Semi-automatic methods and segmentation of hyphal images

The underlying aim of this research is to characterise the growth and morphology of arbuscular mycorrhizal hyphal distributions (i.e. spatial distributions of hyphae from a type of fungus), in order to enable assessment of efficiency of hyphal colonisation of a growth medium relative to localised variations in nutrient availability. Patterns of distribution are likely to have a profound influence on nutrient acquisition and relocation within a medium. Although software exists to measure the hyphae (once extracted manually from an image) segmenting each hypha is inaccurate, tedious and time consuming. Hyphal segmentation is successfully tackled using three different computer-assisted approaches with varying degrees of user-input. Results are reported from an evaluation study comparing the methods both quantitatively and qualitatively; this involved several test subjects segmenting simulated images. This work forms part of a wider study to investigate a number of semi-automatic techniques for performance, flexibility and practicality of use