Joint reconstruction is relevant for a variety of medicalimaging applications, where multiple images are acquired in parallel orwithin a single scanning procedure. Examples include joint reconstruction of different medical imaging modalities (e.g. CT and PET) and various MRI applications (e.g. different MR imaging contrasts of the same patient). In this paper we present an approach for joint reconstruction of two MR images, based on partial sampling of both. We assume each MR image has a limited number of edges, that is, low total variation,but they are similar in the sense that many of the edges overlap. We examine synthetic phantoms representing T1 and T2 imaging contrasts and realistic T1-weighted and T2-weighted images of the same patient. We show that our joint reconstruction approach outperforms conventional TV-based MRI reconstruction for each image solely. Results are shown both visually and numerically for sampling ratios of 4%-20%, and consist of an improvement of up to 3.6 dB.
|Publication status||Published - 2017|
|Event||6th Signal Processing with Adaptive Sparse Structured Representations workshop 2017 - Lisbon, Portugal|
Duration: 5 Jun 2017 → 8 Jun 2017
|Workshop||6th Signal Processing with Adaptive Sparse Structured Representations workshop 2017|
|Abbreviated title||SPARS 2017|
|Period||5/06/17 → 8/06/17|