Figures & data
Figure 3. A typical example of the SSNMF decomposition. These grand-averaged spectrograms were obtained from fifteen adult participants when the 500 sweeps were included in the averaging procedure.
![Figure 3. A typical example of the SSNMF decomposition. These grand-averaged spectrograms were obtained from fifteen adult participants when the 500 sweeps were included in the averaging procedure.](/cms/asset/85efe121-b923-406b-a95c-baeba6ec1e75/iija_a_2071345_f0003_c.jpg)
Figure 4. Application of the SSNMF algorithm on EEG recordings obtained in adult participants. Grand-averaged spectrograms of the input data (A), spectral-basis matrix (B), information-coding matrix (C), enhanced FFR (D), and extracted noise (E).
![Figure 4. Application of the SSNMF algorithm on EEG recordings obtained in adult participants. Grand-averaged spectrograms of the input data (A), spectral-basis matrix (B), information-coding matrix (C), enhanced FFR (D), and extracted noise (E).](/cms/asset/30890fa6-9708-4c5d-9fea-325ad536af91/iija_a_2071345_f0004_c.jpg)
Figure 5. Application of the SSNMF algorithm on EEG recordings obtained in neonatal participants. Grand-averaged spectrograms of the input data (A), spectral-basis matrix (B), information-coding matrix (C), enhanced FFR (D), and extracted noise (E).
![Figure 5. Application of the SSNMF algorithm on EEG recordings obtained in neonatal participants. Grand-averaged spectrograms of the input data (A), spectral-basis matrix (B), information-coding matrix (C), enhanced FFR (D), and extracted noise (E).](/cms/asset/126723eb-e84a-4d60-93dd-d90bb985b265/iija_a_2071345_f0005_c.jpg)
Figure 6. SSNMF performance in adult participants. A. Correlation coefficients before (i.e., the AVG condition) and after (i.e., the AVG + SSNMF condition) the SSNMF decomposition. B. FFR Enhancement as a function of the number of sweeps. C. RMSEs before (i.e., the AVG condition) and after (i.e., AVG + SSNMF condition) the SSNMF decomposition. D. Noise Reduction with increasing number of sweeps. Each shaded area represents the SSNMF performance in terms of FFR Enhancement and Noise Reduction. Δ Correlation = correlation coefficients obtained at the AVG + SSNMF condition – correlation coefficients obtained at the AVG condition. Δ RMSE = RMSE derived at the AVG + SSNMF condition – RMSE derived at the AVG condition. Each error bar indicates one standard error.
![Figure 6. SSNMF performance in adult participants. A. Correlation coefficients before (i.e., the AVG condition) and after (i.e., the AVG + SSNMF condition) the SSNMF decomposition. B. FFR Enhancement as a function of the number of sweeps. C. RMSEs before (i.e., the AVG condition) and after (i.e., AVG + SSNMF condition) the SSNMF decomposition. D. Noise Reduction with increasing number of sweeps. Each shaded area represents the SSNMF performance in terms of FFR Enhancement and Noise Reduction. Δ Correlation = correlation coefficients obtained at the AVG + SSNMF condition – correlation coefficients obtained at the AVG condition. Δ RMSE = RMSE derived at the AVG + SSNMF condition – RMSE derived at the AVG condition. Each error bar indicates one standard error.](/cms/asset/816e6160-d40a-4d00-a0fd-8f6adc9d7fed/iija_a_2071345_f0006_b.jpg)
Figure 7. SSNMF performance in neonatal participants. A. Correlation coefficients before (i.e., the AVG condition) and after (i.e., the AVG + SSNMF condition) the SSNMF decomposition. B. FFR Enhancement as a function of the number of sweeps. C. RMSEs before (i.e., the AVG condition) and after (i.e., AVG + SSNMF condition) the SSNMF decomposition. D. Noise Reduction with increasing number of sweeps. Each shaded area represents the SSNMF performance in terms of FFR Enhancement and Noise Reduction. Δ Correlation = correlation coefficients obtained at the AVG + SSNMF condition – correlation coefficients obtained at the AVG condition. Δ RMSE = RMSE derived at the AVG + SSNMF condition – RMSE derived at the AVG condition. Each error bar represents one standard error.
![Figure 7. SSNMF performance in neonatal participants. A. Correlation coefficients before (i.e., the AVG condition) and after (i.e., the AVG + SSNMF condition) the SSNMF decomposition. B. FFR Enhancement as a function of the number of sweeps. C. RMSEs before (i.e., the AVG condition) and after (i.e., AVG + SSNMF condition) the SSNMF decomposition. D. Noise Reduction with increasing number of sweeps. Each shaded area represents the SSNMF performance in terms of FFR Enhancement and Noise Reduction. Δ Correlation = correlation coefficients obtained at the AVG + SSNMF condition – correlation coefficients obtained at the AVG condition. Δ RMSE = RMSE derived at the AVG + SSNMF condition – RMSE derived at the AVG condition. Each error bar represents one standard error.](/cms/asset/2ea69eb1-2c84-4912-a87d-e1eeee4baaa6/iija_a_2071345_f0007_b.jpg)