A spatio-spectral approach for movement decoding from cortical and subcortical recordings in patients with Parkinson's disease [article]

Victoria Peterson, Timon Merk, Alan Bush, Vadim Nikulin, Andrea A Kühn, Wolf-Julian Neumann, Robert Mark Richardson
2021 bioRxiv   pre-print
The application of machine learning to intracranial signal analysis has the potential to revolutionize deep brain stimulation (DBS) by personalizing therapy to dynamic brain states, specific to symptoms and behaviors. Machine learning methods can allow behavioral states to be decoded accurately from intracranial local field potentials to trigger an adaptive DBS (aDBS) system, closing the loop between patients' needs and stimulation patterns. Most decoding pipelines for aDBS are based on single
more » ... hannel frequency domain features, neglecting spatial information available in multichannel recordings. Such features are extracted either from DBS lead recordings in the subcortical target and/or from electrocorticography (ECoG). To optimize the simultaneous use of both types of signals, we developed a supervised online-compatible decoding pipeline based on multichannel and multiple recording site recordings. We applied this pipeline to data obtained from 11 patients with Parkinson's disease performing a hand movement task during DBS surgery targeting the subthalamic nucleus, in which in addition a research temporary ECoG electrode was placed. Spectral and spatial features were extracted using filter-bank analysis and spatial pattern decomposition. The learned spatio-spectral features were used to train a generalized linear model with sparse regularized regression. We found that movement decoding was successful using 100 ms time windows, epoch time that is well-suited for aDBS applications. In addition, when 9 out of 16 features were selected, decoding performance was improved up to 15% when the multiple recording site features were used as compared to the single recording site approach. The prediction value was inversely correlated with both the UPDRS score and the distance of the ECoG electrode position to the hand knob motor cortex. Further evaluation of the selected features revealed that ECoG signals contribute more to decoding performance than subthalamic signals. This novel application of spatial filters to decode movement from combined cortical and subcortical recordings is an important step toward the use of machine learning for the construction of intelligent aDBS systems.
doi:10.1101/2021.06.06.447145 fatcat:wlfpxe4slbbofccpm5xojp6xta