Content Generation Models in Computational Pathology: A Comprehensive Survey on Methods, Applications, and...

Area of Science:

• Computational Pathology
• Artificial Intelligence in Medicine

Background:

• Content generation modeling is a rapidly advancing field in computational pathology.
• It offers significant potential for data-efficient learning, synthetic data augmentation, and task-oriented generation.

Purpose of the Study:

• To provide a comprehensive review of recent progress in content generation for computational pathology.
• To synthesize key developments, datasets, evaluation protocols, and limitations in the field.

Main Methods:

• Systematic analysis of over 150 representative studies in content generation for computational pathology.
• Categorization of research into image generation, text generation, molecular profile-morphology generation, and specialized applications.

Main Results:

• Evolution of architectures from Generative Adversarial Networks (GANs) to diffusion models and vision-language models.
• Identification of common datasets and evaluation metrics.
• Highlighting limitations such as high-fidelity whole slide image generation and clinical interpretability.

Conclusions:

• Content generation holds promise for computational pathology but faces challenges in fidelity, interpretability, and ethical considerations.
• Future research should focus on integrated, clinically deployable generation systems.