4114
The feasibility of using computerized analysis features of magnetic resonance imaging to predict the outcomes of neoadjuvant chemotherapy for breast cancer1RADIOLOGY, PekingUniversity Shenzhen Hospital, SHENZHEN, China, 2RADIOLOGY, PekingUniversity ShenZhen Hospital, SHENZHEN, China
Synopsis
This study was based on mathematical calculation software to calculate the imaging features of dynamic enhanced magnetic resonance images(DCE-MRI). The feasibility of outcome prediction of neoadjuvant chemotherapy(NAC) for breast cancer was tested by three kinds of computerized analysis features: morphological features, grayscale statistical features, and texture features. The results indicated that DCE-MRI computerized analysis features before and after 2-cycles of NAC can not predict the degree of pathological remission of breast cancer. DCE-MRI computerized analysis features after 4-cycles of NAC can evaluate the degree of pathological remission of breast cancer. In other words, DCE-MRI computerized analysis features has the potential to be a new noninvasive method of evaluating NAC outcomes.
INTRODUCTION
Magnetic resonance imaging(MRI) is a routine method for assessing the efficacy of neoadjuvant chemotherapy for breast cancer. To date, most studies have focused on changes in dynamic enhancement ratio or apparent diffusion coefficients.[1-2] In recent years, the development of high-throughput computing and artificial intelligence has made computer-aided diagnosis(CAD) a great application in the medical field. As previously reported,[3] the imaging features of MRI can reflect the genetic characteristics of breast cancer, thereby reflecting the prognosis of breast cancer. However, these studies are limited to images after neoadjuvant chemotherapy or a single type of imaging feature.[4-5] For the first time, this study combined morphological features, grayscale statistical features, and texture features of images of the efficacy of NAC for breast cancer. The study is aimed to investigate the feasibility of using computerized analysis features to predict the outcomes of neoadjuvant chemotherapy(NAC) in patients with locally advanced breast cancer.METHODS
In this institutional review board approved study, 66 cases of breast cancer between March 2008 to March 2018 were included. Breast magnetic resonance(MR) imaging was performed before treatment(E1), at early stages of NAC(E2) and at end of NAC(E3). Two-dimensional computerized analysis was performed and the computerized analysis features were calculated on Matrix Laboratory(MATLAB). The following parameters were noted: rectangularity, sphericity, average, variance and entropy of radial length(based on the shape); mean, variance, consistency, skewness, kurtosis(based on histogram); energy, entropy, inertia, relevance, homogeneity(based on gray level co-occurrence matrix).These quantitative parameters were analyzed by t-test and least absolute shrinkage and selection operator(LASSO) regression.RESULTS
Major histological response (Miller-Payne grade 4 and 5) was achieved in 20(30.3%) of 66 cases. The computerized analysis features showed significant differences between major histological response and non-major histological response (p<0.05). LASSO regression model showed a good performance for the judgement of NAC outcomes(area under the receiver operating characteristic curve = 0.701).CONCLUSION
In locally advanced breast cancer, the MR imaging computerized analysis features after 4 cycles of NAC were significantly associated with pathologic complete response, may be a new noninvasive method of evaluating NAC outcomes.Acknowledgements
We thank the Department of Pathology, Department of Breast of PekingUniversity Shenzhen Hospital for their support. We are also grateful to the director of our department GUANXUN CHENG for supporting our study.References
1. Kim J H, Ko E S, Lim Y, et al. Breast Cancer Heterogeneity: MR Imaging Texture Analysis and Survival Outcomes[J]. Radiology, 2016, 282(3):160261.
2. Kim J J, Kim J Y, Kang H J, et al. Computer-aided Diagnosis-generated Kinetic Features of Breast Cancer at Preoperative MR Imaging: Association with Disease-free Survival of Patients with Primary Operable Invasive Breast Cancer[J]. Radiology, 2017:162079.
3. Chamming'S F, Ueno Y, Ferré R, et al. Features from Computerized Texture Analysis of Breast Cancers at Pretreatment MR Imaging Are Associated with Response to Neoadjuvant Chemotherapy.[J]. Radiology, 2017, 286(2):170143.
4. Henderson S, Muhammad N, Purdie C, et al. Breast cancer: influence of tumour volume estimation method at MRI on prediction of pathological response to neoadjuvant chemotherapy.[J]. British Journal of Radiology, 2018, 91(1087):20180123.
5. Fukada I, Araki K, Kobayashi K, et al. Pattern of Tumor Shrinkage during Neoadjuvant Chemotherapy Is Associated with Prognosis in Low-Grade Luminal Early Breast Cancer.[J]. Radiology, 2018, 286(1):161548.
Figures
