A deep learning-based background field removal method for brains containing high susceptibility sources

Xuanyu Zhu, Yang Gao, Feng Liu, ‪Stuart Crozier‬, Hongfu SunProceedings on CD-ROM - International Society for Magnetic Resonance in Medicine. Scientific Meeting and Exhibition/Proceedings of the International Society for Magnetic Resonance in Medicine, Scientific Meeting and Exhibition 2023 · DOI

Abstract

Background field removal (BFR) is a critical step in quantitative susceptibility mapping (QSM). Eliminating the background field in brains containing high susceptibility sources, such as intracranial hemorrhages, is challenging due to the relatively large scale of the local field induced from these sources. This study proposed a new deep learning-based method, "BFRnet", and compared it with several conventional BFR methods in processing two simulated and two in vivo brain datasets. The BFRnet method was effective in background field removal for acquisitions of arbitrary orientations and performed significantly better than other methods in the regions with high susceptibility sources.
BibTeX
@inproceedings{zhu2023deep,
  author = {Xuanyu Zhu and Yang Gao and Feng Liu and ‪Stuart Crozier‬ and Hongfu Sun},
  title = {{A deep learning-based background field removal method for brains containing high susceptibility sources}},
  booktitle = {Proceedings on CD-ROM - International Society for Magnetic Resonance in Medicine. Scientific Meeting and Exhibition/Proceedings of the International Society for Magnetic Resonance in Medicine, Scientific Meeting and Exhibition},
  year = {2023},
  doi = {10.58530/2022/2937},
}
Figure