作者: Robert Cierniak
DOI: 10.1007/978-3-642-13208-7_63
关键词:
摘要: This paper presents a novel neural network approach to the problem of image reconstruction from projections obtained by spiral tomography scanner. The process is performed during minimizing energy function in recurrent network. Our method great practical use discrete cone-beam projections. Experimental results show that appropriately designed able reconstruct an with better quality than used commercial conventional algorithms.