Single-Molecule Localization Microscopy  •  Software Benchmarking

Collection of reference datasets

The benchmarking of SMLM software package mainly relies on the usage of common reference datasets.

banner

Real Experiments • 3D Astigmatism

The experimental datasets consist of sequences of frames from real stained biologicial samples. A Z-stack of beads is provided for the axial calibration.

Tubulin-A647-3D

3D astigmatic image of microtubules imaged using dSTORM with a cylindrical lens (112'683 frames).
Microtubules in U-2 OS cells, labeled with anti-alpha tubulin primary and Alexa Fluor 647-coupled secondary antibodies.
Calibration data for the tubulin dataset: fluorescent beads adsorbed to a glass coverslide in water are stepped in Z at high SNR.

Courtesy of Jonas Ries (EMBL)

Reference: Li et al., Nature Methods 15, 2018. (raw data of Figure S6)

NPC-A647-3D

3D astigmatic dSTORM image of nuclear pore complex.
Nup107-SNAP-BG-AF647 in U-2 OS cells, imaged using dSTORM with a cylindrical lens (150'000 frames).
Calibration data for the dataset: fluorescent beads adsorbed to a glass coverslide in water are stepped in Z at high SNR.

Courtesy of Jonas Ries (EMBL)

Reference: Li et al., Nature Methods 15, 2018. (raw data of Figure S9)

Realistic Simulations • 3D • Experimental PSF

The experimental datasets consist of sequences of frames from real stained biologicial samples. A Z-stack of beads is provided for the axial calibration. The same 3D simulated structure is imaged in different experimental conditions and for different modality: 2D, 3D astigmatism (AS), 3D biplane (BP), and 3D double-helix (DH). As a testing phase, we have also the quad-plane (QP) datasets for high-density conditions. The training datasets are provided with ground-truth localizations.

picture1
Microtubules • Contest datasets

Published June 7, 2016 as training dataset for the SMLM challenge 2016

Three crossing microtubules in a field of view of 6.4 x 6.4 x 1.5 μm

Two experimental imaging conditions:

N1: Typical photon counts and background levels for Alexa647 labelled STORM sample

N2: Photoswitchable fluorescent protein labelled sample such as mEos2 or Dendra2

MT1.N1.LD
Low density Molecule density: 0.2 High SNR (N1) 19'996 frames AS • DH • BP
MT2.N1.HD
High density Molecule density: 2 High SNR (N1) 3'125 frames AS • DH • BP • QP
MT3.N2.LD
Low density Molecule density: 0.2 Low SNR (N2) 20'000 frames 2D • AS • DH • BP
MT4.N2.HD
High density Low SNR (N2) 3'020 frames 2D • AS • DH • BP • QP

picture1
Endoplasmic Reticulum • Contest datasets

Published June 7, 2016 as training dataset for the SMLM challenge 2016

Simulation of cellular organelle (endoplasmic reticulum/mitochondria) inspired structure in the field of view of 6.4 x 6.4 x 0.7 μm

One experimental imaging conditions:

N3: Typical photon counts and level of background for site specific dye-labelled live cell STORM sample such as ER Tracker

ER1.N3.LD
Low density Molecule density: 0.2 Very low SNR (N3) 19'620 frames 2D
ER2.N3.HD
High density Molecule density: 5 Very low SNR (N3) 3'020 frames 2D

picture1
Microtubules • Training datasets

Published June 7, 2016 as training dataset for the SMLM challenge 2016

Three crossing microtubules in a field of view of 6.4 x 6.4 x 1.5 μm

Two experimental imaging conditions:

N1: Typical photon counts and background levels for Alexa647 labelled STORM sample

N2: Photoswitchable fluorescent protein labelled sample such as mEos2 or Dendra2

Ground-truth
localizations
MT0.N1.LD
Low density Molecule density: 0.2 High SNR (N1) 2DASBPDH
MT0.N1.HD
High density Molecule density: 2 High SNR (N1) 2DASBPDH
MT0.N2.LD
Low density Molecule density: 0.2 Low SNR (N2) 2DASBPDH
MT0.N2.HD
High density Molecule density: 2 Low SNR (N2) 2DASBPDH

Real Experiments • 2D

The experimental datasets consist of sequence of frames from real stained biologicial samples. The standard acquisition parameters are also provided.

