Application of texture analysis to dynamic contrast enhanced breast magnetic resonance imaging

The aim of this thesis is to retrospectively evaluate dynamic contrast enhanced (DCE) breast magnetic resonance imaging (MRI) data. Texture analysis is applied to extract features from this data and test the feasibility of using these texture features for creating a computer aided diagnostic system (CAD).; Initial processing involved development of tools for automatic registration of DCE series. Computation time for texture feature extraction for the entire 3D dataset precludes whole breast analysis. Reduction of data can be best accomplished through the generation of angiogenesis (parametric maps) based on contrast agent kinetics and the knowledge that malignancy exhibits rapid uptake and plateau or slow washout on the DCE curve. Breast image data that exhibits these high-risk characteristics can be segmented into volumes-of-interest (VOIs). These volumes-of-interest can be projected on high resolution post contrast (T1) images using direction cosines information stored in the image header. The corresponding projected VOI on the high resolution, post contrast MR image can undergo feature extraction.; A simple 3D cursor viewer for selection of volumes-of-interest for feature extraction from the data was developed. The ability to port the angiogenesis maps on to high resolution post-contrast images helps us to select volumes in the high resolution T1 images as opposed to lower resolution dynamic series for texture feature extraction. The viewer also facilitates interactive surface rendering of the selected volume for a radiologist to examine the morphology of the lesion.; Texture features such as grey level co-occurrence matrix (GLCM ), second orientation pyramids (SOP), wavelets and gabor feature maps were extracted for these VOIs. In all a total of 53 texture features and 3 geometric features for each volume-of-interest was generated. Feature ranking was done using Fisher coefficient values. Classification was performed using off the shelf classifiers and also with minimum enclosing ball (MEB) classifier designed by us. The efficacy of texture features to classify suspect VOIs as malignant or benign was evaluated.; Results show that texture features have the ability to produce good sensitivity, specificity and accuracy for automatic lesion classification. Classification using the first 14 Fisher ranked coefficients produces the best results for mean sensitivity, specificity and accuracy.; In conclusion, results presented for machine classification of breast cancer detection from DCE MRI is encouraging. Clearly additional work in this area is required. The hypothesis that feature extraction and machine learning alone has produced good sensitivity and specificity can be proven with 95% significance level only by the study of much larger datasets.