Automatic section thickness determination using an absolute gradient focus function

Area of Science:

• Bioscience research
• Neuroscience
• Microscopy

Background:

• Computerized stereology relies on accurate quantitative analysis of microstructures.
• Current methods for section thickness determination are manual and time-consuming.
• Autofocus functions typically use global maximums, which are insufficient for precise tissue boundary detection.

Purpose of the Study:

• To develop a novel, automated algorithm for precise section thickness determination.
• To improve the accuracy and efficiency of computerized stereology.
• To enable fully automatic computerized stereology by addressing section thickness measurement.

Main Methods:

• Analysis of 14 grey-scale focus functions to identify optimal image processing techniques.
• Utilizing two sharp 'knees' on the focus curve to detect transitions between unfocused and focused planes.
• Developing and testing novel functions, including the 'modified absolute gradient count' function.

Main Results:

• The 'modified absolute gradient count' function demonstrated superior performance.
• Achieved an average error of 0.56 μm on a similar test set and 0.39 μm on a diverse test set (different staining, brain regions, subjects).
• The algorithm accurately determines section thickness using out-of-focus planes.

Conclusions:

• A novel algorithm for automatic section thickness determination has been developed.
• This method significantly enhances the accuracy of computerized stereology.
• The algorithm is a critical prerequisite for fully automated stereological analysis in biosciences.