Multiresolution approach to automatic detection of spherical particles from electron cryomicroscopy images

Describes an approach for automatic detection and extraction of spherical virus particle projection images from electron cryomicroscopy images. High resolution virus structural studies require combination of thousands of low contrast projection images to generate the resulting three-dimensional reconstruction. Thus, effective and efficient automatic methods are needed for the detection of the numerous low contrast particles. The authors' detection approach combines a Fuzzy C-Mean classification preprocessing step to identify and correct spatial image variation with multiresolution image decomposition to increase image contrast. Following preprocessing, particles are detected using template matching.