News

Have you tried out palm3d? We'd love to hear how it went. We need your feedback to make sure the software works on everyone's data, not just ours.

Introduction

Fluorescence microscopes are an awesome tool for biologists, but get blurry if you zoom in too much. It turns out you can remove a lot of this blur if your fluorescent molecules glow one at a time. (See PALM, STORM, FPALM for examples).

A drawback of this trick is that the raw data from your microscope looks like gibberish:

http://www.youtube.com/watch?feature=player_embedded&v=ffFDMgxcC_8' target='_blank'>http://img.youtube.com/vi/ffFDMgxcC_8/0.jpg' width='425' height=344 />

You need software to turn this data into a three-dimensional image like this one:

http://www.youtube.com/watch?feature=player_embedded&v=yEM4gwMavKI' target='_blank'>http://img.youtube.com/vi/yEM4gwMavKI/0.jpg' width='425' height=344 />

That's what 'palm3d' does.

Getting started

Start with our step-by-step tutorial. After you're used to the code, try processing some of your own data, and let us know how it goes.

Why use palm3d?

Other software exists to do the same thing. For example, QuickPALM is an excellent project that tackles a similar problem. 'palm3d' tries to be hardware-agnostic and reasonably fast, with a focus on three-dimensional super-resolution imaging.

We developed palm3d while PALM-imaging thick cells tagged with fluorescent proteins. palm3d overcomes many difficulties presented by these samples: * Water-immersion objectives are better than oil objectives for thick samples, but can be inconsistent; palm3d deals well with these inconsistencies. * Drift tracking is difficult in thick samples, but palm3d corrects drift using a simple trick. * Genetically expressed fluorescent proteins are dim compared to other PALM/STORM tags, and thick samples give noisy images due to autofluorescence. We've tested palm3d in samples up to ~10 microns thick, and it works well with fluorescent proteins.

What hardware do I need?

Short answer

A scope that can do 2D PALM/STORM, with a precise axial sample positioner.

Long answer

Read our paper, where we describe how to build a 3D PALM for thick samples. If you build the same scope, it should work for samples thicker than 10 microns, tagged with two-photon activatable fluorophores. We've had the best luck with PA-mCherry, but we suspect there are other good two-photon activatable fluorophores. Let us know what you try!

If you don't have a femtosecond laser, or don't want to use two-photon activatable fluorophores, you can use one-photon activation instead. One-photon activation works great for thin samples (~1 micron or less), and should be ok even for moderately thick samples (~3 microns).

Most existing 2D PALM/STORM scopes should work fine with palm3d, as long as they have a precise axial positioner for calibration and jump-tracking. We recommend a water-immersion objective to avoid depth-dependent aberrations. Our paper describes how we put a cylindrical lens in our microscope's imaging path to increase z-resolution, but this is optional. palm3d works either way, which makes it easy to experiment with different hardware configurations.

If you plan to take a lot of PALM data, consider getting a nice computer for processing. Most of the PALM data for our paper was taken on a single-core workstation with a single internal hard drive, but we were taking data faster than we could process it. We recently upgraded to a multi-core workstation with a small RAID array, and now we process data faster than we acquire it. It's much nicer to see your PALM image as you take it.