ELASTIC IMAGE REGISTRATION BASED ON STRAIN ENERGY MINIMIZATION: APPLICATION TO PROSTATE MAGNETIC RESONANCE IMAGING

The use of magnetic resonance (MR) imaging in conjunction with an endorectal coil is currently the clinical standard for the diagnosis of prostate cancer. This approach has become recognized for its comparatively high sensitivity and specificity. However, imaging in this manner provides images and spectra of the prostate in the deformed state because of contact with the endorectal coil. Such deformation may lead to uncertainties in the localization of prostate cancer during therapy. In this dissertation we propose a novel 3-D elastic registration method. Unlike the traditional elastic registration methods which solve the force balance equations to achieve transformation between the source and target images, this algorithm derives the transformation by minimizing a physically motivated strain energy function. The minimization process is constrained by the matching features (i.e., contours or surfaces) in the source and target images that have been identified automatically by active contour/surface approaches. The Gauss-Seidel method was used in the numerical implementation of the strain energy minimization algorithm. The registration procedure was validated on synthetic digital images, on a physical prostate phantom and in vivo prostate MR images. The registration error on the synthetic digital images was 0.2 0.1 pixels. On the physical phantom and in vivo patient data, the errors were 1.0 0.6 pixels (0.6 0.4 mm), and 1.8 0.7 pixels (1.1 0.4 mm), respectively. Registration also improved image similarity (normalized cross correlation) from 0.72 0.10 to 0.96 0.03 on patient data. Registration results from the three forms of evaluation demonstrated that the proposed registration procedure can be used to significantly (p < 0.0001) improve the accuracy of image-guided therapies such as brachytherapy or external-beam therapy and can be valuable in the longitudinal follow-up of patients after therapy. When applied to in vivo prostate MR images the SEMT algorithm was significantly (p