Detecting particles in cryo-EM micrographs using learned features

Abstract

A new learning-based approach is presented for particle detection in cryo-electron micrographs using the Adaboost learning algorithm. The approach builds directly on the successful detectors developed for the domain of face detection. It is a discriminative algorithm which learns important features of the particle's appearance using a set of training examples of the particles and a set of images that do not contain particles. The algorithm is fast (10 s on a 1.3 GHz Pentium M processor), is generic, and is not limited to any particular shape or size of the particle to be detected. The method has been evaluated on a publicly available dataset of 82 cryoEM images of keyhole lympet hemocyanin (KLH). From 998 automatically extracted particle images, the 3-D structure of KLH has been reconstructed at a resolution of 23.2 A which is the same resolution as obtained using particles manually selected by a trained user.

Automatic particle selection: results of a comparative study

“…Class II methods are based on feature recognition where algorithms work by way of recognizing local or global salient features inherent to particle images without the use of a 3D reference structure, called feature-based approaches, including BajajÕs, HallÕs, MallickÕs, VolkmannÕs, and ZhuÕs algorithms. Unlike the other feature-based approaches reported here, MallickÕs algorithm uses machine learning as the basic tool to learn both discriminative features and a cascade of classifiers for particle detection (Mallick et al, 2004). There are distinct advantages to each of these approaches and these are described later in this section.…”

Towards automated screening of two-dimensional crystals

“…An approach like this was used by the Fujiyoshi laboratory where targets were identified based on the contrast of the identified object relative to the carbon background as well as the overall shape. Learning-based approaches that have been successfully used for particle detection (Mallick et al, 2004) are another promising avenue to pursue.…”

Reference-free particle selection enhanced with semi-supervised machine learning for cryo-electron microscopy

“…Sorzano et al (2009) and Mallick et al (2004) apply a classifier cascade to perform cost-sensitive learning, a learning methodology that minimizes the number of false positives. Ogura and Sato (2004b) developed a pyramidal neural network algorithm to both characterize and discriminate particles from contaminants/noise.…”