Table 1.
Relevant patient-specific information and disease characteristics.
Fig 1.
Ultrasound B-mode and QUS-derived parametric maps for representative responder and non-responder patients (responder—left, non-responders—right panel) acquired at baseline, and weeks 1 and 4 of treatment.
Abbreviations: MBF (dB): mid-band fit, SS (dB/MHz): spectral slope, ASD (μm): average scatterer diameter, AAC (dB/cm3): average acoustic concentration, SI (dB): spectral intercept. Scale bar represents 2 cm.
Fig 2.
Statistically significant QUS parameters between responders and non-responders at weeks 1 and 4 of neoadjuvant chemotherapy.
Error bars represent ± one standard error of the mean, and significance was determined at p < 0.05. Abbreviations: AAC (dB/cm3): average acoustic concentration, ACE: Attenuation Coefficient Estimate.
Fig 3.
Scatter plots of QUS parameters comparing responders and non-responders at week 1.
Error bars represent ± one standard error of the mean, and significance was determined at p < 0.05.
Fig 4.
Scatter plots of QUS parameters comparing responders and non-responders at week 4.
Error bars represent ± one standard error of the mean, and significance was determined at p < 0.05.
Table 2.
Statistically significant QUS mean values and textural parameters between response groups at week 1 and week 4 into neoadjuvant chemotherapy.
Table 3.
Optimal multivariate-feature classification analysis using machine learning algorithms in week 1 and week 4 during neoadjuvant chemotherapy.
Fig 5.
Receiver operating characteristic curves of QUS feature selection using machine learning algorithms from data acquired before initiation and at weeks 1 and 4 of neoadjuvant chemotherapy.
Area under curve (AUC) values are indicated in the respective curves.