Automated segmentation and quantitative study of retinal pigment epithelium cells for photoacoustic microscopy imaging

We develop an improved region growing method to realize automatic retinal pigment epithelium (RPE) cell segmentation for photoacoustic microscopy (PAM) imaging. The minimum bounding rectangle of the segmented region is used in this method to dynamically update the growing threshold for optimal segmentation. Phantom images and PAM imaging results of normal porcine RPE are applied to demonstrate the effectiveness of the segmentation. The method realizes accurate segmentation of RPE cells and also provides the basis for quantitative analysis of cell features such as cell area and component content, which can have potential applications in studying RPE cell functions for PAM imaging.