Japanese
TitleECTにおける種々の逐次型画像再構成法の比較
Subtitle原著
Authors村山秀雄*, 田中栄一*, 野原功全*, 富谷武浩*, 山本幹男*
Authors(kana)
Organization*放射線医学総合研究所物理研究部
Journal核医学
Volume24
Number6
Page797-807
Year/Month1987/6
Article原著
Publisher日本核医学会
Abstract「要旨」ECTにおける6つの逐次型画像再構成法をアルゴリズムの比較によって分析すると同時に, 計算機によるシミュレーション実験により逐次近似画像を作成してこれらの方法の特性および収束性を分析した. 期待値最大化アルゴリズム(EM)法は加算式同時逐次近似(ASIRT)法および最急降下式最小自乗逐次近似(GRADY)法にアルゴリズムが類似しており, EM法はASIRT法よりも収束性が良好であった. 像空間逐次近似アルゴリズム(ISRA)法は乗算式同時逐次近似(MSIRT)法を簡素化したアルゴリズムであるが収束速度は劣化した. 共役傾斜最小自乗逐次近似(CONGR)法は高周波成分の補正を増強する効果があるため, 上述の6つの方法の中では最も収束性が良好であった. また, 逐次近似画像の収束性を示すための4つの適合関数について相互比較を行い, 各適合関数の特色や相互の関連性を分析した.
Practice臨床医学:一般
KeywordsECT, Image reconstruction, Iterative method, EM algorithm, Computer simulation.
English
TitleA Comparison of Several Iterative Reconstruction Methods for ECT
SubtitleOriginal Articles
AuthorsHideo MURAYAMA, Eiichi TANAKA, Norimasa NOHARA, Takehiro TOMITANI, Mikio YAMAMOTO
Authors(kana)
OrganizationDivision of Physics, National Institute of Radiological Scienses
JournalThe Japanese Journal of nuclear medicine
Volume24
Number6
Page797-807
Year/Month1987/6
ArticleOriginal article
PublisherTHE JAPANESE SOCIETY OF NUCLEAR MEDICINE
Abstract[Summary]Six iterative reconstruction methods appropriate for emission computed tomography were compared in regard to their algorithms. It was shown that the Expectation Maximization algorithm (EM) method is classified in an algorithmic type similar to the Additive Simultaneous Iterative Reconstruction Technique (ASIRT) method and the Gradient (GRADY) method, and that the Image Space Reconstruction Algorithm (ISRA) method has a simplified algorithm of the Multiplicative Simultaneous Iterative Reconstruction Technique (MSIRT) method. The performances of the six methods were investigated by computer simulation generating their iterative reconstruction images of a mathematical phantom. Four fitness functions were introduced to analyze the convergence of the iterative reconstruction methods quantitatively. Each fitness function is a measure of the difference between the reconstruction image and the mathematical phantom, or a measure of the rate of covergence. The EM method achieved faster convergence than the ASIRT method, while the ISRA method gave slower convergence than the MSIRT method. Of the six methods, the fastest convergence was obtained using the Conjugate Gradient (CONGR) method, and the slowest was the ISRA method. The characteristics of each fitness function were also analyzed by its behavior as a function of iteration number for the first 100 iterations of the six methods. These results suggest that the selection of the fitness function for a measure of the convergence is important to use any iterative reconstruction methods.
PracticeClinical medicine
KeywordsECT, Image reconstruction, Iterative method, EM algorithm, Computer simulation.

【全文PDF】