Recent advances in microscopy techniques enable whole-brain activity imaging of small animals at cellular resolution. However, image analysis techniques for detecting and tracking neurons and for measuring neural activities from such imaging data have not yet been established. The soil nematode C. elegans is the only animal that all neurons and the wiring diagram of the whole nervous system are known completely, and is one of the ideal model animals for understanding the mechanisms of information processing of nervous system. The neuronal nuclei are densely packed in the head region of the nematode, and some nuclei will be overlooked. For more accurate nucleus detection, we have developed a curvature-based segmentation method and a gaussian mixture-based image approximation method. The main objective of this project is combining the two methods and developing a framework of bio-image analysis for detecting, tracking, and measuring densely distributed cells in three-dimensional time-lapse images.