With the performance of central processing units (CPUs) having effectively reached a limit, parallel processing offers an alternative for applications with high computational demands. Modern graphics processing units (GPUs) are massively parallel processors that can execute simultaneously thousands of light-weight processes. In this study, we propose and implement a parallel GPU-based design of a popular method that is used for the analysis of brain magnetic resonance imaging (MRI). More specifically, we are concerned with a model-based approach for extracting tissue structural information from diffusion-weighted (DW) MRI data. DW-MRI offers, through tractography approaches, the only way to study brain structural connectivity, non-invasively and in-vivo. We parallelise the Bayesian inference framework for the ball & stick model, as it is implemented in the tractography toolbox of the popular FSL software package (University of Oxford). For our implementation, we utilise the Compute Unified Device Architecture (CUDA) programming model. We show that the parameter estimation, performed through Markov Chain Monte Carlo (MCMC), is accelerated by at least two orders of magnitude, when comparing a single GPU with the respective sequential single-core CPU version. We also illustrate similar speed-up factors (up to 120x) when comparing a multi-GPU with a multi-CPU implementation.

No abstract

With the performance of central processing units (CPUs) having effectively reached a limit, parallel processing offers an alternative for applications with high computational demands. Modern graphics processing units (GPUs) are massively parallel processors that can execute simultaneously thousands of light-weight processes. In this study, we propose and implement a parallel GPU-based design of a popular method that is used for the analysis of brain magnetic resonance imaging (MRI). More specifically, we are concerned with a model-based approach for extracting tissue structural information from diffusion-weighted (DW) MRI data. DW-MRI offers, through tractography approaches, the only way to study brain structural connectivity, non-invasively and in-vivo. We parallelise the Bayesian inference framework for the ball & stick model, as it is implemented in the tractography toolbox of the popular FSL software package (University of Oxford). For our implementation, we utilise the Compute Unified Device Architecture (CUDA) programming model. We show that the parameter estimation, performed through Markov Chain Monte Carlo (MCMC), is accelerated by at least two orders of magnitude, when comparing a single GPU with the respective sequential single-core CPU version. We also illustrate similar speed-up factors (up to 120x) when comparing a multi-GPU with a multi-CPU implementation.

To better understand the physiological mechanisms responsible for the differential motor cortex functioning in aging, we used transcranial magnetic stimulation to investigate interhemispheric interactions and cortical representation of hand muscles in the early phase of physiological aging, correlating these data with participants’ motor abilities. Right-handed healthy subjects were divided into a younger group (n = 15, mean age 25.4 ± 1.9 years old) and an older group (n = 16, mean age 61.1 ± 5.1 years old). Activity of the bilateral abductor pollicis brevis (APB) and abductor digiti minimi (ADM) was recorded. Ipsilateral silent period (ISP) was measured in both APBs. Cortical maps of APB and ADM were measured bilaterally. Mirror movements (MM) were recorded during thumb abductions. Motor abilities were tested using Nine Hole Peg Test, finger tapping, and grip strength. ISP was reduced in the older group on both sides, in terms of duration (p = 0.025), onset (p = 0.029), and area (p = 0.008). Resting motor threshold did not differ between groups. APB and ADM maps were symmetrical in the younger group, but were reduced on the right compared to the left hemisphere in the older group (p = 0.008). The APB map of the right hemisphere was reduced in the older group compared to the younger (p = 0.021). Older subjects showed higher frequency of MM and worse motor abilities (p < 0.001). The reduction of right ISP area correlated significantly with the worsening of motor performances. Our results showed decreased interhemispheric interactions in the early processes of physiological aging and decreased cortical muscles representation over the non-dominant hemisphere. The decreased ISP and increased frequency of MM suggest a reduction of transcallosal inhibition. These data demonstrate that early processes of normal aging are marked by a dissociation of motor cortices, characterized, at least, by a decline of the non-dominant hemisphere, reinforcing the hypothesis of the right hemi-aging model.