%0 Journal Article %1 ankenbrand_sensitivity_2021 %A Ankenbrand, Markus J. %A Shainberg, Liliia %A Hock, Michael %A Lohr, David %A Schreiber, Laura M. %D 2021 %J BMC Medical Imaging %K bmd cctb markusankenbrand myown %N 1 %P 27 %R 10.1186/s12880-021-00551-1 %T Sensitivity analysis for interpretation of machine learning based segmentation models in cardiac MRI %U https://doi.org/10.1186/s12880-021-00551-1 %V 21 %X 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.

@article{ankenbrand_sensitivity_2021, abstract = {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.}, added-at = {2022-01-11T18:58:15.000+0100}, author = {Ankenbrand, Markus J. and Shainberg, Liliia and Hock, Michael and Lohr, David and Schreiber, Laura M.}, biburl = {https://www.bibsonomy.org/bibtex/2ad4cf0af2e57e7673e8c61c12fb0810d/iimog}, copyright = {All rights reserved}, doi = {10.1186/s12880-021-00551-1}, file = {Springer Full Text PDF:/home/markus/Zotero/storage/VQNNNDCR/Ankenbrand et al. - 2021 - Sensitivity analysis for interpretation of machine.pdf:application/pdf}, interhash = {f7466bdfb7a79155e36c1bf866eb19b7}, intrahash = {ad4cf0af2e57e7673e8c61c12fb0810d}, issn = {1471-2342}, journal = {BMC Medical Imaging}, keywords = {bmd cctb markusankenbrand myown}, language = {en}, month = feb, number = 1, pages = 27, timestamp = {2022-01-11T18:58:15.000+0100}, title = {Sensitivity analysis for interpretation of machine learning based segmentation models in cardiac {MRI}}, url = {https://doi.org/10.1186/s12880-021-00551-1}, urldate = {2021-09-16}, volume = 21, year = 2021 }