Brain disease research based on functional magnetic resonance imaging data and machine learning: a review

Area of Science:

• Neuroimaging
• Artificial Intelligence
• Neurology

Background:

• Brain diseases pose a significant public health challenge.
• Functional magnetic resonance imaging (fMRI) offers valuable insights into brain activity for diagnosis.
• Machine learning (ML) methods are increasingly effective for brain disease diagnosis.

Purpose of the Study:

• To review recent advancements in ML-based brain disease diagnosis using fMRI data.
• To focus on Alzheimer's disease/mild cognitive impairment, autism spectrum disorders, schizophrenia, and Parkinson's disease.
• To identify future research directions for AI-aided diagnosis.

Main Methods:

• Systematic review of 55 articles published in the last three years.
• Analysis of studies based on subject sample size, feature extraction, selection, classification models, validation, and accuracy.
• Synthesis of findings to highlight trends and challenges.

Main Results:

• Machine learning methods demonstrate superior performance in brain disease diagnosis compared to traditional approaches.
• Key factors influencing diagnostic accuracy include subject sample size, feature engineering, and model selection.
• High accuracies are reported across various brain disease classifications using fMRI data.

Conclusions:

• ML applied to fMRI data holds significant potential for accurate and early diagnosis of brain diseases.
• Further research is needed to optimize feature selection, model generalizability, and clinical integration.
• Interdisciplinary collaboration is crucial for advancing AI-driven neuroimaging diagnostics.