Radiology and radiology research are primarily concerned with extracting information from images. Image registration could be a key technology in this process, as it enables automated comparisons between images (e.g. for temporal monitoring of tissue change), atlas generation (by averaging images from multiple subjects) and image fusion (e.g. showing information from multiple modalities in a single image). Image registration is computationally expensive. Lag periods between input and output currently make it impractical for interactive applications and limit its general use. Grid parallel computing promises to alleviate this problem. This exhibit demonstrates the Registration Workbench, a web-based interface that lets users submit image registration tasks to a Grid cluster, keeps track of submitted tasks and stores intermediate and final results, while hiding the complexity of Grid processing, secure data transfers, job management, etc. from the user. Visitors are invited to experiment with the Workbench by submitting registration tasks using sample two- and three-dimensional image datasets for remote execution on a Grid-enabled server.
1. Understand the principles and benefits of submitting image registration tasks to a Grid cluster, 2. Evaluate a web-based workbench interface for task submission and results storage, 3. Appreciate the potential of interactive image registration for radiology research and clinical practice. (R.H., Grant, The Dunhill Medical Trust; D.R., Grant, Engineering and Physical Sciences Research Council.)
Questions about this event email: email@example.com