Dedicated computing environments are becoming increasingly important in neuroimaging applications. Indeed we are experiencing a fast development of non-invasive technologies that progress the research on human brain. These advanced techniques, such as magnetic resonance imaging (MRI), positron emission tomography (PET) and electroencephalography (EEG), give us the possibility to visualize and analyse brain function and structure in exceptional detail, but they have led to the necessity of storing and processing very large amounts of data. Moreover, the growth in the complexity of algorithms developed to analyse brain images, has involved an augmenting demand of high-performance resources for data storage and management, and computing systems for image processing and quantitative analysis, e.g. Graphics Processing Units (GPU). The main motivations for using GPU in neuroimaging are the time saving and the possibility to apply advanced algorithms instead of simple ones. The aim of this study is to describe the issues related to the development of a computing environment for neuroimaging applications. The dedicated farm we built, consisting at the moment of a computing node and a storage unit, has been implemented in the Pisa INFN computing centre. It has been designed to guarantee the secure data handling, storage and the access to fast cloud-based computational resources.
GPUs parallel computing exploitation for neuroimaging
FANTACCI, MARIA EVELINA;
2017-01-01
Abstract
Dedicated computing environments are becoming increasingly important in neuroimaging applications. Indeed we are experiencing a fast development of non-invasive technologies that progress the research on human brain. These advanced techniques, such as magnetic resonance imaging (MRI), positron emission tomography (PET) and electroencephalography (EEG), give us the possibility to visualize and analyse brain function and structure in exceptional detail, but they have led to the necessity of storing and processing very large amounts of data. Moreover, the growth in the complexity of algorithms developed to analyse brain images, has involved an augmenting demand of high-performance resources for data storage and management, and computing systems for image processing and quantitative analysis, e.g. Graphics Processing Units (GPU). The main motivations for using GPU in neuroimaging are the time saving and the possibility to apply advanced algorithms instead of simple ones. The aim of this study is to describe the issues related to the development of a computing environment for neuroimaging applications. The dedicated farm we built, consisting at the moment of a computing node and a storage unit, has been implemented in the Pisa INFN computing centre. It has been designed to guarantee the secure data handling, storage and the access to fast cloud-based computational resources.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.