Abstract
DOAJ is a unique and extensive index of diverse open access journals from around the world, driven by a growing community, committed to ensuring quality content is freely available online for everyone.
Introduction
Phase Contrast (PC) MRI is a powerful non-invasive tool for measuring blood flow and hemodynamics. However, its clinical utility is often limited by long scan times, as it requires multiple acquisitions with different magnetic gradients to calculate velocity. This paper introduces a novel Compressed Sensing (CS) approach that significantly speeds up PC MRI by exploiting the "complex difference" (CD) between flow-encoded images as a sparsifying transform.
Methodology
The researchers proposed a CS reconstruction algorithm utilizing complex difference sparsity. In PC MRI, two images ($m_1$ and $m_2$) are typically acquired with different velocity-encoding gradients. By subtracting these images, static tissue cancels out, leaving only the moving blood. This makes the CD image inherently sparse. The reconstruction minimizes an objective function balancing data fidelity with both standard transform sparsity and the new CD sparsity term ($\|m_1 - m_2\|_1$). An iterative "handshake" process ensures that the intermediate reconstruction of one image informs the other, enhancing the recovery of signal from under-sampled data.
Experimental Setup
The study involved 14 healthy adult subjects. Axial slices of the ascending aorta were acquired using a 2D PC MRI sequence. The researchers used Gaussian random under-sampling to simulate acceleration rates of 3x and 5x, comparing the results against fully-sampled "gold standard" data and standard CS reconstructions that did not utilize the CD sparsity term.
Key Insights
By leveraging the underlying physics of phase contrast imaging rather than generic mathematical transforms, the authors achieved higher efficiency. The mutual information shared between coupled bipolar images allows the algorithm to accurately "fill in" missing data, preserving flow rate curves even at high acceleration rates where traditional methods might introduce noise or blurring.
Results and Discussion
The method demonstrated high accuracy in blood volume measurements across the cardiac cycle. Statistical analysis via Bland-Altman plots showed excellent agreement with the gold standard. Furthermore, linear regression confirmed that flow rate curves remained highly correlated with the original data at both 3x and 5x acceleration, proving the robustness of the CD sparsity term.
Conclusion and Future Work
While the retrospective results are promising, prospective studies are needed to validate the technique during live scans. Additionally, testing on larger patient cohorts with complex cardiovascular diseases, such as stenosis or aneurysms, will be essential to confirm the robustness of the CD sparsity assumption in clinical settings where flow may be more turbulent.
Summary updated