Artificial intelligence in digital pathology diagnosis and analysis: technologies, challenges, and future prospects

Area of Science:

• Pathology
• Computational Pathology
• Artificial Intelligence

Background:

• Histopathological images are the gold standard in diagnostics, offering rich morphological and molecular data.
• Artificial intelligence (AI) presents significant potential to enhance disease interpretation in pathology.
• AI-driven computational pathology tools are rapidly developing but require systematic evaluation.

Purpose of the Study:

• To systematically review AI applications across the entire diagnostic continuum in pathology.
• To present a technical taxonomy of AI algorithms and foundation models used in pathology.
• To identify challenges and propose a roadmap for clinical translation of AI in pathology.

Main Methods:

• Systematic evaluation of AI applications in histopathological image analysis.
• Technical taxonomy of AI algorithms and foundation models.
• Benchmarking AI performance across diverse diagnostic tasks.
• Comparative analysis of AI tools for tumor classification, prognostic stratification, and biomarker discovery.

Main Results:

• AI applications span image preprocessing, tumor classification, prognostic stratification, and biomarker discovery.
• A technical taxonomy of AI algorithms and foundation models is presented.
• Performance benchmarks across various diagnostic tasks are provided.
• Critical challenges such as computational scaling, noisy annotations, interpretability, and domain shifts are identified.

Conclusions:

• AI has transformative potential in pathology, revolutionizing disease interpretation.
• Systematic evaluation and benchmarking are crucial for clinical translation.
• Addressing challenges in computational scaling, interpretability, and domain shifts is key.
• A roadmap is proposed to accelerate the integration of AI into diagnostic workflows for precision oncology.