NIRS Data Augmentation Technique to Detect Hemodynamic Peaks During Self-Paced Motor Imagery.

Optical brain monitoring, such as near-infrared spectroscopy (NIRS), has facilitated numerous brain studies, including those based on machine learning techniques. A large and diverse dataset is necessary for training machine learning algorithms to avoid overfitting a limited amount of data. However, recruiting sufficient subjects is challenging owing to time and budget constraints. Therefore, we propose an NIRS data generation algorithm that scales NIRS signal components, such as hemodynamic response function, physiological systemic noise, and instrumental spike noise, based on the source-detector distance to augment the training data. Experimental self-paced left- and right-hand motor imagery data were augmented with generated NIRS data to train a convolutional neural network and classify the motor imagery data. Augmenting the training dataset with 1000 generated data increased the classification accuracy to 86.3 +/- 4.1%, a 26% increase compared with training on experimental data only. In addition, we applied Guided Gradientweighted Class Activation Mapping (Grad-CAM) to visualize the class discriminative features of the input data. The peaks of Guided Grad-CAM heatmaps aligned with the oxy-hemoglobin peaks during selfpaced motor imagery. We concluded that the increased cerebral oxygenation, especially in the contralateral hemisphere, was the class-discriminative feature for classifying left- and right-hand motor imagery.