TY - GEN
T1 - A new, fast, relaxation-free, convergent, hessian-based, ordered-subsets algorithm for emission tomography
AU - Hsiao, Ing Tsung
AU - Rangarajan, Anand
AU - Khurd, Parmeshwar
AU - Gindi, Gene
PY - 2004
Y1 - 2004
N2 - We propose a fast, convergent, positivity preserving, OS-type (ordered-subsets) maximum likelihood (ML) reconstruction algorithm for emission tomography (ET) which takes into account the Hessian information in the ML Poisson objective. In contrast to recent approaches, our proposed algorithm is fundamentally not based on the well known EM-ML algorithm for ET . Our new algorithm is based on an expansion of the ML objective using a second order Taylor series approximation w.r.t. the projection of the source distribution similar to the approach in [1]. Defining the projection of the source as an independent variable, we construct a new objective function in terms of the source distribution and the projection. This new objective function contains the Hessian information of the original Poisson negative log-likelihood. After using a separable surrogates transformation of the new Hessian-based objective, we derive an ordered subsets, positivity preserving algorithm which is guaranteed to asymptotically reach the maximum of the original ET log-likelihood. Preliminary results show that this new algorithm is about as fast as RAMLA [2] after a few initial iterations. However, in contrast to RAMLA, the new algorithm does not require any user-specified, relaxation parameters.
AB - We propose a fast, convergent, positivity preserving, OS-type (ordered-subsets) maximum likelihood (ML) reconstruction algorithm for emission tomography (ET) which takes into account the Hessian information in the ML Poisson objective. In contrast to recent approaches, our proposed algorithm is fundamentally not based on the well known EM-ML algorithm for ET . Our new algorithm is based on an expansion of the ML objective using a second order Taylor series approximation w.r.t. the projection of the source distribution similar to the approach in [1]. Defining the projection of the source as an independent variable, we construct a new objective function in terms of the source distribution and the projection. This new objective function contains the Hessian information of the original Poisson negative log-likelihood. After using a separable surrogates transformation of the new Hessian-based objective, we derive an ordered subsets, positivity preserving algorithm which is guaranteed to asymptotically reach the maximum of the original ET log-likelihood. Preliminary results show that this new algorithm is about as fast as RAMLA [2] after a few initial iterations. However, in contrast to RAMLA, the new algorithm does not require any user-specified, relaxation parameters.
UR - http://www.scopus.com/inward/record.url?scp=17144424661&partnerID=8YFLogxK
M3 - 会议稿件
AN - SCOPUS:17144424661
SN - 0780383885
T3 - 2004 2nd IEEE International Symposium on Biomedical Imaging: Macro to Nano
SP - 1408
EP - 1411
BT - 2004 2nd IEEE International Symposium on Biomedical Imaging
T2 - 2004 2nd IEEE International Symposium on Biomedical Imaging: Macro to Nano
Y2 - 15 April 2004 through 18 April 2004
ER -