Parallelizing I/O-intensive image access and processing applications

2002

J. Electron. Imaging

Abstract. Although many three-dimensional (3D) medical imaging visualization methods exist, 3D volume slicing remains the most commonly used technique for visualizing medical data from modalities such as CT, MRI, and PET. We propose to extend the possibilities of oblique slicing to developable curved surfaces that can be flattened and displayed in two dimensions without deformation. Such surfaces can be used to follow curved anatomical structures while preserving distance metrics at visualization time. They may also be useful for the staging of tumors, i.e., to evaluate the spatial extension of a tumor. We propose an out of core algorithm that runs in parallel on a multi-PC architecture and is able to extract surfaces from very large 3D datasets such as the visible human data set (man: 13 GB, woman: 49 GB). Experimental performance results are presented which demonstrate that parallel surface extraction is scalable and has a reasonable overhead compared with traditional oblique planar slicing. Surface extraction is made available to the public as one of the services offered by EPFLs visible human web server (http://visiblehuman.epfl.ch).

Section: Varying the Number Of Server Nodesmentioning

confidence: 99%

Section: Figueiredo and Herschmentioning

confidence: 99%

Section: Algorithm Overviewmentioning

confidence: 99%

See 1 more Smart Citation

Parallel unfolding and visualization of curved surfaces extracted from large three-dimensional volumes

Figueiredo

2002

J. Electron. Imaging

“…1). A number of parallel file systems were designed ( [2], [3], [4], [5], [6], [7], [8]), which rely on the parallel file striping paradigm.…”

Section: Motivation/introductionmentioning

confidence: 99%

SFIO a striped file I/O library for MPI

Gabrielyan

2001 Eighteenth IEEE Symposium on Mass Storage Systems and Technologies

2001

This paper presents the design and evaluation of a Striped File I/O (SFIO) library for parallel I/O within an MPI environment. We present techniques for optimizing communications and disk accesses for small striping factors. Using MPI derived datatype capabilities, we transmit fragmented data over the network by single MPI transfers. We present the I/O performance of the SFIO library on DEC Alpha clusters, both for the Fast Ethernet and for the TNET communication networks.

“…For enabling parallel access to its data [12], the Visible Human 3D volume is segmented into sub-volumes (stored as extents) of size 54x54x19 RGB voxels, i.e. 162.3 KB, which are striped over a number of CD-ROMs.…”

Section: An Application Examplementioning

confidence: 99%

Parallel file striping on optical jukebox servers

Tárraga

Proceedings. IEEE International Conference on Multimedia and Expo

Self Cite

In the near future, large digital media servers are expected to offer storage capacities in the order of petabytes. Servers made of clusters of PC's connected to jukeboxes may represent an interesting alternative compared with servers made of arrays of magnetic disks. However, due to disk exchange overhead, higher seek times and lower data transfer rates, access to data located on optical disks is significantly slower than access to data located on magnetic disks. In the present contribution, we analyze the benefits and limitations of striping files across multiple optical disks in order to speedup the retrieval of large pieces of information. We also show that server applications requiring both computer power and I/O bandwidth may be distributed over several processing nodes and access files striped over multiple optical disks. For such applications, as long as disk exchanges are not necessary, a predictable speedup can be obtained by ensuring a high enough number of server computers, optical disk drive units, and enough communication bandwidth.