Non-Uniformly Sampled MR Correlated Spectroscopic Imaging in Breast Cancer and Nonlinear Reconstruction
Abstract
Magnetic Resonance Imaging (MRI) is an excellent anatomical tool to image tissue non-invasively. Dynamic contrast enhanced MRI (DCE-MRI) has excellent sensitivity but with varying specificity. MR Spectroscopy (MRS) enables biochemical characterization non-invasively of metabolites. Three major hypotheses are: (1) Accelerated 5-dimensional (5D) echo-planar imaging based correlated spectroscopic imaging (EP-COSI) acquisition produces multi-slice based multi-voxel two-dimensional (2D) MRS in a clinically feasible time. (2) Incorporating non uniform under sampling (NUS) for spectral/spatial sampling into the 5D EP-COSI data acquisition and the group sparsity (GS)-based reconstruction will reduce the total acquisition time by at least a factor of 12. (3) ADC values derived from the multi-slice DWI data will correlate negatively with choline groups and positively with lipids quantified by the EP-COSI technique. Three goals are proposed: i) NUS schemes will be combined with 5D EP-COSI sequence. ii) GS- based CS reconstruction schemes will be developed for accelerated acquisition and optimized to reconstruct the NUS EP-COSI data with better reliability . iii) Alterations in metabolite and lipid levels will be correlated with ADC changes in breast cancer patients compared to healthy women which will improve the diagnostic accuracy. The study patient cohort will include 50 patients with malignant breast carcinoma, 20 patients with benign breast tumor and 20 healthy women.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 2020
- Accession Number
- AD1129165
Entities
People
- Michael A Thomas
Organizations
- University of Southern California