Radiomics and Machine Learning with Multiparametric Breast MRI for Improved Diagnostic Accuracy in Breast Cancer...

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence

Background:

• Accurate breast cancer detection is crucial for timely treatment.
• Multiparametric MRI, including dynamic contrast-enhanced (DCE) and diffusion-weighted imaging (DWI), shows promise in characterizing breast lesions.
• Radiomics and machine learning (ML) offer advanced analytical tools for medical image interpretation.

Purpose of the Study:

• To evaluate the effectiveness of radiomics analysis combined with ML for breast cancer detection.
• To compare the performance of separate DCE-MRI and DWI radiomics models with a combined multiparametric model.
• To assess the potential of this approach in improving the diagnostic accuracy of suspicious breast tumors.

Main Methods:

• A multicenter retrospective study included 93 patients with BI-RADS 4 breast lesions.
• Radiomics features were extracted from DCE-MRI and DWI datasets.
• Machine learning models, including support vector machine (SVM), were developed to differentiate between malignant and benign lesions using selected radiomics features.

Main Results:

• The DWI radiomics model achieved an AUC of 0.79.
• The DCE-MRI radiomics model yielded an AUC of 0.83.
• The combined multiparametric radiomics model demonstrated the highest performance with an AUC of 0.85 and diagnostic accuracy of 81.7%.

Conclusions:

• Radiomics analysis coupled with ML of multiparametric MRI significantly improves the evaluation of suspicious breast tumors.
• The combined DCE- and DWI-based radiomics model offers superior diagnostic performance compared to individual modalities.
• This approach facilitates accurate breast cancer diagnosis and may help reduce unnecessary benign breast biopsies.