Using GPUs to accelerate computational diffusion MRI: From microstructure estimation to tractography and connectomes

The great potential of computational diffusion MRI (dMRI) relies on indirect inference of tissue microstructure and brain connections, since modelling and tractography frameworks map diffusion measurements to neuroanatomical features. This mapping however can be computationally highly expensive, particularly given the trend of increasing dataset sizes and the complexity in biophysical modelling. Limitations on computing resources can restrict data exploration and methodology development. A step forward is to take advantage of the computational power offered by recent parallel computing architectures, especially Graphics Processing Units (GPUs). GPUs are massive parallel processors that offer trillions of floating point operations per second, and have made possible the solution of computationally-intensive scientific problems that were intractable before. However, they are not inherently suited for all problems. Here, we present two different frameworks for accelerating dMRI computations using GPUs that cover the most typical dMRI applications: a framework for performing biophysical modelling and microstructure estimation, and a second framework for performing tractography and long-range connectivity estimation. The former provides a front-end and automatically generates a GPU executable file from a user-specified biophysical model, allowing accelerated non-linear model fitting in both deterministic and stochastic ways (Bayesian inference). The latter performs probabilistic tractography, can generate whole-brain connectomes and supports new functionality for imposing anatomical constraints, such as inherent consideration of surface meshes (GIFTI files) along with volumetric images. We validate the frameworks against well-established CPU-based implementations and we show that despite the very different challenges for parallelising these problems, a single GPU achieves better performance than 200 CPU cores thanks to our parallel designs.

Region-Specific Microstructure in the Neonatal Ventricles of a Porcine Model

The neonate transitions from placenta-derived oxygen, to supply from the pulmonary system, moments after birth. This requires a series of structural developments to divert more blood through the right heart and onto the lungs, with the tissue quickly remodelling to the changing ventricular workload. In some cases, however, the heart structure does not fully develop causing poor circulation and inefficient oxygenation, which is associated with an increase in mortality and morbidity. This study focuses on developing an enhanced knowledge of the 1-day old heart, quantifying the region-specific microstructural parameters of the tissue. This will enable more accurate mathematical and computational simulations of the young heart. Hearts were dissected from 12, 1-day-old deceased Yorkshire piglets (mass: 2.1–2.4 kg, length: 0.38–0.51 m), acquired from a breeding farm. Evans blue dye was used to label the heart equator and to demarcate the left and right ventricle free walls. Two hearts were used for three-dimensional diffusion-tensor magnetic resonance imaging, to quantify the fractional anisotropy (FA). The remaining hearts were used for two-photon excited fluorescence and second-harmonic generation microscopy, to quantify the cardiomyocyte and collagen fibril structures within the anterior and posterior aspects of the right and left ventricles. FA varied significantly across both ventricles, with the greatest in the equatorial region, followed by the base and apex. The FA in each right ventricular region was statistically greater than that in the left. Cardiomyocyte and collagen fibre rotation was greatest in the anterior wall of both ventricles, with less dispersion when compared to the posterior walls. In defining these key parameters, this study provides a valuable insight into the 1-day-old heart that will provide a valuable platform for further investigation the normal and abnormal heart using mathematical and computational models.

A transverse isotropic constitutive model for the aortic valve tissue incorporating rate-dependency and fibre dispersion: Application to biaxial deformation

This paper presents a continuum-based transverse isotropic model incorporating rate-dependency and fibre dispersion, applied to the planar biaxial deformation of aortic valve (AV) specimens under various stretch rates. The rate dependency of the mechanical behaviour of the AV tissue under biaxial deformation, the (pseudo-) invariants of the right Cauchy-Green deformation-rate tensor Ċ associated with fibre dispersion, and a new fibre orientation density function motivated by fibre kinematics are presented for the first time. It is shown that the model captures the experimentally observed deformation of the specimens, and characterises a shear-thinning behaviour associated with the dissipative (viscous) kinematics of the matrix and the fibres. The application of the model for predicting the deformation behaviour of the AV under physiological rates is illustrated and an example of the predicted σ-λ curves is presented. While the development of the model was principally motivated by the AV biomechanics requisites, the comprehensive theoretical approach employed in the study renders the model suitable for application to other fibrous soft tissues that possess similar rate-dependent and structural attributes.