Multi-frequency Electromagnetic Tomography for Acute Stroke Detection Using Frequency Constrained Sparse Bayesian Learning IEEE Transactions on Medical Imaging, 2020, in press,
Jinxi Xiang (Tsinghua University & University of Edinburgh), Yonggui Dong (Tsinghua University), Yunjie Yang (University of Edinburgh)
Note: This figure is a typical conductivity spectrum of biological tissue. However, monotonically increasing conductivity is not a prerequisite for applying FCSBL.
Two motivations:
- Reconstruct multiple measurements simultaneously;
- Enhance the reconstructed quality especially when SNR is low.
Multiple Measurement Model (MMV) + Sparse Bayesian Learning (SBL).
Benefits: exploit spatial correlation and frequency correlation to reconstruct better images.
- Total variation. In the paper, DOI: 10.1109/TMI.2009.2022540 is used and the code is available from EIDORS. Here, a more flexible TV method (10.1109/TIP.2009.2028250) is provided.
- SA-SBL (DOI: 10.1109/TMI.2018.2816739). I don't have the copyright to make the original SASBL code public.
- FCSBL (proposed)
MATLAB 2019b, 32GB RAM memory, and a 6-core Intel, i7-8700 CPU. Please ensure sufficient RAM capacity, as the MMV model solves problems in higher dimensions.