Psychiatric disorders are characterized by intricate changes in neural activity, making accurate detection crucial for effective treatment. Traditional methods often struggle with the scarcity of fMRI data and the challenges posed by the diverse types of information these scans provide. To address these issues, the proposed Frequency-Enhanced Network (FENet) adopts a unique approach by incorporating both time-domain and frequency-domain insights into the analysis of fMRI data. This dual-focus approach enhances the framework’s ability to identify and interpret patterns relevant to psychiatric disorders, particularly when data is limited.
FENet constructs multi-view brain networks that leverage the natural properties of fMRI, thereby allowing for a more comprehensive understanding of the underlying neural mechanisms involved in psychiatric disorders. By introducing domain-specific encoders, FENet captures crucial temporal-spectral characteristics, focusing on high-frequency information that has been shown to correlate significantly with various psychiatric conditions. This emphasis on frequency information is pivotal, as it allows for improved precision in disorder detection and highlights previously overlooked aspects of fMRI data, ultimately contributing to more effective diagnostic processes.
👉 Pročitaj original: arXiv AI Papers
