On asymptotic convergence of the block-iterative Fisher scoring algorithm

Malcolm Hudson*, Jun Ma

*Corresponding author for this work

    Research output: Contribution to journalArticlepeer-review


    This short paper studies convergence properties, particularly asymptotic convergence, of the block-iterative Fisher scoring (BFS) algorithms recently proposed by Ma and Hudson (2008). While applicable in other inverse problem domains (e.g. astronomy, geophysics, signal processing or remote sensing), this class of algorithms was designed for tomographic image reconstruction from projections in medicine. A BFS algorithm is used to reconstruct the patient's internal structural or functional activity from collected projection data. We briefly introduce the BFS algorithm and a general convergence result provided in Ma and Hudson (2008). This result is used to prove the asymptotic convergence of two specific BFS algorithms under new conditions.

    Original languageEnglish
    Pages (from-to)922-925
    Number of pages4
    JournalStatistics and Probability Letters
    Issue number11-12
    Publication statusPublished - Jun 2010


    Dive into the research topics of 'On asymptotic convergence of the block-iterative Fisher scoring algorithm'. Together they form a unique fingerprint.

    Cite this