A Machine Learning-Based Model for Breast Volume Prediction Using Preoperative Anthropometric Measurements.

Area of Science:

• Plastic Surgery
• Biomedical Engineering
• Machine Learning Applications

Background:

• Accurate breast volume assessment is crucial for cosmetic and reconstructive breast surgery.
• Preoperative planning and intraoperative judgment benefit from precise volume data.
• Machine learning offers a novel approach to breast volume estimation.

Purpose of the Study:

• To develop an accurate formula for preoperative breast volume assessment using machine learning.
• To create a user-friendly application for real-time clinical application of the formula.

Main Methods:

• A prospective study involving 39 female-to-male transgender patients undergoing bilateral mastectomy.
• Collection of eight preoperative anthropometric measurements from 78 breasts.
• Development of a predictive model using a Gradient Boosted Model (Python CatBoostClassifier).

Main Results:

• Key measurements correlating with breast volume include breast width, nipple to IMF, sternal notch to nipple, and breast vertical perimeter.
• A validated formula was established with a high R-squared value (0.93) and low RMSE (62.4).
• An Android application, 'Breast Volume Predictor,' was developed for practical use.

Conclusions:

• The derived formula provides an accurate method for preoperative breast volume assessment.
• The 'Breast Volume Predictor' app facilitates real-time utilization of the formula in clinical settings.
• The application is freely available on the Google Play Store.