Segmentation of endothelial cells of the cornea from the distance map of confocal microscope images

We propose a novel algorithm for segmenting cells of the cornea endothelium layer on confocal microscope images. To get an inter-cellular space with minimum gray-scale value and to enhance cell borders, we apply a difference of Gaussian filter before image binarization by thresholding with the minimum gray-scale value. Removal of segmented noise and artifacts is performed by automatic thresholding (using an image frequency analysis to obtain a global threshold value per image). Final segmentation of cells is achieved by fitting the largest inscribed circles into the centers of cell regions defined by the distance map of the binary images. Parameters of interest such as cell count and density, pleomorphism, polymegathism, and F-measure are computed on a publicly available data-set (Confocal Corneal Endothelial Microscopy Data Set - Rotterdam Ophthalmic Data Repository) and compared against the results of the segmentation methods included with the data set, and the results of state of the art automatic methods. The obtained results achieve higher accuracy compared to the results of the segmentation included with the data set (e.g., -proposed versus dataset in R2 and mean relative error-, cell count: 0.823, − 0.241 versus 0.017, 0.534; cell density: 0.933, − 0.067 versus 0.154, 0.639; cell polymegathism: 0.652, − 0.079 versus 0.075, 0.886; cell pleomorphism: 0.242, − 0.128 versus 0.0352, − 0.222, respectively), and are in good agreement with the results of the state of the art method. [Display omitted] ●A difference of Gaussian is applied to enhance cell borders and to obtain the inter-cellular space with minimal value, which we then used as a threshold for obtaining a binary image.●A frequency-domain analysis allows obtaining a global adaptive threshold value per image for automatically removing artifacts in the binary images.●Local maxima, selected in the distance map of the binary image, are used as centers of circles inscribed in the cells. A selection of the inscribed circles is intended to detect individual cells, aiming also to separate those cells that appear merged in the image.●The segmentation of the cell's contours is achieved simply by labeling each background pixel according to its distance to the nearest foreground object.●Quantification and comparison with ground truth and two methods are achieved for measuring clinical parameters of interest and assessing the results, including cell count, cell density, polymegathism, and pleomorphism. Result