Research Project
6485449
DESCRIPTION (provided by applicant): Dynamic contrast-enhanced MRI of the breast is a useful tool in locating lesions and differentiating between malignant and benign lesions. There are a variety of methods used to acquire and analyze the dynamic MRI data resulting in wide variability in the subsequent interpretation. We propose to develop a diagnostic system for extracting quantitative information from the dynamic MRI to assist the physician in the interpretation of the image data. The first step is a pre-processing step that applies deformable registration to the images to correct for any patient motion that may have occurred during the course of the exam. This correction will improve the results of subsequent processing and will improve the quality of the pre and post contrast subtraction images used to visualize potential lesions. Many researchers have found diagnostic utility in fitting the dynamic MRI to a pharmacokinetic model and extracting parametric information such as rate of enhancement, peak enhancement and rate of washout. We propose to implement and evaluate both a pharmacokinetic model and a non-parametric fit using B-splines and use the fitted data to extract the diagnostic parametric information and generate parametric images. In a novel approach, we propose to apply multi-channel segmentation to segment the parametric image data. The proposed system will be validated on a retrospective set of dynamic breast MR about from 35 patients acquired from our collaborators. PROPOSED COMMERCIAL APPLICATION: This dynamic MR image analysis package will be developed for Insightful's existing advanced and medical imaging workstation, IQuantify. Several commercialization avenues will be explored for this package: (1) We will offer it as an IQuantify module to pharmaceutical companies and CROs for clinical trials, (2) We will offer it as an OEM package to MR contrast agent companies and also to MR scanner manufacturera. (3) We will offer a clinical workstation for analysis of dynamic contrast MR images.
