Intracranial Error Detection via Deep Learning

Abstract

Deep learning techniques have revolutionized the field of machine learning and were recently successfully applied to various classification problems in noninvasive electroencephalography (EEG). However, these methods were so far only rarely evaluated for use in intracranial EEG. We employed convolutional neural networks (CNNs) to classify and characterize the error-related brain response as measured in 24 intracranial EEG recordings. Decoding accuracies of CNNs were significantly higher than those of a regularized linear discriminant analysis. Using time-resolved deep decoding, it was possible to classify errors in various regions in the human brain, and further to decode errors over 200 ms before the actual erroneous button press, e.g., in the precentral gyrus. Moreover, deeper networks performed better than shallower networks in distinguishing correct from error trials in all-channel decoding. In single recordings, up to 100 % decoding accuracy was achieved. Visualization of the networks' learned features indicated that multivariate decoding on an ensemble of channels yields related, albeit non-redundant information compared to single-channel decoding. In summary, here we show the usefulness of deep learning for both intracranial error decoding and mapping of the spatio-temporal structure of the human error processing network.

Diffusion property and functional connectivity of superior longitudinal fasciculus underpin human metacognition

“…The IPL might be involved in the information sampling and selection processes (e.g., Baddeley, 2000), and thus the more firstorder memory-related inputs received from the precuneus, the higher and more accurate confidence rating would be. The PrG or a broader presupplmentary motor area (pre-SMA) region, which interconnects with the DLPFC and ACC, might be involved in cognitive control processes (Morales et al, 2018;Völker et al, 2018), and thus when conflicting signals are detected, this region might access more first-order memory-related information from the precuneus to guide a more accurate confidence rating. Therefore, our findings on the relationship of the right IPL -precuneus and right PrG -precuneus rs-FCs with mnemonic metacognitive ability lend support to the hypothesis that mnemonic metacognition relies on the read out of memory trace (Nelson & Narens, 1990).…”

Deep learning-based electroencephalography analysis: a systematic review

“…Their results showed that in most cases, shallower fully convolutional models outperformed their deeper counterpart and architectures with residual connections. However, the authors later found the weight initialization to be critical in successfully training deeper architectures such as ResNet on an intracranial task [210], suggesting hyperparameter tuning might be key to using deeper architectures on neurophysiological data (personal communication, April 17, 2019).…”

Hybrid Brain-Computer-Interfacing for Human-Compliant Robots: Inferring Continuous Subjective Ratings With Deep Regression

“…They additionally can be applied to the raw EEG data, greatly simplifying the design of BCI pipelines. We further demonstrated the usefulness of CNNs for error decoding from noninvasive (Völker et al, 2018c) and intracranial EEG (Völker et al, 2018b).…”