# Overlap‐Kernel EPI: Estimating MRI Shot‐to‐Shot Phase Variations by Shifted‐Kernel Extraction From Overlap Regions at Arbitrary k‐Space Locations

## TL;DR

This paper introduces a new method to correct phase inconsistencies in multi-shot MRI scans, improving image quality and flexibility in scan design.

## Contribution

A novel self-navigation strategy using shift-invariant kernel extraction from arbitrary k-space overlaps in multi-shot EPI.

## Key findings

- The method enables robust phase correction using overlap regions at any k-space location.
- Self-navigation EPI achieves reconstruction quality comparable to navigator corrections.
- The technique is effective for two-shot and four-shot mosaic EPI with improved SNR.

## Abstract

In multi‐shot EPI, shot‐dependent phase fluctuations can introduce ghost artifacts, undermining advantages for enhancing resolution or reducing distortion, particularly in diffusion scans. Here, a novel self‐navigation strategy based on shift‐invariant kernel extraction is proposed, enabling robust estimation of phase inconsistencies from overlap bands at any k‐space locations. It is inspired by RF sensitivity auto‐calibration in parallel imaging, and extends traditional self‐navigation trajectories beyond oversampled k‐space center.

Our shot‐to‐shot correction method only requires shot‐to‐shot sampling trajectories to slightly overlap at arbitrary k‐space locations. Such k‐space overlaps can be defined as shot‐dependent auto‐calibrating‐signal regions, where a GRAPPA/ESPIRiT‐type operation can be applied to explicitly extract shot‐to‐shot signal variation maps in one step. This inherits GRAPPA/ESPIRiT's robustness, and offers much greater flexibility in designing self‐navigated multi‐shot sequences.

Our kernel extraction techniques in four self‐navigated 2D SE‐EPI and GE‐EPI sequences are successfully demonstrated, as well as minor applications in extracting maps from navigator data and B
0 maps. Given sufficient SNR, self‐navigation EPI can reach identical/better reconstruction quality compared to navigator corrections, applicable to various trajectories. Additionally, it is particularly suitable for the two‐shot and four‐shot mosaic EPI, where stronger shot‐dependent eddy currents can also be removed, and the self‐navigated ACS has higher SNR than the refocused navigator echoes.

This paper rigorously analyzes subspace algorithms for phase‐stabilized multi‐shot MRI, significantly broadens the self‐navigation concept, and provides a more complete picture of the connections between signal fluctuations calibration in EPI and RF sensitivity calibration in parallel imaging.