Test-retest reliability of resting-state functional magnetic resonance imaging during deep brain stimulation for...

Patients implanted with modern deep brain stimulation (DBS) hardware can now undergo functional magnetic resonance imaging (fMRI), leading to its increased used to study DBS' mechanisms and predict optimal therapy settings. To accurately interpret fMRI data and realize its clinical potential for DBS, a better understanding of reliability is needed.

Sixteen patients with Parkinson's disease (PD) and DBS targeting the subthalamic nucleus or pallidum underwent 3T test-retest resting-state fMRI with and without concurrent stimulation. Effects of stimulation and device-metal artifacts on reliability of fMRI brain connectivity and moment-to-moment brain variability were explored, plus factors influencing between-subject variations in reliability such as motion.

The brain variability fMRI metric yielded higher intra-class correlation coefficients than the connectivity metric (range across whole brain, motor, limbic, and cognitive networks: 0.36-0.85 and 0.68-0.99, respectively). Average network connectivity appeared less reproducible when DBS was ON versus OFF during fMRI, and fMRI metric reliability for brain areas affected by metal artifacts was significantly higher (brain variability) or lower (connectivity) than unaffected areas (p_uncorrected < 0.05). Motion and DBS target best explained between-subject variations.

DBS hardware and active stimulation may alter fMRI reliability. To develop clinically useful fMRI biomarkers for DBS and aid assessments of reproducibility across studies, the reliability of single study results need reporting.