STARTS: A Self-Adapted Spatio-Temporal Framework for Automatic E/MEG Source Imaging.

Area of Science:

• Neuroscience
• Biophysics
• Signal Processing

Background:

• Electro- and magneto-encephalography (E/MEG) source imaging is crucial for understanding brain activity.
• Accurate source localization is challenging due to the ill-posed nature of E/MEG data.
• Existing methods often require complex manual adjustments and may struggle with noise.

Purpose of the Study:

• To introduce STARTS, a novel, automatic spatio-temporal-constrained framework for E/MEG source imaging.
• To enhance the accuracy and neurophysiological plausibility of brain source reconstruction.
• To provide an efficient and effective alternative to current E/MEG source imaging techniques.

Main Methods:

• Developed STARTS, incorporating block-diagonal covariance for spatial homogeneity and temporal basis functions (TBFs) for noise reduction.
• Estimated and updated TBFs in a data-driven manner for improved source localization.
• Validated STARTS using simulations and real epileptic and resting-state EEG data.

Main Results:

• STARTS demonstrated superior performance in simulations compared to benchmark algorithms (LORETA, EBI-Convex, BESTIES, SI-STBF).
• Neurophysiologically plausible results were obtained on epileptic and resting-state EEG data.
• A computationally efficient version, smooth STARTS, achieved comparable performance with reduced execution time.

Conclusions:

• STARTS offers an effective and efficient approach for E/MEG source imaging through advanced spatio-temporal constraints and self-adapted updates.
• The framework automates source reconstruction, improving accuracy and reducing noise influence.
• STARTS provides a valuable tool for advancing neuroscience research using E/MEG data.