Quantitative Ultrasound Radiomics for Predicting and Monitoring Neoadjuvant Chemotherapy Response in Breast Cancer: A Systematic Review.

: Quantitative ultrasound (QUS) radiomics extracts microstructure-sensitive spectral features from radiofrequency data and may provide contrast-free, early indicators of neoadjuvant chemotherapy (NAC) response in breast cancer. This review synthesized open access human studies evaluating QUS radiomics for a priori prediction and early on-treatment monitoring. : Following PRISMA-2020, we included English, free full-text clinical studies of biopsy-proven breast cancer receiving NAC that reported QUS spectral parameters (mid-band fit, spectral slope/intercept) ± textures/derivatives and machine learning models against clinical/pathologic response. Data on design, RF acquisition/normalization, features, validation, and performance (area under the curve (AUC), accuracy, sensitivity/specificity, balanced accuracy) were extracted. : Twelve cohorts were included. A priori baseline models achieved accuracies of 76-88% with AUCs 0.68-0.90; examples include 87% accuracy in a multi-institutional study, 82% accuracy/AUC 0.86 using texture-derivatives, 86% balanced accuracy with transfer learning, 88% accuracy/AUC 0.86 with deep learning, and AUC 0.90 in a hybrid QUS and molecular-subtype model. Early monitoring improved discrimination: week-1 results ranged from AUC 0.81 to 1.00 and accuracy 70 to 100%, noting that the upper bound was reported in a small cohort using combined QUS and diffuse optical spectroscopy features, while week 4 typically peaked (AUC 0.87-0.91; accuracy 80-86% in observational cohorts), and one series reported week-8 accuracy of 93%. Across reporting cohorts, mean AUC increased with a 0.05 absolute gain. A randomized feasibility study reported prospective week-4 model accuracy of 98% and demonstrated decision impact. : QUS radiomics provides informative a priori prediction and strengthens by weeks 1-4 of NAC, supporting adaptive treatment windows without contrast or radiation. Standardized radiofrequency (RF) access, normalization, region of interest (ROI)/margin definitions, and external validation are priorities for clinical translation.