Out-of-focus brain image detection in serial tissue sections

Area of Science:

• Neuroimaging
• Computational Biology
• Image Analysis

Background:

• Visual inspection of brain imaging data is crucial for quality control but is time-consuming.
• Classifying images as in-focus or out-of-focus (OOF) is a particularly laborious step in the quality control workflow.

Purpose of the Study:

• To develop an automated method for identifying OOF brain images within large datasets of serial tissue sections.
• To enhance the efficiency and accuracy of quality control in brain imaging analysis pipelines.

Main Methods:

• Utilized steerable filters (STF) to compute a focus value (FV) for each image.
• Implemented an outlier detection algorithm with a dynamic threshold for focus classification.
• Applied the method to large-scale datasets exceeding 1.5 petabytes.

Main Results:

• The automated method demonstrated high accuracy in identifying OOF images compared to manual visual inspection.
• Achieved a minimal number of false positives, indicating robust performance.
• Successfully validated the algorithm against existing methods and on simulated OOF image datasets.

Conclusions:

• Presented a practical and automated solution for distinguishing OOF images in large serial tissue section datasets.
• The method is suitable for integration into automated pre-processing pipelines for brain imaging analysis.
• Offers a scalable approach to improve the quality control of extensive neuroimaging data.