VASCULAR IMAGING New techniques for the reconstruction of complex vascular anatomies from MRI images DAVID H. FRAKES, 1,2, * MARK J. T. SMITH, 3 JAMES PARKS, 4 SHIVA SHARMA, 5 MARK FOGEL, 6 and AJIT P. YOGANATHAN 1 1 Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology, Atlanta, Georgia, USA 2 4-D Imaging, Inc., Atlanta, Georgia, USA 3 School of Electrical and Computer Engineering, Purdue University, West Lafayette, Indiana, USA 4 Sibley Heart Center, Children’s Healthcare of Atlanta, Atlanta, Georgia, USA 5 Atlanta Cardiology Associates, Atlanta, Georgia, USA 6 Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA The accurate representation of two-dimensional images in three dimensions has become important for many medical imaging applications and for cardiac magnetic resonance imaging (MRI) in particular. Reconstruction methods applied after data acquisition can produce three- dimensional information from two-dimensional data and make applications such as surgical planning more effective. Current reconstruction techniques usually demand contrast agents, and can suffer due to poor segmentation and sampling constraints that cause surface irregularities and distort dimensions. The novel technique presented here for anatomical modeling uses adaptive control grid interpolation (ACGI) to approximate data not captured by scanning, and a progressive shape-element segmentation technique to complete reconstruction. Quantitative validations conducted on models of pediatric cardiac malformations have confirmed the theoretical advantages of this technique, and that higher quality is achieved than with competing methods based on geometric parameters. Vascular diameters from reconstructions showed errors of less than 1% for a known geometry as compared to over 9% for competing methods. Qualitatively, models produced with the new methodology displayed substantial improvement over alternatives. Approximately 50 rare cardiac structures, including surgically altered Fontan and atypical aortic anatomies, have been reconstructed. All data used to create these reconstructions were acquired using standard pulse sequences and without contrast agents. Benefits of the new technique are particularly evident when complex vascular configurations complicate reconstruction. The proposed methodology enables a powerful tool allowing physicians to analyze and manipulate highly accurate and clearly presented vascular structures in an interactive medium. Key Words: Reconstruction; Segmentation; Surgical planning; Vascular; Fontan 1. Introduction The advent of magnetic resonance imaging (MRI) has equipped clinicians with valuable technology for the non- invasive analysis of anatomy and physiology. In one of its most common forms, two-dimensional (2D) imaging, MRI acquires parallel planar samples from a three-dimensional (3D) data source. These samples can later be reconstructed to model specific 3D structures from the original imaging target. Cardiovascular medicine is one area that has been advanced by MRI, and now relies heavily on both the modality itself and reconstruction to enable applications such as surgical planning. Like all imaging modalities, MRI has its shortcomings, one of which is the tradeoff between spatial resolution and signal-to-noise ratio (SNR). Averaging greater numbers of signals and performing oversampling can offer benefits, but do nothing to change this fundamental compromise. For many MR applications based on 2D imaging, in-plane resolution is deemed more important and out-of-plane resolution is sacrificed in order to maintain SNR. Under other circum- stances, an image stack composed of isotropic voxels is more desirable, and is acquired at the expense of in-plane resolution. Both of these cases present problems that relate to the interpolation and segmentation components of reconstruction scenarios. In the first case, data sets are commonly gathered with in-plane pixel dimensions of less than 1 mm but with slice thicknesses of 3 to 5 mm, which becomes problematic for 3D reconstruction. To counteract this, interpolation is used to approximate information lost to undersampling. Examples of raw and interpolated data sets are offered in Fig. 1. Like all interpolation problems, the quality of approximated data in the MR case is a function of the methods used and can vary significantly among techniques. The data in Fig. 1(b) were interpolated with the novel approach to be presented here. Journal of Cardiovascular Magnetic Resonance (2005) 7, 425–432 Copyright D 2005 Taylor & Francis Inc. ISSN: 1097-6647 print / 1532-429X online DOI: 10.1081/JCMR-200053637 Received 18 June 2004; accepted 11 November 2004. *Address correspondence to David H. Frakes, 4-D Imaging Inc., 75 Fifth St., Suite 331, Atlanta, GA 30308, USA; Fax: (404) 385- 4153; E-mail: dave@bme.gatech.edu 1097-6647 D 2005 Taylor & Francis Inc. 425 Order reprints of this article at www.copyright.rightslink.com