Novel approaches to Cellular Automata with applications in medical image segmentation
Funding source: Romanian National Authority for Scientific Research and Innovation CNCS-UEFISCDI
Code: PN-II-RU-TE-2014-4-1130
Contract no.: 262/01.10.2015
Computer-aided medical image analysis is used to reconstruct anatomical features from imaging modalities such as tomography (CT) scans. These are useful for diagnosis, monitoring growth of tumours, planning surgery or assessing the effects of radiotherapy. Medical image segmentation techniques are still far from being able to identify features such as tumours or even simple organs in scans.
It is in this process that we propose to make use of Cellular Automata (CA) where, through appropriate choice of evolution rules and topologies, we can identify cells which belong together. The main objective of the project is to build an innovative suite of techniques for unsupervised medical image segmentation using CA. Three separate strands in approaching this problem will constitute separate objectives: techniques using CA with hyper-connected topologies; techniques using CA with sparse topologies; techniques using CA with hierarchical topologies.
The success of the project is ensured by the synergy between the research group from Cluj-Napoca, with strong experience in cellular automata and intelligent computational optimization techniques and the research group from Oxford with strong experience in medical imaging. Besides their in depth specific knowledge, each member has a good understanding of the overall project view.