Application of vessel enhancement filtering for automated classification of human In-Vitro fertilized (IVF) images

Visualization, classification and precise analysis of human embryo is an important prerequisite for fertility clinic. This work reports techniques to improve recognition of embryo features such as circular shape, width of edges of the detected cells, alignment of 4-cell and 8-cell structures of the blastomere, etc leading to the goal of blastomere segmentation. Paper describes a method for the detection of cell cleavage quality of human blastomere via image processing. Boundaries or ridges of blastomeres are first extracted by Hessian-based ridge detector followed by non-maxima suppression and hysteresis thresholding. Hough circle transform is applied to the acquired boundaries to detect the presence of Oocytes/embryos with due consideration of boundary orientation. Then a screening algorithm is used to verify the detected embryo cells' circles. Finally, the location of validated cells is approximated with circles and ellipses. The higher order filtering methods such as Multi_Scale structures or second order confined structure of an image (Hessian) may be used to get precise parameters for the implementation of vessel enhancement filter which in turn can identify the ridges in the embryo images.