Journal
FRONTIERS IN ONCOLOGY
Volume 13, Issue -, Pages -Publisher
FRONTIERS MEDIA SA
DOI: 10.3389/fonc.2023.1057841
Keywords
breast cancer; radiomics; multiparametric MRI; neoadjuvant chemotherapy; concentric shrinkage
Categories
Ask authors/readers for more resources
This study aimed to evaluate the tumor shrinkage pattern and response to neoadjuvant chemotherapy (NACT) using preoperative multiparametric magnetic resonance imaging (MRI). The results showed that multiparametric MRI features, combined with molecular subtype information, can be of important clinical value for the preoperative prediction of treatment response and shrinkage pattern.
PurposeDuring neoadjuvant chemotherapy (NACT), breast tumor morphological and vascular characteristics are usually changed. This study aimed to evaluate the tumor shrinkage pattern and response to NACT by preoperative multiparametric magnetic resonance imaging (MRI), including dynamic contrast-enhanced MRI (DCE-MRI), diffuse weighted imaging (DWI) and T2 weighted imaging (T2WI). MethodIn this retrospective analysis, female patients with unilateral unifocal primary breast cancer were included for predicting tumor pathologic/clinical response to NACT (n=216, development set, n=151 and validation set, n=65) and for discriminating the tumor concentric shrinkage (CS) pattern from the others (n=193; development set, n=135 and validation set, n=58). Radiomic features (n=102) of first-order statistical, morphological and textural features were calculated on tumors from the multiparametric MRI. Single- and multiparametric image-based features were assessed separately and were further combined to feed into a random forest-based predictive model. The predictive model was trained in the testing set and assessed on the testing dataset with an area under the curve (AUC). Molecular subtype information and radiomic features were fused to enhance the predictive performance. ResultsThe DCE-MRI-based model showed higher performance (AUCs of 0.919, 0.830 and 0.825 for tumor pathologic response, clinical response and tumor shrinkage patterns, respectively) than either the T2WI or the ADC image-based model. An increased prediction performance was achieved by a model with multiparametric MRI radiomic feature fusion. ConclusionsAll these results demonstrated that multiparametric MRI features and their information fusion could be of important clinical value for the preoperative prediction of treatment response and shrinkage pattern.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available