Abstract #3126
Use of L1-norm solution to Impose Spatial Smoothness Constraints in Quantitative T2 Relaxometry
Dushyant Kumar 1,2 , Susanne Siemonsen 1,2 , Margherita Porcelli 3 , Jens Fiehler 1 , Christoph Heesen 4 , and Jan Sedlacik 1
1
Dept. of Neuroradiology,
Universittsklinikum Hamburg-Eppendorf, Hamburg,
Hamburg, Germany,
2
Multiple
Sclerosis Imaging Section (SeMSI), Universittsklinikum
Hamburg-Eppendorf, Hamburg, Hamburg, Germany,
3
Mathematics,
University of Bologna, Bologna, Bologna, Italy,
4
Institute
for Neuroimmunology and Clinical MS Research,
Universittsklinikum Hamburg-Eppendorf, Hamburg,
Hamburg, Germany
Problem: A moderately high SNR (~200) QT2R data is
needed for robust tissue-water-fraction-map
reconstruction if L2-norm based spatial smoothness is
implemented. We are testing L1-norm-solver as other
possible candidate. Methods: We are developing
L1-norm-solver in this context and its performance is
compared against L2-norm-solver in context of imposing
spatial constraints. Results & Conclusions: Results
using L2- and L1-norms are similar at high SNR (>200);
however, L2-norm-solver performs better at lower SNR. In
near future, we would develop hybrid filter to impose
smoothness and sparsity simultaneously to make L1-norm
performs better at low SNR.
This abstract and the presentation materials are available to members only;
a login is required.
Join Here
