A dedicated storage scheme for the system matrix to the model based tomographic X-ray perfusion imaging

An appropriate mathematical framework for perfusion imaging based on slow projection acquisition like originating from C-arm systems has already been presented in literature. Due to the huge system matrix this approach in general has an extremely high computational effort and memory consumption, thus preventing the storage in the random access memory of a computer for realistic objects and spatial resolution. To overcome this difficulty we present an efficient storage scheme for the system matrix. The main idea is to effectively utilize the underlying geometric symmetries. In this paper we investigated the performance of this the beam-form independent framework focusing on the reduction of memory consumption. The obtained results demonstrate a reduction of the overall memory requirements up to three orders of magnitude compared to savings due to the simply utilization of the matrix sparsity of the matrixes. Future work will focus on the implementation of the method for large three dimensional objects as well as using higher order symmetries.