Segmentation of Breast Region in Magnetic Resonance Images - Evaluation of Template-Based Approach

Automatic detection of suspicious lesions in breast MRI images is hindered by the presence of internal organs of similar intensities in post-contrast MRI images. In this study we applied a template-based algorithm to mask internal organs, efficiently removing anatomical structures outside the region of interest. In the absence of manual annotations of the region of interest, quantitative and qualitative analysis revealed that the proposed approach achieved satisfactory results in identification of internal organs in 99.07% subjects. After successful masking of internal organs, breast and axillary fossae region segmentation is efficiently performed using clustering and morphological operations.