Cell volume estimation from a sparse collection of noisy confocal image slices

Measurement and quantification of the volume of cells are very useful for analyzing and understanding cellular behavior, such as cell growth and differentiation. One popular practice is to use Confocal Laser Scanning Microscopy(CLSM) to image cell slices and then reconstruct the 3-D volume of the cells from these serial optical slices. However, all of the current methods of volume estimation using CLSM imaging require large number of cell slices (i.e. high volume of image data). But in many practical situations, especially in case of CLSM based live cell imaging, such high depth resolution is not feasible in order to avoid photodynamic damage to cells from prolonged exposure to laser. In this work, we have addressed this problem of finding out the volume of a plant cell using CLSM when the amount of data is as limited as two to three slices per cell. We assume an ellipsoid model for a cell and estimate the minimum volume ellipsoid that encloses all three cell slices in the 3D stack. The level set based segmentation often tends to produce cell slice areas less than the actual due to poor signal to noise ratio in images and that is the reason why we choose a Minimum Volume Enclosing Ellipsoid over mean or best-fit ellipsoids. We tested the proposed computational method on time-lapse CLSM images of Shoot Apical Meristem (SAM) cells of model plant Arabidopsis Thaliana. The volume estimates of individual cells are reasonably close to what is expected from biological point of view.