• Chinese Optics Letters
  • Vol. 18, Issue 10, 101701 (2020)
Yiwei Chen1, Yi He1, Jing Wang1、2, Wanyue Li1、2, Lina Xing1, Feng Gao1, and Guohua Shi1、2、3、*
Author Affiliations
  • 1Jiangsu Key Laboratory of Medical Optics, Suzhou Institute of Biomedical Engineering and Technology, Chinese Academy of Sciences, Suzhou 215163, China
  • 2Department of Biomedical Engineering, University of Science and Technology of China, Hefei 230041, China
  • 3Center for Excellence in Brain Science and Intelligence Technology, Chinese Academy of Sciences, Shanghai 200031, China
  • show less
    DOI: 10.3788/COL202018.101701 Cite this Article Set citation alerts
    Yiwei Chen, Yi He, Jing Wang, Wanyue Li, Lina Xing, Feng Gao, Guohua Shi. Automated superpixels-based identification and mosaicking of cone photoreceptor cells for adaptive optics scanning laser ophthalmoscope[J]. Chinese Optics Letters, 2020, 18(10): 101701 Copy Citation Text show less
    Diagram depicting the image processing of the proposed algorithmic steps.
    Fig. 1. Diagram depicting the image processing of the proposed algorithmic steps.
    Example of image denoising: (a) before denoising and (b) after denoising.
    Fig. 2. Example of image denoising: (a) before denoising and (b) after denoising.
    Cone photoreceptor cell number estimation: (a) denoised image, (b) power of discrete Fourier transform (DFT) of a log10 compressed, (c) averaged slice of (b), fitted curve in red, and (d) subtraction outcome of fitted curve (highlighted in red) from the blue curve in (c).
    Fig. 3. Cone photoreceptor cell number estimation: (a) denoised image, (b) power of discrete Fourier transform (DFT) of a log10 compressed, (c) averaged slice of (b), fitted curve in red, and (d) subtraction outcome of fitted curve (highlighted in red) from the blue curve in (c).
    Simple linear iterative clustering (SLIC) superpixels segmentation: (a) original image patch and (b) segmented image with oversegmentation.
    Fig. 4. Simple linear iterative clustering (SLIC) superpixels segmentation: (a) original image patch and (b) segmented image with oversegmentation.
    Superpixels merging process.
    Fig. 5. Superpixels merging process.
    Example of superpixels merging outcome: (a) before merging and (b) after merging.
    Fig. 6. Example of superpixels merging outcome: (a) before merging and (b) after merging.
    Performance of the proposed method: (a) input AO-SLO image, (b) identification of cells and segmented image, and (c) mosaic image.
    Fig. 7. Performance of the proposed method: (a) input AO-SLO image, (b) identification of cells and segmented image, and (c) mosaic image.
    PrecisionPercentage of actual cells in identified cells77.3%
    RecallPercentage of actual cells identified95.2%
    F1-score2×Precision×Recall/(Precision+Recall)85.3%
    Table 1. Evaluation of the Effectiveness of Cell Identification: Overall Precision, Recall, and F1-Score Outcomes
    Yiwei Chen, Yi He, Jing Wang, Wanyue Li, Lina Xing, Feng Gao, Guohua Shi. Automated superpixels-based identification and mosaicking of cone photoreceptor cells for adaptive optics scanning laser ophthalmoscope[J]. Chinese Optics Letters, 2020, 18(10): 101701
    Download Citation