Adaptive Image Deconvolution Algorithm (AIDA)

Welcome and thank you for visiting!  Here, you will find source code, news, and updates on the AIDA package developed mainly by Erik Hom in John Sedat's lab at UCSF, as described in J. Opt. Soc. Am. A 24:1580-1600.

AIDA is a de novo implementation and extension of the MISTRAL myopic deconvolution method developed by Mugnier et al. (2004) [J. Opt. Soc. Am. A 21:1841-1854]. The MISTRAL approach, used thus far to process astronomical images, has been shown to yield object reconstructions with excellent edge preservation and photometric precision. AIDA improves upon the original MISTRAL implementation in the following ways:

  1. AIDA is free and intended for further open source development - please improve it!
  2. AIDA is written in open source Numerical Python instead of the commercial Interactive Data Language (IDL, Research Systems Inc.)
  3. AIDA typically runs at least 15 times faster
  4. AIDA includes a scheme to automatically estimate the 2 hyperparameters used to control noise suppression and edge-preserving regularization in the deconvolution process; this frees the user from quite a bit of ad hoc parameter guessing and in practice, speeds up the production of satisfactory reconstructions by an additional order of magnitude
  5. AIDA can deconvolve multiple frame data and three-dimensional image stacks encountered in adaptive optics and light microscopic imaging.

Platform Requirements: Priithon

AIDA runs on Mac OS 10.4 (including PowerPCs and IntelMacs), Windows 2000 and XP, and Linux (32- or 64-bit machines; AIDA was actually built and tested on Debian).

Importantly, you will need Python 2.4 installed on your system along with "Priithon" -- a free numerical Python platform developed and assembled by Sebastian Haase (Sedat Lab, UCSF).  You may choose to download Priithon for your operating system here, or directly from the links below.  (N.B. These are either .zip or .tbz archives and should be uncompressed using "gunzip [filename.zip]" or "tar -xjvf [filename.tbz]", respectively.)
For a primer on Priithon, check out the Priithon handbook.

Disclaimer

The AIDA package continues to be a work in progress (for better or worse, this code has undergone multiple re-writes/streamlining efforts).  We provide this software as is and cannot guarantee support or assistance.  However, feel free to email bug notices and constructive feedback; we will try to address concerns/requests as we are able (email: aida-feedback/at/freshboom/period/com).

For announcements about updates or additions to the software available here, please subscribe to the AIDA mailing list by sending mail to: aida-subscribe/at/freshboom/period/com.

Click here for news/update information

Download Current Version

In order to download, you will need to agree to a license agreement and provide your name, email address, and snail-mail address.

(N.B. Uncompress tar-bzip2 archive using: "tar -xjvf AIDA1p2p1.tbz"

AIDA 1.2.1    (Please make sure your version of Priithon is current for this)

Download Test Dataset

You can test AIDA using the following dataset: testdataset.tar.bz2

(uncompress using: "tar -xvjf testdataset.tar.bz2")

last modified November 6, 2007

AIDA Development Team (aida-feedback/at/freshboom/period/com)
Priithon, Numerical Python platform (http://www.ucsf.edu/sedat/Priithon)
Sedat Lab, UCSF (http://www.ucsf.edu/sedat)
Macromolecular Structure Group, UCSF (http://www.msg.edu)