Search

This study decodes consumers' preference levels using a convolutional neural network (CNN) in neuromarketing. The classification accuracy in neuromarketing is a critical factor in evaluating the intentions of the consumers. Functional near-infrared spectroscopy (fNIRS) is utilized as a neuroimaging modality to measure the cerebral hemodynamic responses. In this study, a specific decoding structure, called CNN-based fNIRS-data analysis, was designed to achieve a high classification accuracy. Compared to other methods, the automated characteristics, constant training of the dataset, and learning efficiency of the proposed method are the main advantages. The experimental procedure required eight healthy participants (four female and four male) to view commercial advertisement videos of different durations (15, 30, and 60 s). The cerebral hemodynamic responses of the participants were measured. To compare the preference classification performances, CNN was utilized to extract the most common features, including the mean, peak, variance, kurtosis, and skewness. Considering three video durations, the average classification accuracies of 15, 30, and 60 s videos were 84.3, 87.9, and 86.4%, respectively. Among them, the classification accuracy of 87.9% for 30 s videos was the highest. The average classification accuracies of three preferences in females and males were 86.2 and 86.3%, respectively, showing no difference in each group. By comparing the classification performances in three different combinations (like vs. so-so, like vs. dislike, and so-so vs. dislike) between two groups, male participants were observed to have targeted preferences for commercial advertising, and the classification performance 88.4% between “like” vs. “dislike” out of three categories was the highest. Finally, pairwise classification performance are shown as follows: For female, 86.1% (like vs. so-so), 87.4% (like vs. dislike), 85.2% (so-so vs. dislike), and for male 85.7, 88.4, 85.1%, respectively.

Mild cognitive impairment (MCI) is an intermediate stage leading to Alzheimer's disease (AD). Diagnosis for MCI patients at an early stage can reduce the chances of developing into a severe condition for cognition. This study aims to identify the healthy control (HC) and MCI through the neural images in the specific time points during the mental tasks by the convolutional neural network (CNN). The signals were acquired by the functional near-infrared spectroscopy (fNIRS). 15 MCI patients and 9 HC subjects are employed in the experiment to perform the N-back task, Stroop task, and verbal fluency task (VFT), respectively. The neural images were generated by the brain map in the specific time points (i.e., 5 sec, 10 sec, 15 sec, 20 sec, 25 sec, 30 sec, 35 sec, 40 sec, 45 sec, 50 sec, 55 sec, 60 sec, and 65 sec). Four layers CNN were applied to classify the neural images of the different time points for three mental tasks (i.e., N-back, Stroop, and VFT). For evaluating the performance of the classifier, we utilized a 5-fold cross-validation method. The CNN results indicated that all the mental tasks obtained a great performance (i.e., averaged accuracy of N-back: 82.59%, Stroop: 85.03%, and VFT: 82.20%). Especially, the highest accuracy of the Stroop task in the 60-sec time point is 98.57%. Thus, these findings demonstrate that the neural images can be useful for the identification of the MCI. The fNIRS could be a next promising non-invasive neural imaging tool for early detection of AD in the clinical field.

One of the advantages of fNIRS systems is their portability for real-world applications. Therefore, the online rejection of the motion artifacts in functional near-infrared spectroscopy (fNIRS) signals is one of the essential research topics. We proposed a square-root cubature Kalman filter using a weighted moving average model to tract the dynamics of motion artifacts. The rationale and feasibility of the method were discussed using simulated data generated using a Balloon model. The results show that the proposed filter has a good suppression of spike-like motion artifacts. Finally, conclusions and further study are discussed.

Mild cognitive impairment (MCI) is the clinical precursor of Alzheimer’s disease (AD), which is considered the most common neurodegenerative disease in the elderly. Some MCI patients tend to remain stable over time and do not evolve to AD. It is essential to diagnose MCI in its early stages and provide timely treatment to the patient. In this study, we propose a neuroimaging approach to identify MCI using a deep learning method and functional near-infrared spectroscopy (fNIRS). For this purpose, fifteen MCI subjects and nine healthy controls (HCs) were asked to perform three mental tasks: N-back, Stroop, and verbal fluency (VF) tasks. Besides examining the oxygenated hemoglobin changes (ΔHbO) in the region of interest, ΔHbO maps at thirteen specific time points (i.e., 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, and 65 s) during the tasks and seven temporal feature maps (i.e., two types of mean, three types of slope, kurtosis, and skewness) in the prefrontal cortex were investigated. A four-layer convolutional neural network (CNN) was applied to identify the subjects into either MCI or HC, individually, after training the CNN model with ΔHbO maps and temporal feature maps above. Finally, we used the 5-fold cross-validation approach to evaluate the performance of the CNN. The results of temporal feature maps exhibited high classification accuracies: The average accuracies for the N-back task, Stroop task, and VFT, respectively, were 89.46%, 87.80%, and 90.37%. Notably, the highest accuracy of 98.61% was achieved from the ΔHbO slope map during 20 s ~ 60 s interval of N-back tasks. Our results indicate that the fNIRS imaging approach based on temporal feature maps is a promising diagnostic method for early detection of MCI and can be used as a tool for clinical doctors to identify MCI from their patients.

