ISBN 978-3-8439-0549-7

978-3-8439-0549-7, Reihe Informatik

Torsten Hopp
Multimodal Registration of X-Ray Mammograms with 3D Volume Datasets

202 Seiten, Dissertation Universität Mannheim (2012), Hardcover, A5

Zusammenfassung / Abstract

X-ray mammography is the most common imaging method for breast cancer diagnosis. In case of unclear findings, additional volumetric modalities like Magnetic Resonance Imaging are often used to clarify the diagnosis. Due to different dimensionality of images and different compression state of the breast, multimodal diagnosis is challenging. The proposed fully-automated image registration method simplifies this radiological task by providing explicit co-localization between mo-dalities. The method is based on biomechanical modeling of the breast to simulate mammographic compression. An intensity-based optimization of the registration process allows automatic patient-specific adaption. The presented work covers a detailed description of the re-gistration method, an evaluation of the method and results obtained with a large number of clinical datasets. Furthermore a first integration into a clinical setting is described including data transfer, fully-automated processing and intuitive visualizations of registration results.