Tubulin ConjAL647

Experimental sequence of 27'529 frames (128x128 pixels) with the parameters of acquisition.

Reference: Suliana Manley, Julia Gunzenhäuser and Nicolas Olivier, Current Opinion in Chemical Biology 15, 2011.

Tubulin AF647

The dataset represents a fixed cell, stained with mouse anti-alpha-tubulin primary antibody and Alexa647 secondary antibody. The intermittent increase in signal is due reactivation with a 405 nm laser.
This dataset is an experimental sequence of 9990 frames of 128x128 pixels.

Courtesy of Nicolas Olivier and Suliana Manley, LEB, EPFL

Tubulin • 2D • Long Sequence

Experimental sequence of 1'500 frames with the parameters of acquisition
High density of fluorophores per frame.

Courtesy of Nicolas Olivier and Debora Keller, LEB, EPFL

Tubulin • 2D • High Density

Experimental sequence of 500 frames with the parameters of acquisition
High density of fluorophores per frame.

Courtesy of Nicolas Olivier and Debora Keller, LEB, EPFL

Realistic Simulations • 2D • Theoretical PSF

These datasets are simulated sequence of realistic frames. The bio-inspired samples are 3D continuous models that imitates biological structures. These datasets were used for the challenge 2013.

Reference: Sage et al. Quantitative evaluation of software packages for SMLM, Nature Methods 12, 2015.

picture1
picture1
picture1
Tubular structure • Constest datasets

Published February 1, 2013 as training dataset for the SMLM challenge 2013

1 LS
Tubulins of various diameters 10'000 frames Low density, long sequence
1 HD
Tubulins of various diameters 1'000 frames High density, short sequence
2 LS
Network of tubulins 12'000 frames Low density, long sequence
2 HD
Network of tubulins 1'200 frames High density, short sequence
3 LS
Helicoidal tubes, deep sample from 0 to 1μm 7'000 frames Low density, long sequence
3 HD
Helicoidal tubes, deep sample from 0 to 1μm 600 frames High density, short sequence

picture1
picture1
Tubular structure • Training datasets

Published February 1, 2013 as training dataset for the SMLM challenge 2013

Two structures:

Tubulins: This sample consists to a realistic structure of 7 tubulins (constant diameter 25 nm) and 1 tubulin (constant diameter 40 nm

Bundled Tubes: Bundles of 8 tubes of 30 nm diameter2

Tubulins I
100000 fluorophores 2400 frames Low level of read-out noise 2D, theoretical PSF
Tubulins II
100000 fluorophores 2400 frames Low level of read-out noise 2D, theoretical PSF
Bundled Tubes Long Sequence
81049 fluorophores 12000 frames 2D, theoretical PSF
Bundled Tubes High Density
81049 fluorophores 168 frames 2D, theoretical PSF

Artificial Constructions

These datasets represents artificial structure with some actived fluorophores at some specific postions. These datasets are mainly used to evaluated particular features of the localization software, like the ability to discriminate two neighboring fluorophores in the same frame.

picture1
Eye

This dataset consists to 4 non-overlapping tubes of 1nm of radius in a field of view 38.4 x 38.4 μm. Only 300 fluorophores are activated in 40 frames. It is an ideal dataset: no perturbation, no additional noise.

picture1
Snow

This dataset contains fluorophores are placed on a regular grid with various depth and various number of photons. The field of view is 50 μm x 50 μm

picture1
Seashell

The synthetic sample has been created articifically using a 3D artificial structure shape. This shape represents a seashell of six branches with various fluorophore densities. This sample is provided in two versions: flat sample (no Z) and thick sample (1 μm of depth).

© 2018 Biomedical Imaging Group, Ecole Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland
Last update: 30 Nov 2018