@article{DawoodBreuerStebanietal.2023,
  author    = {Dawood, Peter and Breuer, Felix and Stebani, Jannik and Burd, Paul and Homolya, Istv{\´a}n and Oberberger, Johannes and Jakob, Peter M. and Blaimer, Martin},
  title     = {Iterative training of robust k-space interpolation networks for improved image reconstruction with limited scan specific training samples},
  series = {Magnetic Resonance in Medicine},
  volume    = {89},
  journal   = {Magnetic Resonance in Medicine},
  number    = {2},
  doi       = {10.1002/mrm.29482},
  url       = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-312306},
  pages     = {812 -- 827},
  year      = {2023},
  abstract  = {To evaluate an iterative learning approach for enhanced performance of robust artificial-neural-networks for k-space interpolation (RAKI), when only a limited amount of training data (auto-calibration signals [ACS]) are available for accelerated standard 2D imaging. Methods In a first step, the RAKI model was tailored for the case of limited training data amount. In the iterative learning approach (termed iterative RAKI [iRAKI]), the tailored RAKI model is initially trained using original and augmented ACS obtained from a linear parallel imaging reconstruction. Subsequently, the RAKI convolution filters are refined iteratively using original and augmented ACS extracted from the previous RAKI reconstruction. Evaluation was carried out on 200 retrospectively undersampled in vivo datasets from the fastMRI neuro database with different contrast settings. Results For limited training data (18 and 22 ACS lines for R = 4 and R = 5, respectively), iRAKI outperforms standard RAKI by reducing residual artifacts and yields better noise suppression when compared to standard parallel imaging, underlined by quantitative reconstruction quality metrics. Additionally, iRAKI shows better performance than both GRAPPA and standard RAKI in case of pre-scan calibration with varying contrast between training- and undersampled data. Conclusion RAKI benefits from the iterative learning approach, which preserves the noise suppression feature, but requires less original training data for the accurate reconstruction of standard 2D images thereby improving net acceleration.},
  language  = {en}
}
@article{GramGenslerWinteretal.2022,
  author    = {Gram, Maximilian and Gensler, Daniel and Winter, Patrick and Seethaler, Michael and Arias-Loza, Paula Anahi and Oberberger, Johannes and Jakob, Peter Michael and Nordbeck, Peter},
  title     = {Fast myocardial T\(_{1P}\) mapping in mice using k-space weighted image contrast and a Bloch simulation-optimized radial sampling pattern},
  series = {Magnetic Resonance Materials in Physics, Biology and Medicine},
  volume    = {35},
  journal   = {Magnetic Resonance Materials in Physics, Biology and Medicine},
  number    = {2},
  issn      = {1352-8661},
  doi       = {10.1007/s10334-021-00951-y},
  url       = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-268903},
  pages     = {325-340},
  year      = {2022},
  abstract  = {Purpose T\(_{1P}\) dispersion quantification can potentially be used as a cardiac magnetic resonance index for sensitive detection of myocardial fibrosis without the need of contrast agents. However, dispersion quantification is still a major challenge, because T\(_{1P}\) mapping for different spin lock amplitudes is a very time consuming process. This study aims to develop a fast and accurate T\(_{1P}\) mapping sequence, which paves the way to cardiac T1ρ dispersion quantification within the limited measurement time of an in vivo study in small animals. Methods A radial spin lock sequence was developed using a Bloch simulation-optimized sampling pattern and a view-sharing method for image reconstruction. For validation, phantom measurements with a conventional sampling pattern and a gold standard sequence were compared to examine T\(_{1P}\) quantification accuracy. The in vivo validation of T\(_{1P}\) mapping was performed in N = 10 mice and in a reproduction study in a single animal, in which ten maps were acquired in direct succession. Finally, the feasibility of myocardial dispersion quantification was tested in one animal. Results The Bloch simulation-based sampling shows considerably higher image quality as well as improved T\(_{1P}\) quantification accuracy (+ 56\%) and precision (+ 49\%) compared to conventional sampling. Compared to the gold standard sequence, a mean deviation of - 0.46 ± 1.84\% was observed. The in vivo measurements proved high reproducibility of myocardial T\(_{1P}\) mapping. The mean T\(_{1P}\) in the left ventricle was 39.5 ± 1.2 ms for different animals and the maximum deviation was 2.1\% in the successive measurements. The myocardial T\(_{1P}\) dispersion slope, which was measured for the first time in one animal, could be determined to be 4.76 ± 0.23 ms/kHz. Conclusion This new and fast T\(_{1P}\) quantification technique enables high-resolution myocardial T\(_{1P}\) mapping and even dispersion quantification within the limited time of an in vivo study and could, therefore, be a reliable tool for improved tissue characterization.},
  language  = {en}
}