SuRVoS: Super-Region Volume Segmentation workbench

Abstract

Segmentation of biological volumes is a crucial step needed to fully analyse their scientific content. Not having access to convenient tools with which to segment or annotate the data means many biological volumes remain under-utilised. Automatic segmentation of biological volumes is still a very challenging research field, and current methods usually require a large amount of manually-produced training data to deliver a high-quality segmentation. However, the complex appearance of cellular features and the high variance from one sample to another, along with the time-consuming work of manually labelling complete volumes, makes the required training data very scarce or non-existent. Thus, fully automatic approaches are often infeasible for many practical applications. With the aim of unifying the segmentation power of automatic approaches with the user expertise and ability to manually annotate biological samples, we present a new workbench named SuRVoS (Super-Region Volume Segmentation). Within this software, a volume to be segmented is first partitioned into hierarchical segmentation layers (named Super-Regions) and is then interactively segmented with the user's knowledge input in the form of training annotations. SuRVoS first learns from and then extends user inputs to the rest of the volume, while using Super-Regions for quicker and easier segmentation than when using a voxel grid. These benefits are especially noticeable on noisy, low-dose, biological datasets.

CytoCensus: mapping cell identity and division in tissues and organs using machine learning

“…Supervised machine learning methods (Luengo et al, 2017;Arganda-Carreras et al, 2017;Logan et al, 2016;Chittajallu et al, 2015;Sommer et al, 2011) require the user to provide training examples by manually identifying (annotating) a variety of cells or objects of interest, often requiring laborious "outlining" of features to achieve optimal results. However, our use of a "point and click" interface ( Figure S2), to simplify manual annotation, and proximity map output, makes it quick and easy for a user to train and retrain the programme.…”

“…However, these three approaches require advanced knowledge of image processing, programming and/or extensive manual annotation. Other software such as Advanced Cell Classifier are targeted at analysis of 2D data, whilst programs such as RACE, SuRVoS, 3D-RSD and MINS are generally tailored to specific applications (Luengo et al, 2017;Stegmaier et al, 2016;Lou et al, 2014;Cabernard & Doe, 2013;Homem et al, 2013;Arganda-Carreras et al, 2017;Logan et al, 2016;Gertych et al, 2015). Recently, efforts to make deep learning approaches easily accessible have made great strides (Falk e t a l .…”

X-ray tomography of cryopreserved human prostate cancer cells: mitochondrial targeting by an organoiridium photosensitiser

“…The final aligned stacks were then generated from the calculated z-axis and pitch angles using linear interpolation, producing tomograms with no positional drifts. Reconstructed tomograms were generated in 3dmod [33,34] and post-processed to remove unwanted information (noise or background), saved as reconstruction files ready for analysis by SuRVoS [35].…”

Segmentation and Modelling of the Nuclear Envelope of HeLa Cells Imaged with Serial Block Face Scanning Electron Microscopy

“…These, and other properties of the sample and image, mean that techniques that work well for imaging techniques like immunohistochemistry and light microscopy (e.g., watersheds) [30][31][32][33] do not usually port well to EM. Shallow [34,35] and deep learning methodologies [36,37] are becoming popular for segmentation and classification of image data. However, these techniques require significant computational power, as well as very large training data sets [38,39], which are rarely available in biological electron microscopy and were not available for our specific data sets.…”