Analysis And Prediction Of Neoadjuvant Chemotherapy Response For Breast Cancer Based On DCE-MRI Breast Imaging

Breast cancer is a serious threat to women's physical and mental health.The morbidity of breast cancer in the country is still an increasing trend in recent years,and is the highest of malignancy in female.At present,the Neoadjuvant chemotherapy(NAC)for breast cancer has become a standard model for breast cancer treatment,and clinical studies have demonstrated the increased survival rate of patients with complete remission after NAC.However,although the pathological response is usually obtained after the NAC,the optimization of the chemotherapy regimen,the choice of the operation time and the implementation of breast conserving therapy require accurate estimation of the pathological response of breast cancer.Therefore,it's important to evaluate and predict the efficacy of NAC in breast cancer,as well as to find a reliable non-invasive evaluation method.Magnetic resonance imaging(MRI)technology has gradually become an important information source for clinical examination of breast diseases due to its rich image information,high soft tissue resolution and no radiation damage.The dynamic contrast-enhanced magnetic resonance imaging(DCE-MRI)technology can be used to understand the functional information of the tumor,such as blood supply,vascular permeability,cell structure,gap between the cells.This technology has become an important clinical diagnostic basis for breast cancer.In addition,the features of breast DCE-MRI enhancement is associated with the biological characteristics of the micro vascular,angiogenesis,providing a basis for the rational development of breast cancer treatment and prognosis assessment.This thesis mainly discusses the value of breast DCE-MR imaging in evaluating and predicting the efficacy of NAC for breast cancer.To this end,fifty-seven cases of malignant breast cancers with MRI examination both before and after two cycle of NAC were retrospectively analyzed.These cases were divided into pathologic responders and non-pathologic responders according to the pathological reaction.Firstly,the morphological features,statistical features,textural features and dynamic contrast-enhanced features were extracted in breast lesions and background,respectively.Then,statistical analysis was performed.Finally,a classification and prediction model of the effect of neoadjuvant chemotherapy were unutilized for prediction of response to NAC.The thesis is organized as follows:(1)Breast DCE-MRI image preprocessing.In order to improve the reliability of the results of the study,this study has carried on the breast target region segmentation to the breast DCE-MR image.Regions of lesion and breast background could be respectively achieved using semi-automatically segmentation algorithm under the guidance of experienced radiologists in breast DCE-MR imaging before and after two cycles of NAC.(2)Breast DCE-MRI image feature extraction of breastIn this study,126 dimensional image features were extracted.Specifically,44 dimensions features in the lesion region were extracted including dynamic feature,texture feature,morphological characteristics and statistical characteristics,and 82 dimensional features were extracted,including dynamic feature and texture feature in the background region of the breast.Since the DCE-MR imaging has spatial characteristics,the dynamic feature and texture feature were extracted in the 3D space.(3)Statistical analysis on Breast DCE-MRI imaging features and the clinical efficacy of NACIn order to investigate whether the features of breast DCE-MR imaging has a predictive value for NAC,the relationship between the former and the latter has been assessed using statistical analyses.Specifically,pairwised t test was used to evaluate differences of image features between MRI examinations before and after NAC.Moreover,the associations of these image features with response to NAC were assessed using logistic regression.(4)Prediction of the effect of NAC for breast cancer based on breast DCE-MRI imagesIn order to further investigate the ability of imaging features in breast DCE-MRI for prediction of the response of NAC in breast cancer,we use genetic algorithm to select the optimal feature subset.As a result,the optimal feature subset selection were used to evaluate the efficacy of DCE-MRI for prediction of response of NAC.In this thesis,we have explored the correlation and predictive ability of DCE-MRI imaging features for response of NAC in breast.The results showed that the characteristics of breast DCE-MR imaging in breast cancer were significantly correlated with the response of NAC.In addition,the DCE-MR imaging features before NAC have predictive ability for response of NAC in breast cancer.