Accurate image reconstruction from reduced data in pulsed electron paramagnetic resonance imaging

Objective:

We investigate and develop algorithms for reconstructing effective probe-density images, and then for obtaining oxygen-concentration images, from data of a subject collected at sparse views (SVs) or over a limited-angular range (LAR) for possibly achieving fast pulsed electron paramagnetic resonance imaging (EPRI). We refer to the effective probe-density image simply as the EPR image in the work.

Methods:

The reconstruction problem of EPR images from SV or LAR data in pulsed EPRI is formulated as an optimization program that includes a constraint either on the total variation (TV) or on the directional-TVs (DTVs) of the EPR image. Two algorithms, referred to as TV and DTV algorithms, are developed then for reconstruction of EPR images, respectively, from SV and LAR data through solving the respective optimization programs. Oxygen-concentration image is estimated subsequently from the EPR images reconstructed.

Results:

Using numerical studies with simulated data of a digital phantom and also with real data of a physical phantom and a mouse model, we demonstrate the potential of the TV and DTV algorithms that yield, respectively, from SV and LAR data, numerically accurate EPR and oxygen-concentration images.

Conclusion:

The TV and DTV algorithms developed can yield numerically accurate EPR and oxygen-concentration images, respectively, from SV and LAR data in pulsed EPRI.

Significance:

The work may yield insights into the design of scans with minimized scanning time, thus potentially enabling basic and preclinical in vivo studies with fast pulsed EPRI.