PSRC Main Site  |  Past & Future Meetings
Plastic Surgery Research Council

Back to 2020 ePosters


3D Image-based Computational Modeling For Breast Surgeries
Audrey L. Cheong1, Urmila Sampathkumar1, Gregory P. Reece2, Mary Catherine Bordes2, Summer E. Hanson2, Fatima A. Merchant1.
1University of Houston, Houston, TX, USA, 2The University of Texas MD Anderson Cancer Center, Houston, TX, USA.

PURPOSE
It is important to ensure that women undergoing cosmetic or reconstructive breast surgeries make well-informed decisions regarding their surgical options. However, communicating appearance-related surgical outcomes to patients and eliciting patients' expectations regarding their desired breast aesthetics is challenging. Currently, plastic surgeons rely on drawings or images of former patients to explain surgical procedures and their possible outcomes. Modeling techniques that can generate realistic simulations of postoperative breast appearance would improve patient-physician communication, but are lacking. We have developed a spherical harmonics (SPHARM) based approach to model breast shape [1] and an example-based simulation technique to estimate postreconstruction breast shape.
METHOD
We created a SPHARM-based computational model of the breasts shape using breast regions extracted from pre- and postoperative 3D surface images of patients' torsos obtained from an IRB approved study, and used an example-based methodology to estimate the postoperative shape of a patient's breast. Examples were drawn from a knowledge database containing shape models of pre- and postoperative breasts from 29 patients. We calculated the root mean square distance (RMSD), a similarity measure that compares shapes of the test and exemplar breasts (smaller RMSD values indicate similar shapes) and selected 5 preoperative breasts similarly shaped as the test breast and their corresponding postoperative breasts from the database. Next, we generated the average pre- and postoperative breast models using the 5 exemplar breast model pairs. The differences in the model parameters between the average pre- and postoperative breast pair were computed and applied to the test preoperative breast model to estimate its postoperative shape.
RESULTS
Our test set consisted of pre- and postoperative 3D images of 53 breasts from 29 patients undergoing reconstructive surgery: Twenty-five breasts had been reconstructed with DIEP/TRAM flaps, 21 with implants, 1 with an LD flap, and 6 revisions of the natural contralateral breasts. Table 1 shows RMSD values between estimated and actual postoperative breast shapes. The RMSDs were observed to be lower in women with a BMI < 25 than in women with BMI > 30.


Figure 1. (a) Test and 5 exemplar pre- and postoperative breast pairs. (b) Representative results (c) Visualization of overlaid preoperative, postoperative, and estimated postoperative breasts.

BMIRMSD
NMinimumMedianMeanMaximumStandard deviation
18-41All Samples536.1710.6711.5622.823.65
< 25Normal136.179.229.6414.932.60
25-30Overweight216.198.8610.2517.112.87
> 30Obese199.2114.4614.3122.823.49

Table 1. RMSD values between actual postoperative and estimated postoperative breast models.
CONCLUSION
Our example-based computational approach can estimate postoperative breast shape in patients undergoing reconstructive surgery after mastectomy. This method can be extended to simulate breast shape deformations such as those that occur with changes in pose (e.g., when the patient is
upright or supine). Future focus group studies will assess the feasibility and utility of this method in the clinical setting.
REFERENCES
[1]. Cheong, Audrey, Computational modeling of breast shape using spherical harmonics, PhD Dissertation, College of Engineering, University of Houston, 2018.


Back to 2020 ePosters