Pattern recognition software and techniques for biological image analysis



Written by Lior Shamir, John D. Delaney, Nikita Orlov, D. Mark Eckley and Ilya G. Goldberg
Date of Publication: 24 November 2010



The increasing prevalence of automated image acquisition systems is enabling new types of microscopy experiments that generate large image datasets. However, there is a perceived lack of robust image analysis systems required to process these diverse datasets. Most automated image analysis systems are tailored for specific types of microscopy, contrast methods, probes, and even cell types. This imposes significant constraints on experimental design, limiting their application to the narrow set of imaging methods for which they were designed. One of the approaches to address these limitations is pattern recognition, which was originally developed for remote sensing, and is increasingly being applied to the biology domain. This approach relies on training a computer to recognize patterns in images rather than developing algorithms or tuning parameters for specific image processing tasks. The generality of this approach promises to enable data mining in extensive image repositories, and provide objective and quantitative imaging assays for routine use. Here, we provide a brief overview of the technologies behind pattern recognition and its use in computer vision for biological and biomedical imaging. We list available software tools that can be used by biologists and suggest practical experimental considerations to make the best use of pattern recognition techniques for imaging assays.


Computer-aided analysis of microscopy images has been attracting considerable attention in the past few years, particularly in the context of high-content screening (HCS). The link between images and physiology is well established, and it is common knowledge that a significant portion of what we know about biology relies on different types of microscopy and other imaging devices. Automated image acquisition systems integrated with laboratory automation have produced image datasets that are too large for manual processing. This trend led to a new type of biological experiment, in which the image analysis must be performed by machines. Clearly, this approach is different than the bulk of the microscopy performed for the past ∼400 years. However, while the availability of automated microscopy, laboratory automation, computing resources, and digital imaging and storage devices has been increasing consistently, in some cases the bottleneck for high-throughput imaging experiments is the efficacy of computer vision, image analysis, and pattern recognition methods [1]. Computer-based image analysis provides an objective method of scoring visual content independently of subjective manual interpretation, while potentially being more sensitive, more consistent, and more accurate [2]. These advantages are not limited to massive image datasets, as they allow microscopy to be used as a routine assay system even on a small scale.

An effective computational approach to objectively analyze image datasets is pattern recognition (PR, see Box 1). PR is a machine-learning approach where the machine finds relevant patterns that distinguish groups of objects after being trained on examples (i.e., supervised machine learning). In contrast, the other approach to machine learning and artificial intelligence is unsupervised learning, where the machine finds new patterns without relying on prior training examples, usually by using a set of pre-defined rules. An example of unsupervised learning is clustering, where a dataset can be divided into several groups based on pre-existing definitions of what constitutes a cluster, or the number of clusters expected. Read the rest of the publication here.

This publication was originally done on Plos. Read the original publication.

Mohamed HAMZA