Machine learning models for volume and weight estimation in breast reconstruction planning.

Area of Science:

• Biomedical Engineering
• Machine Learning Applications
• Medical Imaging and Data Analysis

Background:

• Accurate breast volume and weight estimation is crucial for post-mastectomy reconstruction.
• Current methods often involve high costs or complex procedures.
• A novel machine learning framework was developed to address these limitations.

Purpose of the Study:

• To develop and validate a machine learning framework for accurate breast volume and weight estimation.
• To leverage demographic and anthropometric data for preoperative breast assessment.
• To provide a cost-effective and accessible alternative to existing methods.

Main Methods:

• Data from 199 patients (2021-2023) were collected and pre-processed.
• Feature selection utilized domain expertise, Spearman's rank correlation, and the Boruta algorithm.
• Linear regression, random forest regression, and support vector regression models were trained and evaluated using R² and Pearson's correlation coefficient.

Main Results:

• Significant correlations were found between breast volume/weight and patient characteristics like BMI, cup size, and ptosis severity.
• The optimal linear regression model achieved an R² of 81.8% for breast volume and 72% for breast weight.
• The model integrated domain-expert and statistically selected features.

Conclusions:

• Machine learning integrating demographic and anthropometric data provides an accurate, interpretable, and accessible preoperative breast assessment method.
• This approach eliminates imaging costs and reliance on specialized equipment, utilizing routinely collected clinical data.
• The proposed model offers a practical, efficient, and cost-effective solution for clinical practice, overcoming traditional method limitations.