Chunli Li1,2, Lirong Song2, Jiandong Yin2. 1. Department of Biomedical Engineering, School of Fundamental Sciences, China Medical University, Shenyang, China. 2. Department of Radiology, Shengjing Hospital of China Medical University, Shenyang, China.
Abstract
BACKGROUND: Radiomics has been applied to breast magnetic resonance imaging (MRI) for gene status prediction. However, the features of peritumoral regions were not thoroughly investigated. PURPOSE: To evaluate the use of intratumoral and peritumoral regions from functional parametric maps based on breast dynamic contrast-enhanced MRI (DCE-MRI) for prediction of HER-2 and Ki-67 status. STUDY TYPE: Retrospective. POPULATION: A total of 351 female patients (average age, 51 years) with pathologically confirmed breast cancer were assigned to the training (n = 243) and validation (n = 108) cohorts. FIELD STRENGTH/SEQUENCE: 3.0T, T1 gradient echo. ASSESSMENT: Radiomic features were extracted from intratumoral and peritumoral regions on six functional parametric maps calculated using time-intensity curves of DCE-MRI. The intraclass correlation coefficients (ICCs) were used to determine the reproducibility of feature extraction. Based on the intratumoral, peritumoral, and combined intra- and peritumoral regions, three radiomics signatures (RSs) were built using the least absolute shrinkage and selection operator (LASSO) logistic regression model, respectively. STATISTICAL TESTS: Wilcoxon rank-sum test, minimum redundancy maximum relevance, LASSO, receiver operating characteristic curve (ROC) analysis, and DeLong test. RESULTS: The intratumoral and peritumoral RSs for prediction of HER-2 and Ki-67 status achieved areas under the ROC (AUCs) of 0.683 (95% confidence interval [CI], 0.574-0.793) and 0.690 (95% CI, 0.577-0.804), and 0.714 (95% CI, 0.616-0.812) and 0.692 (95% CI, 0.590-0.794) in the validation cohort, respectively. The combined RSs yielded AUCs of 0.713 (95% CI, 0.604-0.823) and 0.749 (95% CI, 0.656-0.841), respectively. There were no significant differences in prediction performance among intratumoral, peritumoral, and combined RSs. Most (69.7%) of the features had good agreement (ICCs >0.8). DATA CONCLUSION: Radiomic features of intratumoral and peritumoral regions on functional parametric maps based on breast DCE-MRI had the potential to identify HER-2 and Ki-67 status. LEVEL OF EVIDENCE: 3 Technical Efficacy Stage: 2.
BACKGROUND: Radiomics has been applied to breast magnetic resonance imaging (MRI) for gene status prediction. However, the features of peritumoral regions were not thoroughly investigated. PURPOSE: To evaluate the use of intratumoral and peritumoral regions from functional parametric maps based on breast dynamic contrast-enhanced MRI (DCE-MRI) for prediction of HER-2 and Ki-67 status. STUDY TYPE: Retrospective. POPULATION: A total of 351 female patients (average age, 51 years) with pathologically confirmed breast cancer were assigned to the training (n = 243) and validation (n = 108) cohorts. FIELD STRENGTH/SEQUENCE: 3.0T, T1 gradient echo. ASSESSMENT: Radiomic features were extracted from intratumoral and peritumoral regions on six functional parametric maps calculated using time-intensity curves of DCE-MRI. The intraclass correlation coefficients (ICCs) were used to determine the reproducibility of feature extraction. Based on the intratumoral, peritumoral, and combined intra- and peritumoral regions, three radiomics signatures (RSs) were built using the least absolute shrinkage and selection operator (LASSO) logistic regression model, respectively. STATISTICAL TESTS: Wilcoxon rank-sum test, minimum redundancy maximum relevance, LASSO, receiver operating characteristic curve (ROC) analysis, and DeLong test. RESULTS: The intratumoral and peritumoral RSs for prediction of HER-2 and Ki-67 status achieved areas under the ROC (AUCs) of 0.683 (95% confidence interval [CI], 0.574-0.793) and 0.690 (95% CI, 0.577-0.804), and 0.714 (95% CI, 0.616-0.812) and 0.692 (95% CI, 0.590-0.794) in the validation cohort, respectively. The combined RSs yielded AUCs of 0.713 (95% CI, 0.604-0.823) and 0.749 (95% CI, 0.656-0.841), respectively. There were no significant differences in prediction performance among intratumoral, peritumoral, and combined RSs. Most (69.7%) of the features had good agreement (ICCs >0.8). DATA CONCLUSION: Radiomic features of intratumoral and peritumoral regions on functional parametric maps based on breast DCE-MRI had the potential to identify HER-2 and Ki-67 status. LEVEL OF EVIDENCE: 3 Technical Efficacy Stage: 2.