Maximizing Unambiguous Velocity Range in Phase-Contrast MRI with Multipoint Encoding

Shen Zhao, Rizwan Ahmad, Lee C. Potter

TL;DR

This paper proposes a new method for analyzing phase-contrast MRI data that maximizes the range of velocities that can be accurately measured without ambiguity. By jointly processing all available phase difference measurements, the authors demonstrate a significant increase in both the measurable velocity range and the signal-to-noise ratio compared to existing techniques that simplify the data.

What Problem Does This Paper Address?

Phase-contrast MRI (PC-MRI) is a powerful technique for non-invasively measuring blood flow in vivo. It works by encoding the velocity of spins into the phase of the MR signal. However, a fundamental trade-off exists: increasing the sensitivity to velocity (which allows measurement of faster flows) also leads to "phase aliasing," where faster velocities are incorrectly mapped to slower ones. This limits the maximum velocity that can be accurately measured. Existing methods often simplify the complex MRI data by processing only a subset of measurements, which can further reduce the measurable velocity range or the velocity-to-noise ratio (VNR). This paper addresses the challenge of maximizing the unambiguous velocity range and VNR simultaneously by proposing a method to jointly process all available phase difference measurements.

Methods

The core of the paper's contribution lies in its theoretical framework and its application to existing PC-MRI acquisition schemes. The authors leverage the concept of a "fundamental parallelepiped" derived from linear algebra, which mathematically defines the intrinsic unambiguous velocity range for a given encoding scheme. They formulate the problem as a system of congruence equations (Av + ε = φ mod 2π), where 'A' is the encoding matrix, 'v' is the velocity vector, 'ε' is noise, and 'φ' represents the measured phase differences.

Instead of the common practice of reducing the number of equations by pre-processing, the authors advocate for jointly solving all 'N' phase difference equations. This "joint processing" allows them to maintain the fullest extent of the unambiguous range intrinsically provided by the encoding matrix 'A'. They also show that this approach maximizes the velocity-to-noise ratio. The paper then extends this concept to analyze existing multi-point encoding strategies (4-point and 5-point) and demonstrates how their joint processing approach yields larger unambiguous ranges compared to prior pre-processing methods. They also explore how modifying the encoding matrix itself (perturbed encodings) can further expand this range, illustrating the potential for designing new acquisition schemes.

Datasets

Not discussed in the paper. The paper focuses on theoretical analysis and simulation-based demonstrations of the proposed method.

Key Insights and Ideas

  • The Unambiguous Range is a Parallelepiped: The paper elegantly frames the intrinsic unambiguous velocity range as a parallelepiped, providing a clear geometric interpretation of the limits of velocity encoding.
  • Joint Processing is Key: The central insight is that by avoiding pre-processing and instead jointly analyzing all phase differences, one can achieve both a larger unambiguous velocity range and a higher VNR. This directly challenges the conventional wisdom of simplifying data for tractability.
  • Lattice Theory for Velocity Encoding: The application of concepts from number theory (congruence equations, lattices) provides a rigorous mathematical foundation for understanding and optimizing velocity encoding.
  • Encoding Design Potential: The paper highlights that the parallelepiped concept can guide the design of new PC-MRI sequences to further expand the measurable velocity range, suggesting future avenues for acquisition optimization.

Key Results

  • Maximized Unambiguous Range: The proposed joint processing method maintains the fullest extent of the unambiguous velocity range intrinsically provided by the encoding matrix, which is often reduced by prior pre-processing techniques.
  • Maximized VNR: Joint processing also maximizes the velocity-to-noise ratio (VNR).
  • Comparison with Existing Methods: For balanced 4-point encoding, the paper shows that using all phase differences results in a significantly larger unambiguous velocity range compared to pre-processing methods like P91 and P10. For example, P91 leads to a 4 times smaller parallelepiped volume.
  • Perturbed Encoding Benefits: Demonstrations with "perturbed" 4-point and 5-point encodings show that modifying the encoding matrix can lead to 4x and 2x larger unambiguous velocity ranges, respectively, when analyzed using the proposed joint processing method.

Limitations and Future Work

  • Computational Complexity: While joint processing maximizes the range and VNR, it may involve solving a larger system of equations, potentially increasing computational demands. The paper mentions searching for wrapping integers, which can be computationally intensive.
  • Optimized Acquisition Design: The authors acknowledge that an "optimized design" would involve combining the benefits of the parallelepiped range, noise sensitivity, and probability of wrapping errors. This optimized design is stated to be beyond the scope of the current manuscript, suggesting it as a direction for future research.
  • Practical Implementation: While the theoretical framework is strong, the practical implementation and validation of these concepts on real patient data would be a crucial next step.

Summary updated

Abstract

In phase-contrast magnetic resonance imaging (PC-MRI), the velocity of spins at a voxel is encoded in the image phase. The strength of the velocity encoding gradient offers a trade-off between the velocity-to-noise ratio (VNR) and the extent of phase aliasing. Phase differences provide invariance to an unknown background phase. Existing literature proposes processing a reduced set of phase difference equations, simplifying the phase unwrapping problem at the expense of VNR or unaliased range of velocities, or both. Here, we demonstrate that the fullest unambiguous range of velocities is a parallelepiped, which can be accessed by jointly processing all phase differences. The joint processing also maximizes the velocity-to-noise ratio. The simple understanding of the unambiguous parallelepiped provides the potential for analyzing new multi-point acquisitions for an enhanced range of unaliased velocities; two examples are given.
BibTeX
@inproceedings{zhao2022maximizing,
  author = {Shen Zhao and Rizwan Ahmad and Lee C. Potter},
  title = {{Maximizing Unambiguous Velocity Range in Phase-Contrast MRI with Multipoint Encoding}},
  booktitle = {ISBI},
  year = {2022},
  doi = {10.1109/isbi52829.2022.9761589},
}
Figure