Modeling of 3D Fibers Structure of Human Heart Using DTI Images and Moving Weighted Effective Sphere (MWES) Filter

Understanding the complex fiber architecture of the heart is critical for modeling its electrical and mechanical behavior. Traditional methods such as dissection are invasive, time-consuming, and prone to error. Diffusion Tensor Imaging (DTI) provides a non-invasive alternative to map the anisotropic fiber structure of the myocardium. However, DTI data often suffer from noise and misaligned vectors, particularly at boundaries between excitable and non-excitable tissues. This work proposes a novel filter—Weighted Effective Sphere (WES)—that effectively corrects and smooths misaligned vectors without significantly altering neighboring fiber directions. A refined version, Moving Weighted Effective Sphere (MWES), is also introduced for enhanced spatial reconstruction and visualization of heart fibers. The proposed algorithms outperform traditional filters in preserving anatomical fidelity, reducing noise, and enabling realistic rescaling of DTI datasets, thereby improving the accuracy of cardiac simulations.