Statistical evaluation of methods for quantifying gene expression by autoradiography in histological sections

Area of Science:

• Molecular Biology
• Histology
• Biotechnology

Background:

• In situ hybridisation (ISH) combined with autoradiography is a standard technique for measuring gene expression in tissue sections.
• Current quantification methods for ISH images vary significantly, potentially yielding conflicting results.
• Existing image analysis approaches may not be optimal for accurate gene expression measurement.

Purpose of the Study:

• To evaluate commonly used methods for analysing in situ hybridisation (ISH) images.
• To identify limitations and biases in current gene expression quantification techniques.
• To propose an improved method for analysing ISH images to enhance data accuracy.

Main Methods:

• Analysis of standard image segmentation techniques, including thresholding.
• Evaluation of the impact of region of interest selection on precision and bias.
• Comparison of different methods for converting pixel intensities.

Main Results:

• Image segmentation using thresholding can introduce floor-effects and bias.
• Including the area of interest in calculations leads to reduced precision and increased bias.
• Converting pixel intensities to optical densities or radioactivity units is often unnecessary and can degrade statistical properties.
• A modified method for region of interest selection demonstrates improved bias and precision.

Conclusions:

• Commonly employed methods for analysing ISH images are suboptimal and can introduce bias.
• Optimized methods for image segmentation and region of interest selection are crucial for accurate gene expression analysis.
• The study provides recommendations to reduce bias and increase precision in gene expression data derived from ISH images.