Laser & Optoelectronics Progress, Volume. 61, Issue 22, 2217001(2024)

Intelligent Evaluation of Tumor Calcification Areas Based on Whole Slide Images

Zhenzhen Wan1, Haocheng Li1, Ning Shi1、*, Yuwei Liu1, and Fang Liu2、**
Author Affiliations
  • 1Key Laboratory of Digital Medical Engineering of Hebei Province, College of Electronic and Information Engineering, Hebei University, Baoding 071002, Hebei , China
  • 2Department of Pathology, Baoding Hospital of Beijing Children's Hospital, Capital Medical University, Baoding 071002, Hebei , China
  • show less
    Figures & Tables(12)
    ECR-UNet network structure
    CBAM module
    ECA module
    Flow chart of the outer contour extracted from WSI. (a) Tumor WSI ; (b) grayscale; (c) Gaussian filter ; (d) Canny edge detection; (e) closing operation and opening operation; (f) binary image of the outer contour of the WSI
    Multiple data augmentation examples. (a) Original image; (b) random horizontal flip; (c) random vertical flip; (d) random rotation; (e) center crop; (f) random color jitter; (g) random brightness contrast
    Segmentation results of different networks in calcified regions
    Segmentation results of the outer contour. (a) (c) (e) (g) Tumor WSI; (b) (d) (f) (h) outer contour
    Comparison bar chart of calcified area and pathological image contour area with gold standard. (a)Calcified area; (b) pathological image contour area
    Comparison bar chart of calcification proportion and gold standard
    • Table 1. Performance metrics for ablation experiment

      View table

      Table 1. Performance metrics for ablation experiment

      Data setDiceAccuracyRecallPrecisionJaccard
      U-Net82.2098.3978.0992.7672.71
      R18-UNet82.3598.3585.2484.1472.08
      R34-UNet85.7498.6185.1990.0177.04
      CR-UNet86.4398.6584.5891.0977.69
      ER-UNet85.9798.6886.9088.2377.13
      CER-UNet86.5698.7087.8287.8277.56
      ECR-UNet89.1398.9489.2190.8181.72
    • Table 2. Comparison of performance metrics for different segmentation networks

      View table

      Table 2. Comparison of performance metrics for different segmentation networks

      Data setDiceAccuracyRecallPrecisionJaccard
      FCN-8s76.5697.6281.2278.7564.20
      SegNet79.3297.9584.8278.8767.70
      LinkNet79.6998.2577.5888.5568.96
      PspNet81.3198.1284.1882.6670.54
      U-Net82.2098.3978.0992.7672.71
      ECR-UNet89.1398.9489.2190.8181.72
    • Table 3. Calculation accuracy of calcified region area, outer contour area, and proportion of calcified region

      View table

      Table 3. Calculation accuracy of calcified region area, outer contour area, and proportion of calcified region

      NO.Calcification areaOuter contour areaProportion
      196.7798.7498.00
      297.6797.6295.18
      398.3998.9099.49
      485.2799.2884.43
      589.5799.4590.06
      699.7499.6699.40
      785.9698.8884.66
      886.8499.0685.76
      992.6698.8491.40
      1085.9699.6285.52
      1195.9699.5595.49
    Tools

    Get Citation

    Copy Citation Text

    Zhenzhen Wan, Haocheng Li, Ning Shi, Yuwei Liu, Fang Liu. Intelligent Evaluation of Tumor Calcification Areas Based on Whole Slide Images[J]. Laser & Optoelectronics Progress, 2024, 61(22): 2217001

    Download Citation

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

    Category: Medical Optics and Biotechnology

    Received: Mar. 1, 2024

    Accepted: Mar. 25, 2024

    Published Online: Nov. 19, 2024

    The Author Email: Ning Shi (shiningzhongguo@126.com), Fang Liu (asasfang@163.com)

    DOI:10.3788/LOP240787

    CSTR:32186.14.LOP240787

    Topics