Literature DB >> 27746229

Interpretable deep neural networks for single-trial EEG classification.

Irene Sturm1, Sebastian Lapuschkin2, Wojciech Samek3, Klaus-Robert Müller4.   

Abstract

BACKGROUND: In cognitive neuroscience the potential of deep neural networks (DNNs) for solving complex classification tasks is yet to be fully exploited. The most limiting factor is that DNNs as notorious 'black boxes' do not provide insight into neurophysiological phenomena underlying a decision. Layer-wise relevance propagation (LRP) has been introduced as a novel method to explain individual network decisions. NEW
METHOD: We propose the application of DNNs with LRP for the first time for EEG data analysis. Through LRP the single-trial DNN decisions are transformed into heatmaps indicating each data point's relevance for the outcome of the decision.
RESULTS: DNN achieves classification accuracies comparable to those of CSP-LDA. In subjects with low performance subject-to-subject transfer of trained DNNs can improve the results. The single-trial LRP heatmaps reveal neurophysiologically plausible patterns, resembling CSP-derived scalp maps. Critically, while CSP patterns represent class-wise aggregated information, LRP heatmaps pinpoint neural patterns to single time points in single trials. COMPARISON WITH EXISTING METHOD(S): We compare the classification performance of DNNs to that of linear CSP-LDA on two data sets related to motor-imaginary BCI.
CONCLUSION: We have demonstrated that DNN is a powerful non-linear tool for EEG analysis. With LRP a new quality of high-resolution assessment of neural activity can be reached. LRP is a potential remedy for the lack of interpretability of DNNs that has limited their utility in neuroscientific applications. The extreme specificity of the LRP-derived heatmaps opens up new avenues for investigating neural activity underlying complex perception or decision-related processes.
Copyright © 2016 Elsevier B.V. All rights reserved.

Entities:  

Keywords:  Brain–computer interfacing; Interpretability; Neural networks

Mesh:

Year:  2016        PMID: 27746229     DOI: 10.1016/j.jneumeth.2016.10.008

Source DB:  PubMed          Journal:  J Neurosci Methods        ISSN: 0165-0270            Impact factor:   2.390


  34 in total

1.  Learning Invariant Representations from EEG via Adversarial Inference.

Authors:  Ozan Özdenizci; Y E Wang; Toshiaki Koike-Akino; Deniz ErdoĞmuŞ
Journal:  IEEE Access       Date:  2020-02-04       Impact factor: 3.367

2.  Speech synthesis from ECoG using densely connected 3D convolutional neural networks.

Authors:  Miguel Angrick; Christian Herff; Emily Mugler; Matthew C Tate; Marc W Slutzky; Dean J Krusienski; Tanja Schultz
Journal:  J Neural Eng       Date:  2019-03-04       Impact factor: 5.379

3.  Intelligent inverse treatment planning via deep reinforcement learning, a proof-of-principle study in high dose-rate brachytherapy for cervical cancer.

Authors:  Chenyang Shen; Yesenia Gonzalez; Peter Klages; Nan Qin; Hyunuk Jung; Liyuan Chen; Dan Nguyen; Steve B Jiang; Xun Jia
Journal:  Phys Med Biol       Date:  2019-05-29       Impact factor: 3.609

4.  On Interpretability of Artificial Neural Networks: A Survey.

Authors:  Feng-Lei Fan; Jinjun Xiong; Mengzhou Li; Ge Wang
Journal:  IEEE Trans Radiat Plasma Med Sci       Date:  2021-03-17

5.  Enhanced Performance by Interpretable Low-Frequency Electroencephalogram Oscillations in the Machine Learning-Based Diagnosis of Post-traumatic Stress Disorder.

Authors:  Miseon Shim; Chang-Hwan Im; Seung-Hwan Lee; Han-Jeong Hwang
Journal:  Front Neuroinform       Date:  2022-04-26       Impact factor: 3.739

6.  DeepCOMBI: explainable artificial intelligence for the analysis and discovery in genome-wide association studies.

Authors:  Bettina Mieth; Alexandre Rozier; Juan Antonio Rodriguez; Marina M C Höhne; Nico Görnitz; Klaus-Robert Müller
Journal:  NAR Genom Bioinform       Date:  2021-07-20

Review 7.  Artificial Intelligence in Surgery: Promises and Perils.

Authors:  Daniel A Hashimoto; Guy Rosman; Daniela Rus; Ozanan R Meireles
Journal:  Ann Surg       Date:  2018-07       Impact factor: 12.969

Review 8.  An introduction to deep learning in medical physics: advantages, potential, and challenges.

Authors:  Chenyang Shen; Dan Nguyen; Zhiguo Zhou; Steve B Jiang; Bin Dong; Xun Jia
Journal:  Phys Med Biol       Date:  2020-03-03       Impact factor: 3.609

9.  Operating a treatment planning system using a deep-reinforcement learning-based virtual treatment planner for prostate cancer intensity-modulated radiation therapy treatment planning.

Authors:  Chenyang Shen; Dan Nguyen; Liyuan Chen; Yesenia Gonzalez; Rafe McBeth; Nan Qin; Steve B Jiang; Xun Jia
Journal:  Med Phys       Date:  2020-03-28       Impact factor: 4.071

10.  On the robustness of deep learning-based lung-nodule classification for CT images with respect to image noise.

Authors:  Chenyang Shen; Min-Yu Tsai; Liyuan Chen; Shulong Li; Dan Nguyen; Jing Wang; Steve B Jiang; Xun Jia
Journal:  Phys Med Biol       Date:  2020-12-22       Impact factor: 3.609

View more

北京卡尤迪生物科技股份有限公司 © 2022-2023.