Meeting Banner
Abstract #4003

Quantitative Assessment of Compressed-Sensing Reconstruction Fidelity for 3D He-3 and H-1 Acquisitions in One Breath-Hold

Kun Qing1, Talissa A. Altes2, Nicholas J. Tustison2, Jaime F. Mata2, G. Wilson Miller2, Eduard E. de Lange2, William A. Tobias3, Gordon D. Cates3, James R. Brookeman<s

1Biomedical Engineering, University of Virginia, Charlottesville, VA, United States; 2Radiology and Medical Imaging, University of Virginia, Charlottesville, VA, United States; 3Physics, University of Virginia, Charlottesville, VA, United States

The purpose of this work was to directly and quantitatively evaluate the reconstruction fidelity of undersampled compressed-sensing (CS) reconstructed 3D image sets by acquiring both the fully-sampled and undersampled acquisitions for a given nucleus (He-3 or H-1) in the same breath hold. Undersampled, CS-reconstructed H-1 3D images showed high similarity to their fully-sampled counterparts. While somewhat lower similarity indices were found for undersampled, CS-reconstructed He-3 3D image sets, this can be attributed to factors unrelated to the actual performance of the undersampled acquisition with CS reconstruction, including diaphragm movement, the non-equilibrium hyperpolarized magnetization, and B1 inhomogeneity.

Abstract PDF E-Poster Video

Keywords

acquiring acquisition acquisitions actual adding adjusting although applying approved arise arises assessment bandwidth best biomedical body breath chest clearly close coil common comparing compressed consistency consistent constant consumption containing contrast coronal counterparts decreases described diaphragm direct directly distribution edges engineering equilibrium equipped exchange experimental fidelity fold free full fully general generated good healthy hold holds human implemented included index indicated inhomogeneity insight institutional intensity interface issues larger least loss lung magnetization matrix medical middle miller modifies movement much nature noise nuclear nucleus offers optically optimal option original pattern patterns performance physician physics pixel polarized protocol proton pumped quantitative quantitatively radiology randomly rapid real reconstructed reconstruction reconstructions relatively representative resolution respectively roughly rubidium sampled sampling scaled scanned scanner sensing sets settings setup similarity similarly simultaneously since solutions somewhat sought spatial spin steady structural studies subject subjects suggests supported system technologies third typically uniformly unrelated upper vapor ventilated verified wavelet whole worse yields