Whole tumor histogram analysis of early and late phase contrast-enhanced MRI in predicting residuals after neoadjuvant chemotherapy for mass-like breast cancer / 中国医学影像技术

ABSTRACT

Objective To investigate the diagnostic value of whole tumor histogram analysis of early and late phase contrast-enhanced MRI in predicting residual tumor after neoadjuvant chemotherapy (NAC) for mass-like breast cancer. Methods Totally 52 patients with mass-like breast cancer underwent contrast-enhanced MR scanning after NAC. ROIs were drawn on early phase images to cover the whole enhanced areas. Patients with enhanced signals on site of original tumor were divided into residual group (n=29) and non-residual group (n=9) according to pathological results. Parameters of histogram were compared between the 2 groups, and Logistic regression was used to analyze the impact factors of residual tumors, then diagnostic value of histogram parameters to residual tumors was assessed with ROC curves. Results There were significant differences in volume of enhanced areas, maximal signal intensity on early phase and kurtosis on late phase between residual group and non-residual group (all P<0.05). Logistic regression analysis showed that kurtosis on late phase was an independent factor of residual tumor (P=0.025, OR=4.303, 95%CI [1.205,15.359]). The diagnosis cutoff value of kurtosis on late phase was 2.25, the diagnostic specificity, sensitivity and accuracy rate were 76.19%(16/21), 90.32%(28/31) and 84.62%(44/52), respectively. Conclusion Histogram analysis is a useful method to reflect the internal distribution of signal intensities on early and late enhanced phases. Kurtosis on late phase is an independent factor for diagnosis of residuals after NAC for mass-like breast cancer.