Distributed Computer Aided Detection Algorithms

Today medical computing requirements are growing due to the need to make Computer Aided Detection (CADe) available for all patients. Development of CADe software is an ongoing issue, mostly due to the high computational and storage costs of detection algorithms. The objective of this paper is to identify, study, and test the feasibility of distributed algorithms for computer aided detection, which are able to scale to increasingly larger inputs and larger datasets corresponding to more intense processing. There are two parallel approaches to the problem, the first one tries to respond to the need for greater access to this type of software and diagnostic methods, by creating the premise for massive parallel execution of these algorithms. The second method hinges on the idea of parallelizing the algorithms for detection, making them execute faster and do more intensive processing thus raising the quality of the response. The solution of the proposed subject is based on gProcess, an interactive toolset supporting the flexible description, instantiation, scheduling and execution by Grid processing. Within the framework description of a processing, the workflow is done via Process Description Graphs, which are directed graph of processing nodes.


Saturday, 7 April, 2012 - 08:00 to 09:30