So I just posted my first dataset on figshare. This is really cool. Figshare that is. Well the simulation too. Even though it is simple. It is a simulation that reproduces a bead image that Kurt Thorn of UCSF shared with me. The actual image is here. And the deconvolved of the actual image here. I did the simulation as an additional check of the deconvolution results.
Below is the real bead image and the simulated bead image. At first glance they look almost identical. But an expert looking closely at the data could probably tell which one was simulated. And since the data is on figshare they could do this.
Below are the deconvolution results.
And below are some line profiles of the original images taken through the bead. When we look at the line profiles we see one of them has a larger FWHM (it is the real image). So the simulation isn’t perfect. Since the data is on figshare someone could look it over and figure out why the real-image FWHM is larger.
Here is the line profile from the deconvolved real image
And the deconvolved simulated image
And finally a line profile through the original “object”, that is the ground truth of the simulation. The phantom had an intensity of 200000. The reconstruction is pretty close. It may not seem like a big deal to reconstruct a simulated bead. But there is are so many things that can go wrong in deconvolution, simple simulations and sanity checks are important.
Anyways figshare is a terrific tool. It makes it really easy to share thoughts on data. I suspect as I get better at using figshare I won’t have to actually embed figures in my blog entries. Instead I can just share links to figshare or a similar service.