Linear Regression Modeling Based Scoring System to Reduce Benign Breast Biopsies Using Multi-parametric US with Color...

Area of Science:

• Radiology
• Medical Imaging
• Oncology

Background:

• Benign breast biopsies are common but invasive procedures.
• Accurate differentiation of benign from malignant breast lesions is crucial to avoid unnecessary interventions.

Purpose of the Study:

• To develop and validate a simple ultrasound (US) based scoring system to reduce benign breast biopsies.
• To improve the diagnostic performance of breast lesion characterization using US and shear-wave elastography (SWE).

Main Methods:

• A scoring system was developed using linear regression models based on standard US, color Doppler US (CDUS), and SWE parameters.
• The model was trained on 418 breast lesions and validated on an independent dataset of 187 lesions.
• Decision tree analysis was used to convert continuous parameters into categorical equivalents for regression modeling.

Main Results:

• The combined US and SWE model achieved an area under the curve (AUC) of 0.882, outperforming individual parameters.
• The scoring system could have potentially avoided 37.8% of benign breast biopsies while missing only 1.1% of malignant lesions.
• Validation demonstrated a 100% negative predictive value (NPV) and 74.6% specificity.

Conclusions:

• A scoring system integrating US and SWE parameters offers superior diagnostic performance for breast lesions compared to US alone.
• This novel scoring method has the potential to significantly decrease the rate of benign breast biopsies.
• The validated system shows high accuracy in identifying benign lesions, reducing the need for invasive procedures.