Human-computer interaction and expert systems for three-dimensional studies of biomedical images

Abstract

Three-dimensional reconstruction of serial sections, observed by microscopy, is an important technique in medicine and biology. To view any part of a structure clearly, that part has to be identified and clearly highlighted in its relationship to other features in a structure. The identification process can be time consuming and tedious if many sections are involved, especially for routine applications, since human identification is required. In this paper we describe how image processing, together with other information on the shape and position of features relative to each other on any one section and throughout the structure, could be incorporated into an expert system and we also show how such a system could be designed. An important feature is the use of human-computer interaction to allow the system to evolve under the guidance of the biological or medical expert. An example of feature identification in a plant-parasitic nematode is used.