Fang Liu

Date of Award


Degree Type


Degree Name

Master of Science


Medical Biophysics


Dr. Neil Gelman


Dynamic contrast-enhanced MRI (DCE-MRI) has been widely used in the diagnosis of breast cancer and as an aid in the management of this disease. Although DCE-MRI has a high sensitivity for the detection of malignant breast lesions, distinguishing malignant from benign lesions is more challenging for this method and may depend to some extent on how the images are analysed. Although clinical assessment of these images typically involves qualitative assessment by an expert, there is growing interest in the development of quantitative and automated methods to assist the expert assessment. This thesis involves the quantitative analysis of a particular empirical feature of the time evolution of the DCE-MRI signal known as the time-to-peak ( 7 ^ ) . In particular, this thesis investigates die feasibility of applying measures sensitive to 7 ^ heterogeneity as indicators for malignancy in breast DCE-MRI.

Breast lesions in this study were automatically segmented by K-means clustering. Voxel- by-voxel 7'peak values were extracted using an empirical model. The / 1th percentile values (p = 10, 20...) of the 7’peak distribution within each lesion, as well as the fractional and absolute hot spot volumes were determined, where hot spot volume refers to the volume of tissue with 7 ^ less than a threshold value. Using the area under the receiver operating characteristic curve (AUC), these measures were tested as indicators for differentiating fibroadenomas from invasive lesions and from ductal carcinoma in situ, as well as for differentiating non-fibroadenoma benign lesions from these malignant lesions. For differentiating fibroadenomas from malignant lesions, low percentile values (p = 10) provided high diagnostic performance. At the optimal threshold (3 min), the hot spot volume provided high diagnostic performance. However, non-fibroadenoma benign lesions were quite difficult to distinguish from malignant lesions. This thesis demonstrates that quantitative analysis of the 7’peak distribution can be optimized for diagnostic performance providing indicators sensitive to intra-lesion r peak heterogeneity.



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.