Functional connectivity: Definition and signal origins

Since its introduction in the 1990s, functional magnetic resonance imaging (fMRI) has evolved into a popular neuroscience research method, with applications spanning cognitive neuroscience, clinical diagnosis, and brain development. In this review, we explore both task-based and resting-state fMRI (RS-fMRI), the physiological basis of blood-oxygenation-level-dependent signals, and the analysis methods used to investigate brain connectivity. Resting-state fMRI allows the study of intrinsic brain activity by identifying several RS networks including the default mode network sensorimotor, visual, auditory, and salience networks. Analytical approaches include seed-based correlation, independent and principal component analysis, clustering, and graph theory. These methods reveal both global and localized patterns of functional connectivity. Voxel-wise measures such as ALFF, fALFF, and ReHo further enable detailed analysis of local signal coherence. In clinical settings, fMRI is used in presurgical mapping, diagnosis of neuropsychiatric and neurodegenerative disorders, and evaluation of treatment outcomes. It has proven valuable in understanding epilepsy, schizophrenia, Alzheimer's disease, depression, and attention-deficit hyperactivity disorder. Multimodal approaches that combine fMRI with electroencephalography, positron emission tomography, fNIRS, and magnetoencephalography offer complementary insights by enhancing spatial and temporal resolution. However, challenges remain, including variability in acquisition protocols, motion artifacts, and the need for standardized preprocessing pipelines. Recent advances in machine learning and virtual reality integration promise to expand fMRI's research and clinical utility by improving pattern recognition and simulating real-world scenarios. In summary, fMRI continues to revolutionize neuroscience by enabling a deeper understanding of brain structure, function, and connectivity in both health and disease.