The preprint paper describing this method is referenced below.
You can see some examples in the bottom of the page.
Our new manuscript, ML-SIM, is now on arXiv: https://t.co/BQo1VH4NiN
— Charles Nicklas Christensen (@charlesnchr) March 27, 2020
Sadly I won't get to present the work at @FOMconferences this year (#COVID19 😑), but at least I am happy to provide some non-COVID reading material for those interested. #ML_SIM #group_laser pic.twitter.com/Z9XrcOscQy
Get the source code now at GitHub.
Want
to see how it works?
Give ML-SIM a try with
your own images - simply drop them in the box below. You can also use the ML-SIM Test Images that consist of simulated images and data from two distinct microscopes (SLM-based and interferometric, respectively).
Note that the currently served model is trained on 9-frame SIM stacks of 512x512 resolution.
See the difference between reconstructions from
ML-SIM and standard wide-field microscopy.
Tried the new machine-learning #superresolution structured illumination #microscopy reconstruction algorithm https://t.co/NgeeUhaTGX. Nephrin staining of a kidney comparing widefield and @zeiss_micro + ML-SIM algorithms. Thanks @palankarr for pointing out! https://t.co/9AtHprz4wg pic.twitter.com/WHh1pgzu6z
— Florian Siegerist (@Siegerist) March 29, 2020
Test of ML-SIM posted on Twitter.
Calibration structured illumination image of beads.
Test image with simulated diffraction and illumination patterns.
Structured illumination image of endoplasmic reticulum.