Proc. 26th Annual Int. Conf. of the IEEE Engineering in Medicine and Biology Society (EMBS), San Francisco, CA, 2004. Abstract—We develop an image registration system based on biomechanical modeling of the prostate and surrounding tissues to register cancerous tumor locations for targeted prostate brachytherapy treatment planning. Cancerous tumors can be identified using magnetic resonance spectroscopy (MRS) imaging, which is acquired with an endorectal probe that causes significant nonlinear deformation of the prostate. The probe is removed during magnetic resonance (MR) imaging for brachytherapy planning and therapy. Given 2-dimensional segmented MR and MRS images, our finite element based model defines a mapping between the probe-in/out images by estimating the deformation of the prostate and surrounding tissues due to endorectal probe insertion and balloon inflation. Treating uncertain patient-specific model parameters for tissue stiffness and external forces as variables, we compute a locally optimal solution to maximize image registration quality. We visualize results by applying the computed mapping to the MR image to generate a deformed MR image. We compare deformed MR images to corresponding MRS images for 5 patients and obtain an average Dice Similarity Coefficient (DSC) of 95.6% for the prostate. Using the mapping, we warp a regular spectroscopy grid from the MRS image to the probe-out MR image for use during treatment planning. Keywords—brachytherapy, finite element model, HDR, image registration, magnetic resonance imaging, MR, MRS, optimization, prostate cancer, spectroscopy I. INTRODUCTION Prostate cancer kills over 30,000 Americans each year [1]. It is often treated with brachytherapy, a minimally invasive medical procedure that places radioactive seeds in close proximity to cancerous tumors. High dose rate (HDR) brachytherapy uses a robot to move a single radioactive seed along approximately 20 catheters temporarily implanted inside the prostate. By adjusting the length of time (dwell time) that the seed remains at any location (dwell position) within a catheter, it is possible to generate and optimize over a wide variety of dose distributions [2]. The first step to create a targeted dose distribution is to precisely locate cancerous tumors in magnetic resonance (MR) images. Magnetic resonance spectroscopy (MRS) imaging has been effectively used to diagnose and locate cancerous tumors in the prostate [3, 4]. MRS imaging can be used to measure choline and citrate levels, which increase and decrease, respectively, with prostate cancer. To measure the low concentrations of choline and citrate, a probe containing an acquisition coil must be inserted endorectally to obtain sufficient sensitivity. The probe causes significant nonlinear translation and distortion of the prostate. The probe is removed during imaging for HDR treatment planning and therapy as shown in Fig. 1. To register MRS probe-in images to MR probe-out images, we develop a finite element based model that estimates the deformation of the prostate and surrounding tissues in the plane of the image due to the insertion and inflation of an endorectal probe balloon. A key problem with any biomechanical model is that required patient- specific model parameters are not known, including tissue stiffness properties for the prostate central gland, peripheral zone, and surrounding tissues. Additional unknown parameters include forces due to patient position changes, bladder volume changes, and other factors that differ between the MR and MRS images but are not explicitly included in our linear elasticity soft tissue deformation model. We use an optimization algorithm to solve for locally optimal patient-specific tissue stiffness properties and external forces that maximize image registration quality. The computed deformations result in a nonlinear warping of the spectroscopy grid, as shown in Fig. 3. II. RELATED WORK Past work on image registration of prostate deformations includes registering treatment and interventional MR images Image Registration for Prostate MR Spectroscopy Using Biomechanical Modeling and Optimization of Force and Stiffness Parameters Ron Alterovitz 1 , Ken Goldberg 1,2 , John Kurhanewicz 3 , Jean Pouliot 4 , I-Chow Hsu 4 1 Department of Industrial Engineering & Operations Research, University of California, Berkeley, CA, USA 2 Department of Electrical Engineering & Computer Science, University of California, Berkeley, CA, USA 3 Magnetic Resonance Science Center, University of California, San Francisco, CA, USA 4 Department of Radiation Oncology, University of California, San Francisco, CA, USA (a) MRS probe-in image (b) MR probe-out image Fig. 1. Spectroscopy data is obtained when the rectal balloon is inserted and inflated (a). HDR treatment is performed with the probe removed (b).