Functional near-infrared spectroscopy (fNIRS) is an emerging technique for the non-invasive brain-computer interface (BCI). Quickly obtaining precise brain signals is very crucial for successful BCIs. This paper investigates a real-time filtering technique to remove motion artifact (MA) and low-frequency drift in the fNIRS signals. Optical intensities of two wavelengths are generated using a balloon model in the literature and an experimental paradigm. Two types of MAs (spike-like and step-like) and low-frequency drifts are generated and added to the simulated optical intensities of two wavelengths. A new dual-stage median filter (DSMF) is proposed to recover the uncontaminated signals. Five evaluation metrics are used to determine the best window sizes of the dual filters: 4 s and 9 s for the first and 18 s for the second median filter. The proposed method is compared with a wavelet-based MA correction method and spline interpolation method using the same metrics. The results show that the proposed method outperforms the compared methods in attenuating MAs and signal distortion. Finally, the designed DSMF is applied to experimental data from eight healthy subjects, in which MAs were introduced by asking the subjects to shake their heads. The filtered data of the proposed method demonstrates clean signals with no MAs and low-frequency drifts.

The removal of the physiological noises and the external noises such as motion artifacts is one of the key issues in functional near-infrared spectroscopy (fNIRS) studies. In this paper, we investigate a dual extended Kalman filter (DEKF) for the data from fNIRS, which consists a state estimator for hemodynamics responses and a parameter estimator for path-lengths of individual channels simultaneously. Eight subjects participated in two experimental sessions of mental arithmetic tasks. The obtained signals are processed by the correlation based signal improvement (CBSI) method and the newly proposed method, respectively. We evaluate accuracy of the hemodynamic response estimation using the correlation between the CBSI estimation and the DEKF estimation while that of the path-lengths using a left-tail $t$ -test on the relative errors between sessions. The noise-resistant performance is validated using the metrics of correlation coefficient analysis and within-subject standard deviation (SD). The results show that the DEKF-converted concentrations strongly correlate with that of CBSI and the estimated path-lengths are statistically identical in both sessions. The DEKF method is as resistant to noise as CBSI.

The main objective of this study is to propose a shared control strategy for improving the performance and stability of brain-computer interface (BCI)-based quadcopter movement control. With the advancement of brain imaging devices that are inexpensive, non-invasive, and portable, BCI has recently been implemented in various control applications, for example, wheelchair, telepresence robot, robotic manipulator arm, and quadcopter can be controlled using BCI-based motor imagery task or mental arithmetic task. However, using BCI modalities is still difficult to control a movement of the abovementioned machines with high stability and accuracy. Shared control is a technique that combines inputs and environment information of the system, and then, these inputs will be processed to achieve high efficient output. In this paper, the concept of the shared control will be proposed in three different methods: distance measurement sensor, eye movement detection and Global Positioning System (GPS) for avoiding an obstacle of BCI-based quadcopter control.

Nowadays a quadcopter is used in various applications, such as data collection, news broadcasting, surveying in hazardous areas, quadcopter racing competition, etc. Recently, a quadcopter movement control method has been implemented for people with disabilities by using a brain-computer interface (BCI). However, there are some limitations for BCI-based quadcopter movement control, for example, long response time in generating command signals, and unpredictable noise in human brain signals. In this paper, an image processing technique is proposed for quadcopter movement control. By detecting eye movements (i.e. forwarding gaze, left gaze, and right gaze), the image processing algorithm was implemented to distinguish the different states of eye movements and transform them into the quadcopter movement commands. This implemented technique aims to control a quadcopter without controlling by hand. Furthermore, the detection of eye movements for people who have ptosis or falling of the upper eyelid can be achieved by using the Circular Hough Transform algorithm.