Introduction to Artificial Intelligence and Machine Learning for Pathology.

Area of Science:

• Pathology
• Medical Informatics
• Artificial Intelligence

Background:

• Machine learning (ML) is rapidly advancing, offering potential to enhance or replace human expert functions in various fields.
• In pathology, ML may revolutionize image analysis, interpretation, and patient outcomes prediction.
• Pathologists currently lack familiarity with ML principles, necessitating education for effective adoption and management of these new technologies.

Purpose of the Study:

• Provide pathologists with foundational knowledge of ML algorithms, development, and evaluation.
• Explore the current applications of ML in pathology and define the future roles and responsibilities of pathologists.
• Identify challenges and regulatory gaps in the deployment and management of ML systems in pathology.

Main Methods:

• Literature review encompassing biomedical, engineering, and professional organization sources.
• Inclusion of accessible references for pathologists without specialized statistical or software development backgrounds.
• Synthesis of authors' experience in machine learning applications.

Main Results:

• ML techniques demonstrate significant potential in recent pathology studies.
• Collaborative models where human experts work with ML tools show superior performance compared to either alone.
• Optimal integration of ML in pathology workflows is yet to be determined.

Conclusions:

• ML tools offer promising advancements for pathology practice.
• Further research is crucial to establish best practices for ML deployment, including generalizability, local verification, and performance monitoring.
• Development of a robust regulatory framework is needed for widespread and safe implementation of ML in pathology.