Meeting Banner
Abstract #2043

Maximum Entropy Reconstruction of Non-Uniformly Under-Sampled Multidimensional Spectroscopic Imaging in vivo

Brian L. Burns1, Neil Wilson2, M. Albert Thomas, 23

1Department of Biomedical Engineering, UCLA, Los Angeles, CA, United States; 2Biomedical Physics IDP, UCLA, Los Angeles, CA, United States; 3Radiology, UCLA, Los Angeles, CA, United States

This work accelerates the acquisition of 4D MRSI data (2 spectral x 2 spatial) by under-sampling a spatial and a spectral dimension and reconstructing the missing data using Maximum Entropy (MaxEnt) non-linear reconstruction. Our results show that non-uniform under-sampling combined with MaxEnt is a viable way to reduce the scan times required of 4D MRSI to a clinically acceptable 6 minutes.

Abstract PDF

Keywords

accelerate acceleration acquisition aliased aliasing allow amplitude antigen applied around artifacts audience bandwidth biomedical bottom breast burns cancer channel characteristics characterization characterize citrate clinical clinically coherence coil combination competing conjugate constrained context convex correction correlated cross datasets decreasing deviation diagonal dimensions directly distribution distributions domain easily eddy either engineering ensemble entropy expansion expected filters form full fully going gradient grant healthy highlighted human idea in vivo increments info inverse iteration iterative iteratively long malignant mask maximize maximizing measure metabolites minute minutes mixed multidimensional near noise noisier optimization optimizing peaks percent phantom physics planar potential preserving prior problem processing proportional prospective prostate protocol quantitative radiology reconstruct reconstructed reconstructing reconstruction recording rectum reducing reduction removed represent represents requires researchers resolution resolved respectively retrospective routine sample sampled sampling scanner score select sensing simulated skilling solve spatial spectra spectral spectroscopic spectrum spins stern successfully take theory training trans transform translates truncation typical uniform uniformly unwanted volume year