NeuroDeblur

The Highest Performance Deconvolution Software
MBF Bioscience > NeuroDeblur

Product Overview

NeuroDeblur is the premier deconvolution and artifact removal software for large 3D microscopy datasets. It supports a wide range of microscopy modalities, including light sheet, laser scanning confocal, spinning disc confocal, two-photon, widefield fluorescence, and brightfield. NeuroDeblur uses advanced algorithms and GPU acceleration to produce images that are clearer than the raw images obtained by the microscope.

 

NeuroDeblur is engineered to work with the most demanding image sets. It can be used in two ways – with an easy to use graphical user interface, or as a command-line tool for automated image processing pipelines.

Key Benefits

NeuroDeblur is the leading solution for deconvolution and artifact removal for microscopy images. It is particularly useful for working with large data sets from light sheet and confocal microscopes. With computational optimizations for the fastest performance using GPU.

NeuroDeblur is engineered by experts in microscopy and deconvolution to produce optimal results from a wide array of microscopes.

Even images from the best microscopes can be further improved using NeuroDeblur.

NeuroDeblur contains the following important features:

  • Excellent deconvolution results with light sheet and confocal microscopy without or without PSF measurements
  • Deconvolution using matching accurate PSF models for light sheet and confocal microscopy
  • Deconvolution using a PSF measured from the imaging system
  • Powerful adaptive background artifact correction
  • Fast multiprocessor aware deconvolution with excellent regularizing for noise minimization
  • Ultra-fast GPU based processing with modern Nvidia graphic cards (> 100 million voxels/min)
  • Batch deconvolution for building an automated imaging workflow
  • Automatic block-wise processing of large data sets even on computers with limited RAM
  • Sophisticated stripe-artifact removal for light sheet microscopy data
  • Side-by-side comparison of original and deconvolved data in synchronized display windows
  • Process select ROIs and color channels of proprietary 4D- or 5D-input formats (e.g. Zeiss, Evident, Leica, etc).
  • Optimize the display of deconvolved results via additive and subtractive color channel mixing.
  • Image post-processing by adaptive histogram equilibration (CLAHE) and unsharp masking

Download NeuroDeblur product sheet here.

How does deconvolution software work?

Image generation by a microscope can be modeled by a numerical operation called convolution: The incident light coming from the sample is convolved with the point spread function (PSF) of the microscope to construct the image.  The PSF of a fluorescence microscope can be described as the image you would get from an infinitely small and infinitely bright light-emitting point-source. In practice, it can be (approximately) measured by imaging tiny fluorescent particles that are below the resolution limit of the microscope. The PSF of an ideal microscope would just be a mathematical point. However, in any real microscope it has a certain spatial extent due to the wavelength of the illumination light and the aberrations of optical lenses. The spatial extent of the PSF inevitably causes the limited resolution and blurring observed to varying degrees in any microscopic image.

 

However, If the shape of the PSF of a microscope is known this process can (at least partially) be reverted by mathematically reconstructing the original shape of the sample. This process is called deconvolution:

 

Image = convolution (sample, PSF), sample = deconvolution (Image, PSF)

 

The shape of the PSF can either be measured by imaging small fluorescent particles (e.g. of 200 nm diameter) that are just below the resolution limit of the microscope, or it can be approximated using a theoretical model of the imaging process (synthetic point spread function). Both approaches are supported by NeuroDeblur.


A practical problem with any deconvolution is the potential amplification of the noise inevitably added during the imaging process by the camera and the stochastic spatial nature of photons. There are different mathematical regularization approaches to cope with this problem. The regularization approach applied by NeuroDeblur (flux preserving regularization) has the advantage of preserving the photogrammetry of data, i.e., the light intensities in distinct image regions are not shifted relatively to each other by the deconvolution process. Flux preserving regularization was originally developed for deconvolving astronomical data and to our knowledge has not been used by deconvolution software for microscopy before. Preservation of photogrammetry is especially important if quantitative comparisons of intensity are to be performed with deconvolved data. More details about the deconvolution algorithm used by NeuroDeblur in Scientific Reports, 2019 (doi.org/10.1038/s41598-019-53875-y).

Key Features

Excellent results even without PSF measurements by accurately computed PSFs for light sheet and confocal microscopy

 

A matching point spread function (PSF) is essential for obtaining optimal deconvolution results. With NeuroDeblur, the PSF can either be determined experimentally or it can be derived from a theoretical model of the image formation process in the microscope (synthetic PSFs, figure left). The equations used for modeling the PSF of a light-sheet microscope can be found in this publication in Scientific Reports, 2019 (doi.org/10.1038/s41598-019-53875-y) and have been proved for years to provide excellent results that are comparable with those obtained using a measured PSF. Due to their capability to utilize synthetic PSFs NeuroDeblur eliminates the requirement of laborious and error-prone PSF measurements and makes high quality deconvolution easy.

Key Features

High processing-speed in CPU- and GPU-based deconvolution modes.

 

The time required for deconvolving an image stack depends on the number of voxels that need to be processed as well as the computer hardware. Modern GPUs for the numerous parallel FFT-calculations required for deconvolution bring an enormous increase in processing speed, which is often 100x faster than CPU-based processing. Using high-end NVIDIA graphic cards NeuroDeblur can deconvolve one billion voxels in less than 10 minutes.

Key Features

Sophisticated adaptive image background removal algorithm.

 

Light sheet, as well as confocal microscopy recordings often are compromised by background fluorescence causing blurring and masking of details. While an image background that is approximately constant can be removed easily by plain subtraction, determining and removing an inhomogeneous background is much more difficult. NeuroDeblur addresses this problem by applying a modified rolling ball background removal approach, which was published in 2021 (https://doi.org/10.1002/jbio.202100290) This technique results in a striking improvement in clarity with many types of image data, while preserving fine details of interest as e.g. dendrites, axons, and other fibers.

Key Features

Command-line version for simple embedding of deconvolution into complex work-flows and pipelines.

 

Deconvolution in NeuroDeblur can either be executed via an easy-to-use interface (that also includes functionality for visualizing deconvolution results and post-processing data), or it can be run using command line sequences allowing a seamless integration of deconvolution into automated image processing workflows.

Key Features

Stripe-artifact removal in light sheet microscopy images.

 

Light sheet microscopy images often contain stripeing artifacts, which are caused by light absorbing particles throwing shadows along the propagation direction of the light-sheet. NeuroDeblur reduces these stripe artifacts using a directional frequency filtering approach described in Scientific Reports, 2020 (doi.org/10.1038/s41598-020-71737-w).

Powerful Deconvolution Software for Light Sheet and Confocal Microscopy

 

Binaree Cleared Data Imaged by ClearScope® – the Light Sheet Theta Microscope

 

 

< >
< >

Who Is Using MBF Products?

MBF products are used across the globe by the most prestigious laboratories. 

Cited in Peer Reviewed Scientific Publications

MBF’s software utility is underscored by the number of references it receives in the worlds most important scientific publications.

Testimonials

Robust Professional Support

Our service sets us apart, with a team that includes Ph.D. neuroscientists, experts in microscopy, stereology, neuron reconstruction, and image processing.  We’ve also developed a host of additional support services, including:

 

  • Forums
    We have over 25 active forums where open discussions take place.
    >> Learn More
  • On-Site/Training
    We’ve conducted over 750 remote software installations.
    >> Learn More
  • Webinars
    We’ve created over 55 webinars that demonstrate our products & their uses.
    >> Learn More

Request an Expert Demonstration

At MBF, we’ve spent decades understanding the needs of researchers and their labs — and have a suite of products and solutions that have been specifically designed for the needs of today’s most important and advanced labs. Our commitment to you is to spend time with you discussing the needs of your lab — so that we can make sure the solutions we provide for you are exactly what you’ll need. It’s part of our commitment to supporting you — before, during, and after you’ve made your decision. We look forward to talking with you!