Neural Noise- A Nature Neuroscience article (online 18 March 2026; DOI 10.1038/s41593-026-02229-9) reports longitudinal, long-duration precision fMRI in people receiving deep brain stimulation (DBS).1
- Stimulation-related activity patterns dissociate globus pallidus circuits from primary motor cortex (M1) circuits, with distinct frequency-dependent and time-dependent responses.1
- DBS drives divergent M1 functional connectivity changes across networks: within a somato-cognitive action network, M1 connectivity shifts toward a more normal pattern, while in effector motor networks M1 connectivity moves away from normal (“denormalized”).1
- This constitutes the world’s largest longitudinal DBS-fMRI dataset to date: 14 patients with Parkinson’s disease underwent MRI-compatible DBS across five timepoints spanning one year, each session averaging 11.7 hours of fMRI under seven stimulation conditions (30–172 min per session), totaling >200 hours of high-quality imaging data.2
- The study used 3-T MRI-compatible DBS systems and a deep learning generative model (Yan et al., Nature Communications, 2020) to reconstruct BOLD signals compromised by DBS implant artifacts, enabling precision imaging that would otherwise be unfeasible.1
- Stimulation conditions included variable-frequency DBS alternating between 130 Hz and 60 Hz; 130 Hz is the standard Parkinson’s DBS setting, while variable-frequency protocols target freezing of gait, a symptom often unresponsive to conventional high-frequency stimulation.1
- Clinical benefit from DBS tracked with normalization of connectivity in the somato-cognitive action network (SCAN)—a cortico-subcortical circuit for whole-body action planning, coordination, and interoception—building on a companion Nature 2026 paper (Ren et al., Nature, DOI: 10.1038/s41586-025-10059-1) that defined the SPARK (SCAN-Parkinson’s) circuit as the pathological origin of PD.2
- The SCAN’s hyperconnectivity is a key feature of Parkinson’s disease; DBS therapy exerts its effects primarily by remodeling abnormal functional connectivity within this SPARK circuit rather than simply suppressing subthalamic activity.2
- The divergent effect on effector motor networks (denormalization of M1 within canonical motor circuits) may explain residual motor side-effects and non-responsiveness of certain symptoms despite clinical DBS benefit, providing a mechanistic framework for individualized stimulation parameter optimization.1
- Corresponding institution: a Beijing-led team; this is the precision neuroimaging arm of the same group that published the broader SPARK-PD circuit paper; findings were also publicized by Galaxy Brain Scientific as a basis for non-invasive personalized treatment strategies targeting the SPARK circuit.2
- Clinically, the dataset establishes a circuit-level framework that could guide closed-loop DBS designs—adjusting stimulation frequency based on real-time SCAN/SPARK connectivity states rather than fixed parameters.1