Low-rank iterative infilling for zero echo-time (ZTE) imaging

A new referenceless low-rank reconstruction technique has been introduced to address the issue of missing samples within the Zero Echo Time (ZTE) dead-time gap.

The proposed method reformulates the in-filling of the missing samples as an inverse problem subject to low-rank constraints. Its performance and robustness are evaluated through a comparative analysis that combines Monte Carlo computational simulations and data obtained from in vivo experiments.

The proposed method is tested for dead-time gaps ranging up to 4.5 Nyquist dwells, across signal-to-noise ratio levels of 5, 10, 15, and 20 dB. Consistently superior performance is observed across all cases compared to algebraic and parallel imaging methods. The speed for convergence decreases exponentially as the dead-time gap expands.

The proposed method enables artifact-free reconstruction up to dead-time gap of 4 Nyquist dwells and thereby supports ZTE imaging up to an imaging bandwidth of $\pm 41.67$ kHz (assuming transmit and receive switching less than 30 $\mu$ s). It demonstrates superior performance compared to algebraic and parallel imaging methods.