Deciphering Breast Cancer Complexity: A Study on the Predictive Power of MRI Texture Analysis for Tumor Characterization and Treatment Response

Abstract

This study examined the association between MRI radiomics features of malignant breast masses and their histopathological, molecular characteristics, and response to neoadjuvant treatments. Utilizing a retrospective cohort of 70 breast cancer patients, this study conducted a detailed texture analysis on preoperative MRI scans, including the extraction of texture features, such as entropy, contrast, and homogeneity, and their analysis against histopathological and clinical target variables (e.g., lymph node metastasis) and molecular profiles. Statistical analyses and machine learning algorithms, including logistic regression and support vector machines, were employed to evaluate the predictive power of MRI texture features for determining molecular subtypes and evaluate the association of radiomics markers with neoadjuvant treatments. The findings reveal significant associations between specific MRI texture features and the histopathological and molecular characteristics of breast tumors, demonstrating that certain texture parameters are strongly associated with aggressive tumor phenotypes and poorer response to chemotherapy. Despite the limited dataset, machine learning models showcased promising performance in classifying tumors into molecular subtypes and predicting treatment outcomes, highlighting the potential of MRI texture analysis in clinical decision-making. This study underscores the potential of MRI texture analysis as a non-invasive tool for enhancing the personalized management of breast cancer. The significant associations between MRI texture features and critical tumor characteristics suggest that these features could serve as valuable biomarkers for predicting tumor behavior and treatment efficacy. The findings advocate for further large-scale research into integrating MRI texture analysis into clinical practice to tailor treatment strategies to individual tumor profiles, ultimately aiming to improve patient outcomes in breast cancer treatment.