Basic Science and Pathogenesis.

Area of Science:

• Neuroscience
• Pathology
• Citizen Science

Background:

• Research-grade histopathological annotations from whole slide images (WSIs) require high data quality beyond current automated methods.
• Citizen science engages public volunteers as research collaborators, offering a scalable way to obtain expert-like histopathological evaluations.
• This approach can enhance community outreach and education, particularly for populations disproportionately affected by Alzheimer's disease, like the Hispanic community.

Purpose of the Study:

• To pilot human computation methods engaging public volunteers in annotating amyloid pathologies in human brain WSIs.
• To compare a two-stage crowd-based analysis using "wisdom of crowd" methods against expert-generated gold standard data for specific pathologies.

Main Methods:

• A crowd-powered analytic pipeline broke down histopathological analysis into manageable online tasks for non-expert volunteers.
• "Wisdom of crowd" methods aggregated multiple individual annotations into a single expert-like annotation for each pathological instance.
• The study compared crowd-based analysis to gold standard data for cored plaques and cerebral amyloid angiopathy.

Main Results:

• Methodological development, including annotation interfaces and consensus methods, was tested.
• Sufficient aggregation of non-expert answers achieved agreement with gold standard data, meeting or exceeding expert interrater agreement.
• This demonstrates the feasibility of using citizen science for histopathological annotation.

Conclusions:

• The developed approaches facilitate rapid, high-volume annotation dataset production for machine learning in neurodegenerative disease research.
• Pilot results support the development of an online platform for de novo analysis of human brain histopathology by volunteers.
• The platform will include culturally responsive outreach for impacted communities, such as Hispanic populations, to foster researcher engagement.