Gigabytes to bytes: Automated denoising and feature extraction as applied to the analysis of HIV architecture and variability using electron tomography

Advances in automated data acquisition in electron tomography have led to an explosion in the amount of data that can be obtained about the spatial architecture of a variety of biologically and medieally relevant objects with sizes in the "nano" range of 10-1000nm. The development of methods to analyze the vast amounts of information contained in these tomograms is a major challenge since the electron tomograms are intrinsically noisy. A fundamental step in the automatic analysis of large amounts of data for statistical inference is to segment 3D features in cellular tomograms that can work robustly and rapidly despite of low signal to noise ratios inherent to biological electron microscopy. This work evaluates various denoising techniques on tomograms obtained using dual-axis simultaneous iterative reconstruction (SIRT) technique. Using three-dimensional images of HIV in infected human macrophages as an example, we demonstrate that transform domain-denoising techniques significantly improve the fidelity of automated feature extraction. Importantly, our approaches represent an vital step in automating the efficient extraction of useful information from large datasets in biological tomography, and facilitate the overall goal of speeding up the process of reducing gigabyte-sized tomograms to byte-sized data.