Autoradiograph mask analysis in MATLAB

Introduction

Mask analysis is a deconvolution procedure aimed at locating radioactive sources in biological material. The input data are autoradiograms which include biological structures of various types (e.g. mitochondria) as well as developed photographic grains resulting from radioiostopes bound to some of the biological structures. The goal is to find out which structures bind the radioisotopes, and determine the binding density (source intensity).

The radiation spreads from the labeled sources in a characteristic fashion described in the reference below. As a result, grains appear over biological structures which are not sources. Input data to the code given below comes from all the autoradiograms in an experiment and includes:

• The distances of all grains from certain reference structures. The grain distances are further processed into bins.
• The selection of hypothetical sources which are used to generate "model grains" via a monte-carlo process.
• A scattering matrix which relates how radiation from the hypothetical sources ends up in a grain bin.

The Code

Originally the code was a FORTRAN program running on an IBM System/360 using IMSL optimization routines.

The current recoding converts the FORTRAN program to MATLAB to run on Windows-based systems.

Analysis code

Auxillary subroutine for the function minimizer.

Reference

Salpeter, M., McHenry, F., and Salpeter, E.
Resolution in EM autoradiograms. IV. Analysis of autoradiograms,
J. Cell Biol. vol 76 (1978), 127-145
with the appendix:
Basis for computer program for mask analysis of EM autoradiograms.
by Land, B. and Salpeter, E.