Talks and Conference Presentations
Organization for Human Brain Mapping
10th Annual Meeting
June 13-17, 2004
Automated Processing Pipeline for Anatomical and Functional MRI data
James Dias and James Voyvodic
Functional MRI scanning generates large datasets (typically hundreds of Mbytes) that need to be efficiently and reliably transferred from the scanner to data analysis computers and archival media. Processing of these data sets also typically involves some routine image processing procedures, which can be time consuming because of the amount of data involved. In clinical cases where rapid processing of fMRI data is important, many potentially useful processing steps may often be omitted to save time. In order to make these routine data manipulation steps more efficient and more reliable, therefore, we have implemented a fully automated data processing pipeline that detects MRI data as soon as it is acquired on the scanner and takes care of the initial critical aspects of the analysis. We have also implemented a batch processing scheme to facilitate routine user-initiated post-processing steps.
An XML-based Data Access Interface for Image Analysis and Visualization Software
Syam Gadde, Charles R. Michelich, James Voyvodic
Constructing a robust image analysis pipeline from several, often disparate, software elements is not a trivial process. Many valuable software tools have arbitrary data format requirements. Combining such software tools in analysis requires data conversion steps that may slow down processing and introduce errors.
We have defined a general data access interface in the form of a structured text file, written in XML (Extensible Markup Language). This XML file provides instructions on how to extract image data from files in any uncompressed data format, without explicit conversion of data. This interface supports data of arbitrary dimensionality, data type, and byte order, and is applicable to most MR image formats, physiological and stimulus data.