Automatic particle detection in digitized electron micrographs

Short, Judith M (2005). Automatic particle detection in digitized electron micrographs. PhD thesis The Open University.



High resolution structural analysis of biological complexes can be carried out by single particle electron microscopy where a large number of particle images are available. Many approaches to automate the process of selection of particle positions from digitized electron micrograph images have been described, but so far none has proved as good as manual selection.
This thesis describes a method which I have developed to locate such biological complexes by matching small boxed areas to a set of reference images using the radius of gyration, complemented by a series of other simple criteria. From the reference images, parameters such as the ratio between the average density of the central area and that in its surrounding band, and the density sum and variance are calculated. They are compared with corresponding values from a moving square window of densities extracted from the micrograph, and the coordinates of successfully matched candidate squares are recorded. Since the same particle is detected in a series of overlapping windows, candidates found to be within close proximity are grouped, and the best-fitting one is selected from each cluster. Along with a small stack of boxed reference images, a few specified parameter values, such as the particle radius and the minimum acceptable distance between particle centres are required to select the windows. Micrograph labels and other areas that do not contain appropriate specimens are automatically ignored in order to minimize false positives, and reduce the computing time.
A computer program SLEUTH written to carry out this method of automatic particle detection includes a graphical user interface to assist the user in setting up the parameter values. The program has been tested successfully on a variety of different biological structures, from both negatively stained and ice-embedded specimens.

Viewing alternatives

Download history


Public Attention

Altmetrics from Altmetric

Number of Citations

Citations from Dimensions

Item Actions