C.M. Chen and S. Y. Lee, "On parallelizing the EM algorithm for PET image reconstruction," IEEE Trans. Parallel Distrib. Syst., vol. 5, pp. 86(873, Aug., 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....exploitation of its sparsity and symmetries. For every projection index i, only a narrow chord of voxels, representing just the region of support of coincidence line blurring, is nonzero. The symmetries, which include rotation and reflection in the transverse plane [25] 26] axial reflection [3], and parallel axial redundancies [22] account for a storage reduction factor of 104 for thick slice reconstructions and 451 for thin slice reconstructions. In order to test the convergence properties of the interior point methods and to compare the computational requirements of the ....

C.M. Chen and S. Y. Lee, "On parallelizing the EM algorithm for PET image reconstruction," IEEE Trans. Parallel Distrib. Syst., vol. 5, pp. 86(873, Aug., 1994.

....the figure. plexity analysis based only on multiplications and additions in (2) and (5) The three types of symmetries in the system matrix are illustrated in Figure 3. In plane symmetries [17] involve reflection and rotation about # and provide an 8 fold storage reduction. The axial symmetry [4] involves a reflection about # and yields a 2 fold symmetry (unless # = 0, in which case there is no gain) Axial parallel chord redundancies are the simplest to implement and yield the largest reduction in a 3 D system. The corresponding reduction is NR s per symmetry related group, where NR ....

C.M. Chen, S.Y. Lee, and Z.H. Cho, "On parallelizing the EM algorithm for PETimage reconstruction," IEEE Trans. Parallel Dist. Sys., vol. 5 (1994), pp. 860-873.

....to compare, contrast, and show potential for future research results. Keywords: volume visualization, splatting, ray casting, taxonomy 1 References [98] 49] 16, 48, 113] 44] 52] 107] 111, 81] 94] 16, 15] 34] 101] 40] 66] 73] 90] 116] 26] 20] 68] 6] 54, 52, 63, 28, 67] [7, 1, 42, 4, 62] [31, 75, 30, 38, 86] 41, 21, 119, 112, 118, 45, 58] 55, 65, 14, 29, 79] 115, 53, 70, 74, 87, 19, 117, 32] 72, 57, 92, 36, 7] 111, 12, 90, 97, 78, 5] 15, 27, 35, 43, 87, 28, 51, 22, 23, 59, 60, 64, 63, 66, 68, 71, 80, 81, 91, 93, 99, 47, 108] 103, 116, 83, 106, 107, 105, 2, 76, 110, 77] ....

.... taxonomy 1 References [98] 49] 16, 48, 113] 44] 52] 107] 111, 81] 94] 16, 15] 34] 101] 40] 66] 73] 90] 116] 26] 20] 68] 6] 54, 52, 63, 28, 67] 7, 1, 42, 4, 62] 31, 75, 30, 38, 86] 41, 21, 119, 112, 118, 45, 58] 55, 65, 14, 29, 79] 115, 53, 70, 74, 87, 19, 117, 32] [72, 57, 92, 36, 7] [111, 12, 90, 97, 78, 5] 15, 27, 35, 43, 87, 28, 51, 22, 23, 59, 60, 64, 63, 66, 68, 71, 80, 81, 91, 93, 99, 47, 108] 103, 116, 83, 106, 107, 105, 2, 76, 110, 77] 11, 104, 101, 96, 89, 26, 73, 6, 17, 52, 54, 85, 114, 9, 82, 37, 24] 8, 61, 3, 56, 67, 88, 44, 10, 69, 48, 16, 84, 113, 95, 13, ....

Chung-Ming Chen and Soo-Young Lee. On parallelizing the em algorithm for pet image reconstruction. IEEETPDS, 5(8):860--873, Aug. 1994.

....communication overheads and load imbalances. The efficiencies for the different data and task partitions are reported as very similar. An estimate of performance but no times was reported for a message passing Intel iPSC 2 system and on a shared memory BBN Butterfly GP 1000 multiprocessor. In [92], Chen and Lee add associated integration and broadcast algorithms and study the theoretical performance on hypercube and ring topologies for multiple slice PET. They assume that computation and communication cannot be concurrent. Extendable multiprocessors composed of DSP chips have been proposed ....

C.-M. Chen and S.-Y. Lee, "On parallelizing the EM algorithm for PET image reconstruction," IEEE Trans. Parallel Dist. Sys., vol. 5, no. 8, pp. 860--873, 1994.

....is linearly partitioned into subtasks T 1 , T 2 and T 3 which are pipelined. c) A PMM consisting of multiprocessor modules MM 1 , MM 2 and MM 3 corresponding to T 1 , T 2 and T 3 . Local results are collected or integrated before being transferred to the next stage. Linear Integration Algorithm [4], a fast integration algorithm developed for a ring, is employed in each stage. The size of data involved in each stage of the problem model is large, as in image processing, computer vision, large scale simulation, computer graphics, virtual reality, etc. Therefore, the time for each of ....

....linear array of PEs in the two successive stages such that each integrated local result goes through all PEs in the following stage. During this transfer, each PE in the following stage takes whatever portion of each integrated result it needs for processing. Linear Integration Algorithm (refer to [4] for description of the algorithm) a fast integration algorithm developed for a ring, is used for integration. In this algorithm, the integration procedure requires n i Gamma 1 transmissions of data of size m i n i and n i Gamma 1 link setups when m i is the size of data to be integrated and ....

[Article contains additional citation context not shown here]

C.-M. Chen and S.-Y. Lee, "On Parallelizing the EM Algorithm for PET Image Reconstruction", IEEE Transactions on Parallel and Distributed Systems, Vol. 5, No. 8, pp. 860-873, August 1994.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC