Big data viewer and Analysis in 3D in 2019

Hello everybody,
We are lucky in our University because we have access to imaris 7.5 (2012) and Arivis 2 (2017).
We have confocal microscope, airyscan, multiphoton and lightsheet microscope. The files acquire with the lightsheet are pretty big +/- 240 Go per file.
We don’t want to invest too much in an update because they are expensive.
Could you give me some advice to choose free software for 3D analysis? (and can you tell me if the user need some bioinformatic skills?)

best,

2 Likes

Shameless self promotion, but SciView has the most action in terms of active development of 3D visualization support. It is aimed to replace 3D Viewer and has support for BigDataViewer rendering for 3D volume renders.

5 Likes

Hi Alex,
ParaView is also very good, and free, is based on Python.
Bob

2 Likes

Dear Kephale,
Thank you I’ll try to run the file on SciView do you know if we can do analysis in 3D with this tools ? like 3D segmentation or if it’s only to view data .

Dear Robert,
Thank you for your advice I’ll try this software but I’m very worried about the tool to do denoising or segmentation etc…
Do you have some experience with this software to run 3D analysis ?
best

Not too sure what kind of 3D analysis you need to do, being more specific with the requirements will be helpful for giving advice.

Regardless, you could have a look at Paintera. You can manually create dense 3D segmentations with automatic mesh rendering of the segmentations in 3D. It is not available through Fiji but you can simply install it through pip or conda (instructions on website linked above). It currently does not support time series but that may be a feature that could be added down the road. (Disclaimer: I am the author)

1 Like

Dear hanslovsky,
Usually we segment cells or tumor (counting, volume, MFI…), we perform some 3D network analysis on neuron.
We have nly fluorescent microscopy images from confocal or lightsheet microscope.
best,

Alex,
It is difficult to beat Fiji/ImageJ for denoising or segmentation. Paraview handles VERY large images and is designed for analysis of the data in the image, not to alter the image. Free and worth a try but has a steep learning curve.
Good Luck
Bob

2 Likes

Hey @Alex.h,

also some shameless self-promotion from my side. I’m working on a faster version of the pretty popular 3D objects counter (3D-OC) in Fiji (which can do a lot more than just counting btw.) by re-implementing its core on the GPU using clij. For simple denoising / background removal and segmenting nuclei in 3D light sheet data (in our test case about 100 MB 16-bit image stacks), it might be very well suited as it reduces processing time of the original 3D-OC from about a minute to 3-7 seconds (tested on Intel/AMD/NVidia). It doesn’t have a user interface yet, but you can play with it from ImageJ macro for example. Here are some teaching slides about it (look out for the exercise by the end).

I think by the end of the year there will be a proper user interface at least similar to the good-old 3D-OC.

Cheers,
Robert

12 Likes

Thank you for your help

best,

Hi Alex,

If I’m correct, it looks like you would like to do 3D segmentation and 3D neuron tracing on big data (hundreds of Gb). Is that right?
I fully understand that you do not want to invest in updates for partial solutions, and what Robert is offering is a very good option!

Please just keep in mind that the applications I listed above are challenging with big data. Sometimes, big data is already challenging for data handling, storage, processing, before thinking about any analysis really!
So depending on your needs, I think this is still a fair game to compare what you can have for free and what commercial solutions (Imaris, Aivia - I am part of DRVision company) could offer. Also because you might be interested in combining new tools (AI) to your 3D analysis, meaning getting latest tech, etc.

Here my message has no intention to force toward any direction, just to bring some more general info to your question. Hope this helps you to define your priorities and needs :slight_smile:

Best,
Patrice

4 Likes

A post was merged into an existing topic: Implementing high-performance FFT-based algorithms (Deconvolution, …)

24 posts were split to a new topic: Implementing high-performance FFT-based algorithms (Deconvolution, …)

Thank you for your message!

Hi all,

I would like to jump in this conversation. Do any of the big data visualization tools (napari, bigdataviewer, itk, etc) allow for simple editing and saving of files? Specifically, I would like the ability to crop and rotate my stacks (flipping might also be nice).

Thanks,

1 Like

Hey @jclab,

just two quick counter questions: How large are your stacks? Is it individual 3D data sets or time lapses?

Cheers,
Robert

I have multi-channel 3D image stacks (not time lapse). Image stack sizes are 50-100GB

1 Like