Real-Time Protein Crystallization Image Acquisition and Classification System

Abstract

In this paper, we describe the design and implementation of a stand-alone real-time system for protein crystallization image acquisition and classification with a goal to assist crystallographers in scoring crystallization trials. In-house assembled fluorescence microscopy system is built for image acquisition. The images are classified into three categories as non-crystals, likely leads, and crystals. Image classification consists of two main steps – image feature extraction and application of classification based on multilayer perceptron (MLP) neural networks. Our feature extraction involves applying multiple thresholding techniques, identifying high intensity regions (blobs), and generating intensity and blob features to obtain a 45-dimensional feature vector per image. To reduce the risk of missing crystals, we introduce a max-class ensemble classifier which applies multiple classifiers and chooses the highest score (or class). We performed our experiments on 2250 images consisting 67% non-crystal, 18% likely leads, and 15% clear crystal images and tested our results using 10-fold cross validation. Our results demonstrate that the method is very efficient (< 3 seconds to process and classify an image) and has comparatively high accuracy. Our system only misses 1.2% of the crystals (classified as non-crystals) most likely due to low illumination or out of focus image capture and has an overall accuracy of 88%.

FocusALL: Focal Stacking of Microscopic Images Using Modified Harris Corner Response Measure

“…The images for protein crystallization trial sets were captured using the acquisition system described in [1]. The images are collected at a resolution 2560×1920.…”

“…It is important to process focused images not to miss protein crystals. Once a protein crystal is detected, an expert tries to optimize conditions to improve the quality of a protein crystal to be evaluated using X-ray diffraction [1]. …”

Optimizing Associative Experimental Design for Protein Crystallization Screening

“…#3553), with the protein: precipitant ratios ( v/v ) for the drops being 1:1, 2:2, and 4:1. Plates were imaged using the in-house developed Crystal X2 imager [24] (iXpressGenes/Molecular Dimensions), with the first set of images immediately after set up, on days 1, 2, 4, and thence on a weekly basis for the next 6 weeks. Plates were scored by visual observation, with the scores then adjusted by reference to the fluorescent images [18].…”

“…Sigdel et. al summarize some of the different scoring (or classification categories) methods in [24]. Since Hampton scoring [25] provides additional crystal subcategories with respect to other scoring scales, a revised version of Hampton scoring is used as shown in Table I for optimization and AED analysis.…”

Image Analytical Approach for Needle-Shaped Crystal Counting and Length Estimation

“…Although eq 2 = + y ax b i i (2) can be used to describe this line, in practice when the line approaches vertical, the slope value a becomes infinitely large. Hence from a computational perspective, it is practical to express a line using the mathematical expression in eq 3, 26 which can be derived from eq 2…”