Generative AI for brain image computing and brain network computing: a review

Area of Science:

• Neuroimaging
• Artificial Intelligence
• Computational Neuroscience

Background:

• Non-invasive brain imaging techniques have advanced significantly.
• Generative artificial intelligence (AI) has grown substantially, creating new data from existing patterns.
• Integrating generative AI with neuroimaging offers new possibilities for brain research.

Purpose of the Study:

• To review advanced generative AI models, tasks, challenges, and prospects in brain imaging and network computing.
• To provide a comprehensive overview of current generative AI techniques in neuroimaging.
• To focus on novel methodological approaches and applications in the field.

Main Methods:

• Discussed fundamental theories and algorithms of four classic generative models.
• Systematically surveyed and categorized tasks including co-registration, super-resolution, enhancement, classification, segmentation, cross-modality, brain network analysis, and brain decoding.
• Highlighted challenges and future directions in the latest work.

Main Results:

• Identified key generative AI applications in neuroimaging, such as feature extraction and network reconstruction.
• Categorized various tasks where generative AI is applied within brain imaging.
• Provided insights into the current state and potential of generative AI in neuroscience.

Conclusions:

• Generative AI in neuroimaging is a promising field with significant potential for advancing brain mapping and network analysis.
• Further research is needed to address current challenges and explore future directions.
• This review offers a foundational understanding for researchers in generative AI and neuroimaging.