Joint Projection and Denoising in CARE

Hi there,

The number of CARE users in Institute Curie is growing really fast, and I have been getting lots of questions about it from multiple users that come to our facility and some have asked me about the prospect of doing content aware 3D denoising, skip Premosa and get the projected result straight out of the CARE workflow.

I keep telling them that it is not released yet and they are still working on it but can I give them a time frame in which they can expect to have this workflow as well?

Rest is great, Cheers,
Varun

@frauzufall @fjug @uweschmitt

2 Likes

Dear Varun,

sorry I don’t have an answer for you but we forwarded this question internally, we’ll keep you posted! (the Uwe you linked is not our Uwe :wink: )
Thank you for the feedback and for helping people to use CARE!

Best,
Deborah

1 Like

Good news @kapoorlab… projection-CARE is out.

http://csbdeep.bioimagecomputing.com/doc/models.html#csbdeep-models-projectioncare

Let us know if it works for you! :slight_smile:

4 Likes

@fjug I emailed my bosses to tell them about this 12 minutes after it was released :slight_smile:, it’s game on, on Monday :slight_smile:, am doing MTrack stuff this weekend. <Kisses!>

Does this mean that we can generate models within the CSBDeep Fiji plugin now?

No it does not mean that.

First test ongoing! …Drum Roll…28 >

Running the inference now!!!

And?!? Happy with the outcome?

2 Likes

I have to go to office soon to see what came out, teamviewer has betrayed me :frowning:, gimme some time to reach my desk.

Something is wrong with the training data, I need a day more now

On the bright side, I was doing stats on 3D reconstructed images and computing NMSE and Structural similarity index measure (SIM) of each slice for 76 Z stacks for (GT/low) and (GT/Restored) image pairs and histogrammed the results to show improvement in the image quality. I actually wanted to make a statement that the restored results are better then GT as the network recoganizes the randomness in the noise but I could not find quantitative measures to support that, so just settled for improvement in image quality argument. Figures attached:

Now I am happy! Test Data (3D was too big but trust me, it worked):

DrosophilaDenoisingProjection_40x_bin1_05_5min_1_w2CSU-488_s1_t26.tif (16.0 MB)

2 Likes

Looking good! Congrats!
Seeing this makes me happy! :slight_smile:

1 Like