mm and Ø50 mm).

S1.4 Human participants
Ten women were recruited for preliminary assessment of the 3D-SI system. Two parameters were reported: (i) bra size and (ii) breast ptosis. First, bra size is made up of cup size (in letter) and chest wall diameter (in number). Cup size is calculated by the difference between the fullest part of the breast and the chest circumference. Cup sizes of A and AA tend to have smaller breast volumes (50-250 cm 3 ), and cup sizes of D or more tend to have larger volumes (500 cm 3 or more). In addition to cup sizes, breast volume is also affected by chest diameter which is measured at the inframammary fold. In this paper, bra sizes were calculated by the method of Zheng et al. (2006). 27,28 Second, breast ptosis is a measurement to characterize breast morphology. It is defined when the nipple drops to the level of the inframammary crease (IMC), and the classification of the degree of ptosis is adapted from the work of Kirwan (2002) 29 .
Grade 0 (normal) is classified when the nipple position at least 1 cm above the IMC, Grade 1 ptosis is classified when the nipple position is even with IMC, and Grade 2 ptosis is classified when the nipple position is 1 cm below the IMC. The demographics of the female participants are listed in Table S1.

S1.5 Post-processing of surface scans: stitching
The 3D models from each SLS system were stitched together into a complete surface model by the HP software (HP 3D Scan Pro 5.4.0). The scans were coarsely registered using the fiducial markers, then finely registered by applying an iterative closest point (ICP) algorithm to optimize alignment. 30 The coarse and fine registrations were performed automatically using the fiducial markers between the scans from each SLS system. The registered scans were then exported and combined by merging common vertices of the scans in 3D mesh processing software (Meshlab 2016, GNU General Public License software) 31 to obtain a full mesh.

S1.6 Post-processing of surface scans: estimation of breast volume
The estimated volume of the breast depended on the extracted breast region; this region of interest was selected by examining the surrounding anatomical structures. We manually segmented the breast volumes using a breast volume analysis method modified from techniques described by Yip et al. (2012). 7 Briefly, the full mesh from each surface scan consisting of the breast and the surrounding anatomical structures as shown in Figure S3