Chinese Journal of Liquid Crystals and Displays, Volume. 37, Issue 9, 1190(2022)

Red blood cell image segmentation based on NODE-UNet++ and marker watershed

Ya-qi RONG1,2, Li-juan ZHANG2, Jin-li CUI3, Wei SU4, and Meng-ye GAI1、*
Author Affiliations
  • 1School of Information Technology, Jilin Agricultural University, Changchun 130118, China
  • 2School of Computer Science and Engineering, Changchun University of Technology, Changchun 130012, China
  • 3Department of Radiology, Affiliated Hospital of Changchun University of Chinese Medicine, Changchun 130000, China
  • 4College of Medicine Information, Changchun University of Chinese Medicine,Changchun 130117, China
  • show less
    Figures & Tables(13)
    Image annotation and analysis. (a) Original image;(b) Mask image;(c) Label image.
    Data enhancement
    Flow chart of our method for red blood cell image segmentation
    Structure of residual block
    Structure of ODE block
    Architecture of NODE-UNet++
    Loss values and accuracy in the training process of NODE-UNet++
    Segmentation images of NODE-UNet++. (a) Original images; (b) Label images; (c) Pre-segmentation images.
    Segmentation images of MW algorithm. ‍(a) Foreground marker image; (b) Background marker image; (c) Final image; (d) Reconstruction map of gradient topographic; (e) Segmented RBCs.
    Red blood cell segmentation results from three different algorithms. (a)Original image;(b)Label image;(c) Results using the algorithm in [14];(d) Results using MW-UNet++ algorithm;(e) Results using our algorithm.
    Box-plots of three algorithms
    • Table 1. Confusion matrix

      View table
      View in Article

      Table 1. Confusion matrix

      SamplePrediction(Positive)Prediction(Negative)
      Actual (True)TPFN
      Actual (False)FPTN
    • Table 2. Quantitative results of three methods

      View table
      View in Article

      Table 2. Quantitative results of three methods

      ModelDSCMPAMIoU
      Algorithm in [140.957 70.961 50.942 3
      MW-UNet++ algorithm0.964 30.981 30.955 4
      Our algorithm0.968 90.989 70.963 3
    Tools

    Get Citation

    Copy Citation Text

    Ya-qi RONG, Li-juan ZHANG, Jin-li CUI, Wei SU, Meng-ye GAI. Red blood cell image segmentation based on NODE-UNet++ and marker watershed[J]. Chinese Journal of Liquid Crystals and Displays, 2022, 37(9): 1190

    Download Citation

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

    Category: Research Articles

    Received: Jan. 15, 2022

    Accepted: --

    Published Online: Sep. 14, 2022

    The Author Email: Meng-ye GAI (mengyeg@jlau.edu.cn)

    DOI:10.37188/CJLCD.2022-0009

    Topics