Acta Optica Sinica, Volume. 44, Issue 16, 1617001(2024)

Reconstruction for Fluorescence Molecular Tomography Using Fast Iterative Shrinkage Thresholding Algorithm Based on Restart Strategy

Jiachen Gao1, Sheng Zhong1, Qiong Xie2, Yating Yuan1, and Huangjian Yi1、*
Author Affiliations
  • 1School of Information Science & Technology, Northwest University, Xi’an 710127, Shaanxi , China
  • 2The First Hospital of Hunan University of Chinese Medicine, Changsha 410007, Hunan , China
  • show less
    Figures & Tables(7)
    Digital mouse model
    Reconstruction time of two algorithms under different numbers of excitation sources
    Reconstruction results of two algorithms under different noise levels. (a)-(e) Cross-sectional views of reconstructed results by FISTA at Z=16.0 mm; (f)-(j) cross-sectional views of reconstructed results by R-FISTA at Z=16.0 mm
    Error analysis of FISTA and R-FISTA under different noise levels. (a) Comparison of reconstruction LE; (b) comparison of reconstruction time
    Comparison of reconstruction results. (a) Reconstruction result of R-FISTA; (b) reconstruction result of FISTA; (c) comparison of reconstruction time
    • Table 1. Comparison of the basic steps between R-FISTA and FISTA

      View table

      Table 1. Comparison of the basic steps between R-FISTA and FISTA

      StepAlgorithm 1: FISTAAlgorithm 2: R-FISTA
      Step 0

      Input: L=Lf

      Initialization x0, let y1=x0Rn, t1=1

      Initialization x0, let L0>0, θk>0, y1=x0

      Step k

      (k≥1)

      Let Lk=a×Lk, until the proximal gradient condition is satisfied:

      FpLkykQLkpLkyk,yk:

      xk=pLyk,

      Calculate and judge according to the proximal gradient condition:

      ① If FpLkyk>QLkpLkyk, yk,

      let Lk=1+θk×Lk and calculate:

      xk=pLkyk,

      tk+1=1+1+4tk22,Lk+1Lk11+1+4Lk+1Lktk22,Lk+1Lk<1,

      Step k

      (k≥1)

      tk+1=1+1+4tk22,

      yk+1=xk+tk-1tk+1xk-xk-1,

      k=k+1

      yk+1=xk+tk-1tk+1xk-xk-1.

      Update θk:

      θk=θk-α'×dk,α'×dk<θk0.5,else.

      ② If FpLkykQLkpLkyk,yk:

      if Lk=Lk-1, reboot, Lk=L0;

      else Lk=Lk-1, k=k+1

    • Table 2. Simulation experiment results

      View table

      Table 2. Simulation experiment results

      Number of excitation sourcesAlgorithmNRMSELE /mmTime /s
      4R-FISTA0.01200.8401.70±0.04
      FISTA0.0150±0.0031.000±0.133.78±1.10
      8R-FISTA0.0165±0.00050.538±0.0194.33±0.21
      FISTA0.0200±0.0030.720±0.135.46±1.37
      16R-FISTA0.0170±0.00020.520±0.0069.75±0.33
      FISTA0.0160±0.00040.730±0.1316.56±3.79
      36R-FISTA0.0180±0.00020.480±0.0132.71±2.05
      FISTA0.0230±0.0030.910±0.2442.71±14.99
    Tools

    Get Citation

    Copy Citation Text

    Jiachen Gao, Sheng Zhong, Qiong Xie, Yating Yuan, Huangjian Yi. Reconstruction for Fluorescence Molecular Tomography Using Fast Iterative Shrinkage Thresholding Algorithm Based on Restart Strategy[J]. Acta Optica Sinica, 2024, 44(16): 1617001

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Medical optics and biotechnology

    Received: Feb. 18, 2024

    Accepted: Apr. 24, 2024

    Published Online: Aug. 5, 2024

    The Author Email: Yi Huangjian (yhj2014@nwu.edu.cn)

    DOI:10.3788/AOS240640

    CSTR:32393.14.AOS240640

    Topics