@article{TerekhovElabyadSchreiber2021,
  author    = {Terekhov, Maxim and Elabyad, Ibrahim A. and Schreiber, Laura M.},
  title     = {Global optimization of default phases for parallel transmit coils for ultra-high-field cardiac MRI},
  series = {PLoS One},
  volume    = {16},
  journal   = {PLoS One},
  number    = {8},
  doi       = {10.1371/journal.pone.0255341},
  url       = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-265737},
  year      = {2021},
  abstract  = {The development of novel multiple-element transmit-receive arrays is an essential factor for improving B\(_1\)\(^+\) field homogeneity in cardiac MRI at ultra-high magnetic field strength (B\(_0\) > = 7.0T). One of the key steps in the design and fine-tuning of such arrays during the development process is finding the default driving phases for individual coil elements providing the best possible homogeneity of the combined B\(_1\)\(^+\)-field that is achievable without (or before) subject-specific B\(_1\)\(^+\)-adjustment in the scanner. This task is often solved by time-consuming (brute-force) or by limited efficiency optimization methods. In this work, we propose a robust technique to find phase vectors providing optimization of the B-1-homogeneity in the default setup of multiple-element transceiver arrays. The key point of the described method is the pre-selection of starting vectors for the iterative solver-based search to maximize the probability of finding a global extremum for a cost function optimizing the homogeneity of a shaped B\(_1\)\(^+\)-field. This strategy allows for (i) drastic reduction of the computation time in comparison to a brute-force method and (ii) finding phase vectors providing a combined B\(_1\)\(^+\)-field with homogeneity characteristics superior to the one provided by the random-multi-start optimization approach. The method was efficiently used for optimizing the default phase settings in the in-house-built 8Tx/16Rx arrays designed for cMRI in pigs at 7T.},
  language  = {en}
}
@article{AnkenbrandShainbergHocketal.2021,
  author    = {Ankenbrand, Markus J. and Shainberg, Liliia and Hock, Michael and Lohr, David and Schreiber, Laura M.},
  title     = {Sensitivity analysis for interpretation of machine learning based segmentation models in cardiac MRI},
  series = {BMC Medical Imaging},
  volume    = {21},
  journal   = {BMC Medical Imaging},
  number    = {1},
  doi       = {10.1186/s12880-021-00551-1},
  url       = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-259169},
  pages     = {27},
  year      = {2021},
  abstract  = {Background Image segmentation is a common task in medical imaging e.g., for volumetry analysis in cardiac MRI. Artificial neural networks are used to automate this task with performance similar to manual operators. However, this performance is only achieved in the narrow tasks networks are trained on. Performance drops dramatically when data characteristics differ from the training set properties. Moreover, neural networks are commonly considered black boxes, because it is hard to understand how they make decisions and why they fail. Therefore, it is also hard to predict whether they will generalize and work well with new data. Here we present a generic method for segmentation model interpretation. Sensitivity analysis is an approach where model input is modified in a controlled manner and the effect of these modifications on the model output is evaluated. This method yields insights into the sensitivity of the model to these alterations and therefore to the importance of certain features on segmentation performance. Results We present an open-source Python library (misas), that facilitates the use of sensitivity analysis with arbitrary data and models. We show that this method is a suitable approach to answer practical questions regarding use and functionality of segmentation models. We demonstrate this in two case studies on cardiac magnetic resonance imaging. The first case study explores the suitability of a published network for use on a public dataset the network has not been trained on. The second case study demonstrates how sensitivity analysis can be used to evaluate the robustness of a newly trained model. Conclusions Sensitivity analysis is a useful tool for deep learning developers as well as users such as clinicians. It extends their toolbox, enabling and improving interpretability of segmentation models. Enhancing our understanding of neural networks through sensitivity analysis also assists in decision making. Although demonstrated only on cardiac magnetic resonance images this approach and software are much more broadly applicable.},
  language  = {en}
}