Automated detection of inflammatory cells in whole anterior chamber of a uveitis mouse from swept-source optical coherence tomography images

摘要

Cell grading in a rodent anterior chamber is essential for anterior inflammation evaluation in preclinical vision research. This paper describes a computerized method for detection and counting of the anterior chamber cells from swept-source optical coherence tomography (SS-OCT) images of a experimental rodent model of uveitis. The volumetric anterior segment OCT data is obtained from 100 kHz SS-OCT imaging of mouse eye in vivo. For the OCT cross-sections, each OCT structural image is de-speckled and binarized. After removal of cornea, iris, and crystalline lens structures connected to the binary image border, an area thresholding is then employed for each labeled region to isolate only cell-like objects in the anterior chamber, followed by roundness estimation of the objects to identify potential cell candidates in the data. Eventually, the cell candidates are counted and graded as total number of cells in the anterior chamber.