TL;DR
This paper presents a novel method for reconstructing and visualizing the complex 3D fiber and sheet structure of the human heart using Diffusion Tensor MRI (DTMRI). By developing advanced fiber and sheet tracking algorithms and employing high-quality visualization techniques, the authors provide unprecedented insight into the heart's intricate architecture, crucial for understanding its mechanical and electrical properties.
What Problem Does This Paper Address?
The human heart's mechanical function is intimately tied to the precise arrangement of its muscle fibers into sheets. Understanding this complex, helical 3D structure is vital for accurate cardiac modeling, predicting the effects of conditions like myocardial infarction, and developing effective treatments. However, visualizing this intricate architecture from medical imaging data has been a significant challenge. Previous methods often struggled with noise, the complex interplay of fiber and sheet orientations, and lacked the resolution to fully capture the nuanced structure, especially in regions like the apex.
Methods
The authors tackle this challenge by employing a two-pronged approach:
- Regularized Fiber and Sheet Tracking: They utilize Diffusion Tensor MRI (DTMRI) data, which provides information about water diffusion direction within tissues.
- Anisotropic Least Squares Filtering (MLS): To combat noise and oscillations inherent in DTMRI data, they apply a Moving Least Squares (MLS) method. This filter regularizes the diffusion tensor field, taking into account the local anisotropy, which is crucial for accurate tracking.
- Fiber Tracking: A novel algorithm is developed to track the principal diffusion direction (the first eigenvector of the diffusion tensor), representing the fiber orientation. This involves an integration step using a fifth-order Dormand-Prince method with automatic error estimation to ensure accuracy. A "sense of progression" check is also implemented to maintain consistent tracking direction.
- Sheet Tracking: Reconstructing the sheet structure is more complex, requiring the identification of planes. The authors develop an algorithm that tracks tangent planes to the sheets. This involves projecting a desired radial direction onto the local plane defined by fiber and sheet directions and then using a Newton method for reconstruction. They also employ a rotation to define a second direction for surface expansion.
- High-Quality Visualization: To make the complex 3D structures comprehensible, they use the open-source ray-tracing software POV-Ray. Fibers are rendered as shaded 3D tubes, and sheet structures are represented as triangulated surfaces, enhancing depth perception.
Datasets
The study utilized DTMRI data from a normal excised human heart, acquired at Johns Hopkins Medical Center. This data was made available for downloading on the internet. The dataset was arranged in a 256 × 256 × 134 array with a voxel size of 429.7 µm × 429.7 µm × 1000 µm.
Key Insights and Ideas
- Fiber-Sheet Interplay: The research visually confirms that heart fibers are organized into sheets, with the sheets having concave or convex transmural structures that align with histological findings. This provides a clearer understanding of how these two structures work together.
- Apex Curvature: The study highlights that the sheets exhibit more pronounced curvature in the apex region of the heart, a detail crucial for understanding regional mechanics.
- Anisotropy Filtering is Key: The paper emphasizes the necessity of anisotropic filtering (like MLS) for accurate fiber tracking in cardiac DTMRI, as simple isotropic filters can destroy important directional information.
- DTMRI for Cardiac Architecture: This work demonstrates the power of DTMRI, when combined with sophisticated processing and visualization, to reveal the intricate, often overlooked, architectural details of the heart.
Key Results
- Detailed 3D Reconstruction: The authors successfully generated high-resolution 3D visualizations of both fiber and sheet structures within the human heart.
- Validation of Histological Findings: The visualized fiber orientations and sheet structures showed excellent correlation with existing histological data, validating their approach.
- Smooth Variation of Fiber Angles: The visualizations confirmed the smooth variation of fiber angles across the myocardial wall, from the epicardium to the endocardium.
- Complex Sheet Geometry: The research revealed the complex, spatially varying structure of the sheet surfaces, particularly noting increased curvature in the apex.
- Color-Coded Visualization: The use of color coding for fiber angle and z-component provided an intuitive way to interpret the complex fiber orientations.
Limitations and Future Work
- Noise Sensitivity: The authors acknowledge that fiber tracking can be sensitive to noise, especially in determining the sheet structure, which relies on the less robust second and third eigenvectors. While MLS filtering helps, it can introduce bias.
- Visualization Speed: The use of POV-Ray for visualization, while providing high quality, results in slow rendering times, limiting interactive exploration.
- Generalization of Fiber Angle: The paper suggests that the commonly cited ±60° range for fiber angles is a rough approximation, and the actual myocardial fiber structure is more complex.
- Future Directions:
- Developing interactive visualization tools.
- Studying fiber and sheet structures in animal models to understand remodeling after infarction.
- Investigating other regions of the heart, such as the right ventricle.
- Validating the hypothesis of a single continuous myocardial band.
- Comparing detected sheet planes with major principal strains of cardiac deformation.
Summary updated