As the global population ages, cognitive impairment-particularly mild cognitive impairment (MCI)has emerged as a pressing public health challenge. MCI, a transitional stage between normal aging and dementia, is characterized by subtle declines in cognitive functions such as memory and language, yet these deficits do not meet the diagnostic criteria for dementia. Structural magnetic resonance imaging (sMRI), a non-invasive neuroimaging technique, provides detailed brain anatomical information, making it a valuable tool for studying the relationship between brain structural changes and cognitive dysfunction. This review systematically examines current sMRI-based MCI diagnostic methods, including voxel-based morphometry (VBM), region of interest (ROI)-based analysis, and machine learning/deep learning approaches. Each method presents unique advantages and challenges. VBM enables automated whole-brain analysis but may miss subtle changes in complex regions. ROI analysis offers targeted insights but is subjective and narrow in focus. Machine learning and deep learning methods, despite their powerful feature extraction capabilities, require large public datasets and may suffer from black-box issues. Future research should prioritize optimizing deep learning models, integrating multi-modal and multi-omics data, and establishing multi-center, large-sample MRI databases. These advancements are expected to significantly contribute to the early diagnosis, intervention, and treatment of MCI, addressing the growing challenge of cognitive impairments worldwide.
Wenzhi Li (Wed,) studied this question.