On 2.5D Surface Reconstruction of Cell Cultures

The domain of image processing for cell microscopy presents non-trivial challenges that must be addressed for consistent image quality. One such challenge concerns the loss of focus as a result of cellular processes, where cell objects may move or change their morphology and, as a result, lie outside of the depth-of-view of the lens. This paper presents two approaches to addressing this problem; namely, the multiscale and geometric methods of image depth estimation. These algorithms are applied to a z-stack of images acquired from a standard phase contrast microscope and a total internal reflection microscope. To assess the algorithms in terms of their scalability, 10x, 20x, and 60x lens objectives are used to offer increased spatial resolutions as well as corresponding improvements in image quality.