Computerized pattern recognition used for grain counting in high resolution autoradiographs with low grain densities

Area of Science:

• Biomedical Imaging
• Histology
• Radiochemistry

Background:

• Autoradiography is a crucial technique for visualizing and quantifying the distribution of radioactive isotopes in biological tissues.
• Accurate grain counting is essential for interpreting autoradiographic data, but manual methods are time-consuming and prone to error.
• Automated methods are needed to improve the efficiency and reliability of grain quantification.

Purpose of the Study:

• To develop and validate a fast and reliable automated system for counting autoradiographic grains.
• To integrate grain counting with histological image analysis for precise localization of radioactive signals.

Main Methods:

• A video-image system coupled to a minicomputer was utilized.
• Commercial image handling software and a custom FORTRAN program were employed for grain recognition.
• Autoradiographic grains in dark-field were counted automatically.
• Grain counts were overlaid on bright-field histological images for direct correlation.

Main Results:

• The automated system achieved a grain counting speed of approximately 3.5 minutes for 120,000 square microns.
• The reliability of the system was high, with false scores reported at less than 5%.
• The method allowed direct correlation of grain counts with underlying histological structures.

Conclusions:

• The developed FORTRAN program and video-image system provide a fast, reliable, and accurate method for automated autoradiographic grain counting.
• This automated approach enhances the efficiency of quantitative autoradiography and improves the interpretation of histological data.
• The integration with histological images facilitates precise spatial analysis of radioactive labeling in biological samples.