An Integrated Multi-Channel Deep Neural Network for Mesial Temporal Lobe Epilepsy Identification Using Multi-Modal Medical Data.

Epilepsy is a chronic brain disease with recurrent seizures. Mesial temporal lobe epilepsy (MTLE) is the most common pathological cause of epilepsy. With the development of computer-aided diagnosis technology, there are many auxiliary diagnostic approaches based on deep learning algorithms. However, the causes of epilepsy are complex, and distinguishing different types of epilepsy accurately is challenging with a single mode of examination. In this study, our aim is to assess the combination of multi-modal epilepsy medical information from structural MRI, PET image, typical clinical symptoms and personal demographic and cognitive data (PDC) by adopting a multi-channel 3D deep convolutional neural network and pre-training PET images. The results show better diagnosis accuracy than using one single type of medical data alone. These findings reveal the potential of a deep neural network in multi-modal medical data fusion.

[A research on epilepsy source localization from scalp electroencephalograph based on patient-specific head model and multi-dipole model].

Accurate source localization of the epileptogenic zone (EZ) is the primary condition of surgical removal of EZ. The traditional localization results based on three-dimensional ball model or standard head model may cause errors. This study intended to localize the EZ by using the patient-specific head model and multi-dipole algorithms using spikes during sleep. Then the current density distribution on the cortex was computed and used to construct the phase transfer entropy functional connectivity network between different brain areas to obtain the localization of EZ. The experiment result showed that our improved methods could reach the accuracy of 89.27% and the number of implanted electrodes could be reduced by (19.34 ± 7.15)%. This work can not only improve the accuracy of EZ localization, but also reduce the additional injury and potential risk caused by preoperative examination and surgical operation, and provide a more intuitive and effective reference for neurosurgeons to make surgical plans.

Electroencephalogram-Based Brain Connectivity Analysis in Prolonged Disorders of Consciousness.

Background: Prolonged disorders of consciousness (pDOC) are common in neurology and place a heavy burden on families and society. This study is aimed at investigating the characteristics of brain connectivity in patients with pDOC based on quantitative EEG (qEEG) and extending a new direction for the evaluation of pDOC. Methods: Participants were divided into a control group (CG) and a DOC group by the presence or absence of pDOC. Participants underwent magnetic resonance imaging (MRI) T1 three-dimensional magnetization with a prepared rapid acquisition gradient echo (3D-T1-MPRAGE) sequence, and video EEG data were collected. After calculating the power spectrum by EEG data analysis tool, DTABR ((δ + θ)/(α + ß) ratio), Pearson's correlation coefficient (Pearson r), Granger's causality, and phase transfer entropy (PTE), we performed statistical analysis between two groups. Finally, receiver operating characteristic (ROC) curves of connectivity metrics were made. Results: The proportion of power in frontal, central, parietal, and temporal regions in the DOC group was lower than that in the CG. The percentage of delta power in the DOC group was significantly higher than that in the CG, the DTABR in the DOC group was higher than that in the CG, and the value was inverted. The Pearson r of the DOC group was higher than that of CG. The Pearson r of the delta band (Z = -6.71, P < 0.01), theta band (Z = -15.06, P < 0.01), and alpha band (Z = -28.45, P < 0.01) were statistically significant. Granger causality showed that the intensity of directed connections between the two hemispheres in the DOC group at the same threshold was significantly reduced (Z = -82.43, P < 0.01). The PTE of each frequency band in the DOC group was lower than that in the CG. The PTE of the delta band (Z = -42.68, P < 0.01), theta band (Z = -56.79, P < 0.01), the alpha band (Z = -35.11, P < 0.01), and beta band (Z = -63.74, P < 0.01) had statistical significance. Conclusion: Brain connectivity analysis based on EEG has the advantages of being noninvasive, convenient, and bedside. The Pearson r of DTABR, delta, theta, and alpha bands, Granger's causality, and PTE of the delta, theta, alpha, and beta bands can be used as biological markers to distinguish between pDOC and healthy people, especially when behavior evaluation is difficult or ambiguous; it can supplement clinical diagnosis.

Automatic food detection in egocentric images using artificial intelligence technology.

OBJECTIVE: To develop an artificial intelligence (AI)-based algorithm which can automatically detect food items from images acquired by an egocentric wearable camera for dietary assessment. DESIGN: To study human diet and lifestyle, large sets of egocentric images were acquired using a wearable device, called eButton, from free-living individuals. Three thousand nine hundred images containing real-world activities, which formed eButton data set 1, were manually selected from thirty subjects. eButton data set 2 contained 29 515 images acquired from a research participant in a week-long unrestricted recording. They included both food- and non-food-related real-life activities, such as dining at both home and restaurants, cooking, shopping, gardening, housekeeping chores, taking classes, gym exercise, etc. All images in these data sets were classified as food/non-food images based on their tags generated by a convolutional neural network. RESULTS: A cross data-set test was conducted on eButton data set 1. The overall accuracy of food detection was 91·5 and 86·4 %, respectively, when one-half of data set 1 was used for training and the other half for testing. For eButton data set 2, 74·0 % sensitivity and 87·0 % specificity were obtained if both 'food' and 'drink' were considered as food images. Alternatively, if only 'food' items were considered, the sensitivity and specificity reached 85·0 and 85·8 %, respectively. CONCLUSIONS: The AI technology can automatically detect foods from low-quality, wearable camera-acquired real-world egocentric images with reasonable accuracy, reducing both the burden of data processing and privacy concerns.