ZOLA-3D Aristov, Lelandais, Rensen & Zimmer. Nature - - PowerPoint PPT Presentation

Zernike Optimized Localization Algorithm

ZOLA-3D

Aristov, Lelandais, Rensen & Zimmer. Nature communication 2018 https://github.com/imodpasteur/ZOLA-3D

Benoit Lelandais IMOD - Institut Pasteur Paris, France

Image formation model

ZOLA-3D

PSF({\color{Red} \boldsymbol{\mathbf{\Phi}}},\ boldsymbol{\mathbf{x}},\bol dsymbol{\mathbf{y}},{\color{ Mahogany} \boldsymbol{\mathbf{z}}},{\c

• lor{DarkGreen}

\boldsymbol{\mathbf{f}}})=|\ mathcal{F}_{k_x,k_y}\rho(k_ x,k_y)e^{j{\color{Red} \boldsymbol{\mathbf{\Phi}}}( k_x,k_y)}e^{2\pi j(\boldsymbol{\mathbf{x}}k_x +\boldsymbol{\mathbf{y}}k_y )}e^{2\pi j({\color{Mahogany} \boldsymbol{\mathbf{z}}}{\co lor{NavyBlue} \boldsymbol{\mathbf{k_{z}}}}

• {\color{DarkGreen}

\boldsymbol{\mathbf{f}}}{\col

• r{YellowOrange}

\boldsymbol{\mathbf{k_{f}}}} )}|^2

ZOLA-3D Calibration

• The phase is represented by a

set of K zernike coefficients:

• Phase retrieval by maximum

likelihood approach ZOLA-3D

ZOLA-3D - Localization ZOLA-3D - post-processing

• 3D Drift correction (redundant cross-correlation)
• Temporal grouping (Merging)
• Visualization
• ...

ZOLA-3D

Localization precision Goodness

• f fit

• Many PSF models:
• Model refractive index mismatch:

○ Prediction of the PSF at any depth ○ Z position relative to the coverslip

• ImageJ / Fiji plugin

○ https://github.com/imodpasteur/ZOLA-3D

• Accelerated reconstruction using Nvidia GPU

Main advantages

ZOLA-3D

2D Astigmatism (1µm) Saddle-point (3µm) Tetrapod (6 µm) Bi-astigmatism benoit.lelandais@pasteur.fr