Contractor shall obtain MRI data of a human body scanned in 1 mm increments both in the image sections and between sections. Data may be obtained either as GFE from the AL, from the "visible man" project at the University of Colorado Health Sciences Center, or from our subcontractor, the University of New Mexico's School of Medicine (Center for Non-Invasive Diagnosis). From the MRI data, an anatomical block model of a human will be constructed of 1 mm cells, where each sell will be assumed to be completely composed of 1 of 16 tissues types. Tissue typing information will be extracted from the MRI data by use of contract information. For instance, the scanner parameters can be adjusted to pick up differences in proton density, spin-lattice relaxation time (t1), and spin-spin relaxation time (t2). Thus, fat can, for example, be hyperintense on T1-weighted images and hypointense on T2-weighted images. Due to the huge amount of image data to process, anatomical boundaries will be determined and tissue types classified with maximal use of automation. Specifically, we will implement neural networks to classify tissue type from MRI data using human-identified tissues to train the networks.
Keywords: Human Modeling Human Modeling Specific Absorption Rate (Sar) Specific Absorption Rate (Sar) Mri Mri
