A High Resolution Dynamic Heart Model Based on Averaged MRI Data John Moore 1,2 , Maria Drangova 1 , Marcin Wierzbicki 1 , John Barron 2 , and Terry Peters 1 1 Robarts Research Institute and University of Western Ontario London, Ontario, Canada N6A 5K8 2 Dept. of Computer Science University of Western Ontario London, Ontario, Canada N6A 5B7 {jmoore,mdrangov,mwierz,tpeters}@imaging.robarts.ca Abstract. We are in the process of constructing a high resolution, high signal to noise ratio (SNR) dynamic MRI dataset for the human heart using methodology similar to that employed to construct a low-noise standard brain at the Montreal Neurological Institute. Several high reso- lution, low SNR magnetic resonance images of 20 phases over the cardiac cycle were acquired from a single subject. Images from identical phases and temporally adjacent phases were registered, and the image intensities were averaged together to generate a high resolution, high SNR dynamic magnetic resonance image volume of the human heart. Although this work is still preliminary, and the results still demonstrate residual arti- facts due to motion an sub-optimal alignment of interleaved image slices, our model has a SNR that is improved by a factor of 2.7 over a single volume, spatial resolution of 1.5 mm 3 , and a temporal resolution of 60 ms. 1 Introduction Magnetic resonance imaging (MRI) involves a compromise between spatial res- olution, signal to noise ratio (SNR), and acquisition time, among other factors. In the case of cardiac MRI, this compromise is further complicated by both heart motion and the fact that the images are typically acquired under breath- hold conditions. This compromise usually results in images with high in-plane resolution in two dimensions, but anywhere from 6 mm to 10 mm thick slices. Although super-high resolution may not be required for diagnosing many car- diac diseases, image guided surgery (IGS) would benefit from high resolution isotropic 3D images. The application of IGS to neurosurgery has benefited from the availability of a high resolution brain model [1,2]. Since heart surgery often requires a level of precision similar to that required for neurosurgery, we believe that cardiac IGS [3] can benefit from a high resolution, high SNR dynamic heart model. Linear or higher order interpolation is traditionally used to re-sample a non- isotropically sampled volume, to one that is sampled uniformly in all directions R.E. Ellis and T.M. Peters (Eds.): MICCAI 2003, LNCS 2878, pp. 549–555, 2003. c  Springer-Verlag Berlin Heidelberg 2003