TL;DR
This paper investigates the feasibility of using microfocus X-ray CT (µCT), a more accessible imaging technique, for analyzing cardiac fiber structures compared to refraction-contrast X-ray CT (RCT), which offers superior contrast but requires a synchrotron. The study demonstrates that µCT can provide comparable fiber orientation estimates to RCT, paving the way for its broader application in cardiac research, though limitations related to stitching artifacts need further attention.
What Problem Does This Paper Address?
Understanding the intricate three-dimensional (3D) structure of cardiac muscle fibers, particularly in the left ventricle (LV), is crucial for comprehending cardiac anatomy and diseases like heart failure. While advanced imaging techniques are needed for this, existing methods have drawbacks. Diffusion tensor magnetic resonance imaging (DT-MRI) offers lower resolution. Histopathological methods, while high-resolution, are complex to reconstruct in 3D and involve tissue damage. Refraction-contrast X-ray CT (RCT) provides excellent soft tissue contrast for visualizing these fibers but is not commercially available and requires a synchrotron facility. Microfocus X-ray CT (µCT) is a commercially available alternative, but its contrast for soft tissues is generally lower than RCT. This paper aims to bridge this gap by evaluating the quantitative and objective usefulness of µCT for cardiac fiber analysis by comparing it directly with RCT.
Methods
The researchers employed a two-pronged approach for fiber analysis:
- Fiber Orientation Estimation: They used Structure Tensor (ST) analysis, a common technique for estimating local fiber orientation. This involves analyzing the diffusion of water molecules (in DT-MRI) or local intensity gradients (in CT) within a voxel to determine the direction of least resistance or change. For CT images, this means finding the direction of minimal intensity change around a voxel.
- Fiber Tracking: This process uses the estimated fiber orientations to trace the path of individual fibers through the 3D volume, creating trajectories that visualize the overall fiber architecture. This is achieved through an iterative process where the tracking algorithm moves along the estimated fiber direction from an initial point.
The core of their methodology involved:
- Scanning: A rabbit heart was scanned using both RCT and µCT with a custom protocol.
- Registration: The µCT and RCT volumes were manually cropped and rotated, then non-rigidly registered to align them precisely, compensating for deformation and differences in specimen positioning.
- Comparison: They quantitatively compared fiber orientation estimations from both modalities by calculating the angle difference between µCT and RCT estimates. They also analyzed the "inclination angle," which describes the fiber's orientation relative to the axial plane, and correlated this with the "outside ratio" (a measure of a point's proximity to the epicardium vs. endocardium).
- 3D Visualization: Fiber tracking was performed on both registered volumes to visually compare the resulting fiber trajectories.
Datasets
The study utilized a single rabbit heart specimen. This specimen was scanned using two different modalities:
- Refraction-Contrast X-ray CT (RCT): Performed at the High Energy Accelerator Research Organization (KEK) in Japan, using a synchrotron system.
- Microfocus X-ray CT (µCT): Performed using a desktop scanner (inspeXio SMX-90CT Plus) at Nagoya University.
The paper does not introduce new benchmark datasets but rather uses this single, well-prepared specimen for a direct, side-by-side comparison of the two imaging techniques.
Key Insights and Ideas
- µCT as a Viable Alternative: The most significant insight is that µCT, despite its lower contrast compared to RCT, can yield fiber orientation estimates that are quantitatively comparable to those obtained from the gold-standard RCT. This is a crucial finding for making advanced cardiac fiber analysis more accessible.
- Anatomical Consistency: Both RCT and µCT results align with established anatomical knowledge that cardiac fiber orientations differ significantly between the inner (endocardial) and outer (epicardial) layers of the LV. This provides strong validation for the fiber analysis methods themselves.
- Stitching Artifacts are a Hurdle: A key limitation identified is that the µCT scanning process, which involved stitching multiple scans to cover the entire heart, introduced artifacts. These artifacts led to "flat" and "densely gathered" tracking results, particularly around the boundaries of the stitched regions, which were not observed in the RCT data.
- Contrast Matters for Clarity: While µCT can capture fiber information, the paper visually confirms that fibers appear clearer and more distinct in the RCT volumes, highlighting the advantage of RCT's higher soft tissue contrast.
Key Results
- Fiber Orientation Correlation: The inclination angles measured from µCT and RCT volumes showed a correlation coefficient (CC) of 0.63, indicating a strong resemblance between the two.
- Average Angle Difference: The mean angle difference between µCT and RCT fiber orientations across axial slices was around 20 degrees.
- Anatomical Correlation: The inclination angles from both RCT and µCT showed a significant correlation with the "outside ratio," confirming the expected anatomical variation of fiber orientation from the endocardium to the epicardium.
- 3D Tracking Differences: While fiber tracking generally followed anatomical patterns in both modalities, the µCT results exhibited noticeable artifacts due to the stitching process, leading to less smooth and sometimes distorted fiber trajectories in specific areas.
Limitations and Future Work
The authors acknowledge several limitations:
- Single Specimen: The study was conducted on only one rabbit heart specimen. This limits the generalizability of the findings, and the authors emphasize the need for more extensive validation with a larger number of hearts.
- Parameter Sensitivity: The quantitative validation of fiber orientation results relies on parameters that were manually set. Further investigation into the impact of different parameter sets is needed.
- Artifacts in µCT: The stitching artifacts in the µCT volume led to incorrect tracking results. The authors suggest that smoothing around these boundaries is a necessary future step to improve the accuracy of µCT-based fiber tracking.
- Broader Application: While the study focused on the LV, future work could explore observing fiber structures in other parts of the heart and other tissues.
The paper concludes that µCT is a promising modality for cardiac fiber analysis, but further investigation into the differences in results between µCT and RCT, particularly concerning artifact mitigation and quantitative validation, is required.
Summary updated