DeepSTI: Towards tensor reconstruction using fewer orientations in susceptibility tensor imaging

Zhenghan Fang, Kuo-Wei Lai, Peter C. M. van Zijl, Xu Li, Jeremias Sulam

Abstract: Highlights•We present DeepSTI, a method for STI reconstruction with data-driven priors.•DeepSTI allows for STI from significantly fewer orientations than previous methods.•Promising results are obtained from only one orientation in human in vivo.•High-resolution fiber tractography results are obtained from DeepSTI reconstructions.•We demonstrate an application of DeepSTI for multiple sclerosis characterization.