|
Comprehensive assessment of SMLM software.The single molecule localization microscopy software challenge is an ongoing competition to assess SMLM software on both simulated and real reference datasets using objective assessment metrics. The 3D challenge remains continuously open to new submissions from both 2D and 3D SMLM software. |
This challenge is based on realistic simulated datasets for 4 popular SMLM modalities:
3D astigmatism, biplane, double-helix and 2D.
Simulations include a 4-states photophysics model and
experimentally-derived PSF.
Qualitative assessments against real 3D STORM test structures, microtubules and NPCs, are also performed.
The challenge remains continuously open to new submissions.
News and views of Nature Photonics: Seamus Holden and Daniel Sage Nature Photonics 10, 2016.
Read full PDF
The Grand Challenge Localization Microscopy was presented in the conference IEEE ISBI 2013. Nearly 30 software have been run, mostly by the authors of the software, thus constituting the first world-wide effort in benchmarking the SMLM software. The results are published in Nat. Method 2015 (PDF).
Read more
These challenges are the first attempt to foster the community to use the same common reference datasets to evaluate the performances of the localization software.
Collaboration through competition: Nature Methods Editorial, Nature Methods 11, 695 (2014).
Grand Challenge: These SMLM challenges are part of the Grand-Challenge in Biomedical Image Analysis.
Super-resolution fluorescence microscopy is an emerging multidisciplinary field that has opened up new opportunities in studying the living cell at the nanometer scale. It overcomes the classical diffraction limit of Abbe, yielding images of biological structures that have lateral resolution down to 10 nm. One of the most promising techniques in this field is single molecule localization. It was announced Method of the Year 2008 by the Nature Methods journal [1] and was was recognized with the 2014 Nobel Prize in chemistry [2]. This technique has been gaining growing attention in microscopy, biology, and engineering conferences, among them ISBI, ever since. The method is based on exciting a spatially sparse set of fluorophores at each time instant, and on fitting the acquired images with a PSF (point spread function) model. This technique was originally proposed and introduced by three independent groups in 2006, giving rise to several acronyms: PALM (photo-activated localization microscopy by Eric Betzig et al.), STORM (stochastic optical reconstruction microscopy by Xiaowei Zhuang et al.), and F-PALM (fluorescence PALM by Samuel Hess). Unlike classical fluorescence microscopy, however, the acquired data cannot be visualized and inspected directly, and an additional processing step is required in this regard. The algorithmic aspects of the single molecule method are therefore extremely important, and this is the focus of the challenge we propose here.
There are currently numerous image-analysis software packages that process super resolution data. A typical data set consists of millions of excited fluorophores, distributed over thousands of images. The excited fluorophores are then detected, localized and visualized by these software tools. Automation is an indispensable aspect of the method [3] and a challenging task [4], as the data is very long image sequence.
Multiple updates to the challeng including assessment against real 3D STORM datasets, interactive graphing and ranking tools and more detailed software reports. Check it out here.
A data exploration system is on-line to analyse the results of the assessment across software x modalities x metrics. interactive leaderboard.
The first results are presented on the bioRxiv: Super-resolution fight club: A broad assessment of 2D & 3D single-molecule localization microscopy software
Thanh-an Pham has a talk at the SMLMS 2017 Symposium in London: Developments of the ongoing 3D SMLM software challenge.
© 2018
Biomedical Imaging Group, Ecole Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland
Last update: 30 Nov 2018
