Figure 7
Classification results for the STFT images at 50 × 20 pixels, images of the EEG waveform at 40 × 250 pixels and the concatenated images of three temporally separated images of EEG waveforms at 40 × 750 pixels. (a) The inputs were classified with a convolutional neural network (CNN) for 2D input. (b) The false positive (FP) numbers for the STFT images, waveform images at 40 × 250 pixels and waveform images at 40 × 750 pixels. ***p  < 0.001 vs. STFT. (c) False negative (FN) numbers for the STFT images, waveform images at 40 × 250 pixels and waveform images at 40 × 750 pixels. (d) The receiver operating characteristics (ROC) curve of the classification result for the STFT images. (e) The ROC curve of the classification result for the waveform images at 40 × 250 pixels. (f) The ROC curve of the classification result for the waveform images at 40 × 750 pixels. The area under the curve (AUC) is presented for each ROC curve.
