Dynamical graph neural network with attention mechanism for epilepsy detection using single channel EEG.

Epilepsy is a chronic brain disease, and identifying seizures based on electroencephalogram (EEG) signals would be conducive to implement interventions to help patients reduce impairment and improve quality of life. In this paper, we propose a classification algorithm to apply dynamical graph neural network with attention mechanism to single channel EEG signals. Empirical mode decomposition (EMD) are adopted to construct graphs and the optimal adjacency matrix is obtained by model optimization. A multilayer dynamic graph neural network with attention mechanism is proposed to learn more discriminative graph features. The MLP-pooling structure is proposed to fuse graph features. We performed 12 classification tasks on the epileptic EEG database of the University of Bonn, and experimental results showed that using 25 runs of ten-fold cross-validation produced the best classification results with an average of 99.83[Formula: see text] accuracy, 99.91[Formula: see text] specificity, 99.78[Formula: see text] sensitivity, 99.87[Formula: see text] precision, and 99.47[Formula: see text] [Formula: see text] score for the 12 classification tasks.

Two-photon excited luminescence of structural light enhancement in subwavelength SiO2 coating europium ion-doped paramagnetic gadolinium oxide nanoparticle and application for magnetic resonance imaging.

BACKGROUND: Oxides of lanthanide rare-earth elements show great potential in the fields of imaging and therapeutics due to their unique electrical, optical and magnetic properties. Oxides of lanthanide-based nanoparticles enable high-resolution imaging of biological tissues by magnetic resonance imaging (MRI), computed tomography (CT) imaging, and fluorescence imaging. In addition, they can be used to detect, treat, and regulate diseases by fine-tuning their structure and function. It remains challenging to achieve safer, efficient, and more sensitive nanoparticles for clinical applications through the structural design of functional and nanostructured rare-earth materials. RESULT: In this study, we designed a mesoporous silica-coated core-shell structure of europium oxide ions to obtain near-infrared two-photon excitation fluorescence while maintaining high contrast and resolution in MRI. We designed enhanced 800 nm photoexcitation nanostructures, which were simulated by the finite-difference method (FDM) and finite-difference time-domain method (FDTD). The nanoparticle structure, two-photon absorption, up-conversion fluorescence, magnetic properties, cytotoxicity, and MRI were investigated in vivo and in vitro. The nanoparticle has an extremely strong optical fluorescence response and multiple excitation peaks in the visible light band under the 405 nm continuous-wave laser excitation. The nanoparticle was found to possess typical optical nonlinearity induced by two-photon absorption by ultrafast laser Z-scan technique. Two-photon excited fluorescence of visible red light at wavelengths of 615 nm and 701 nm, respectively, under excitation of the more biocompatible near-infrared (pulsed laser at 800 nm). In an in vitro MRI study, a T1 relaxation rate of 6.24 mM-1 s-1 was observed. MRI in vivo showed that the nanoparticles could significantly enhance the signal intensity in liver tissue. CONCLUSIONS: These results suggest that this sample has applied potential in visible light fluorescence imaging and MRI.

Construction of Exosome SORL1 Detection Platform Based on 3D Porous Microfluidic Chip and its Application in Early Diagnosis of Colorectal Cancer.

Exosomes are promising new biomarkers for colorectal cancer (CRC) diagnosis, due to their rich biological fingerprints and high level of stability. However, the accurate detection of exosomes with specific surface receptors is limited to clinical application. Herein, an exosome enrichment platform on a 3D porous sponge microfluidic chip is constructed and the exosome capture efficiency of this chip is ≈90%. Also, deep mass spectrometry analysis followed by multi-level expression screenings revealed a CRC-specific exosome membrane protein (SORL1). A method of SORL1 detection by specific quantum dot labeling is further designed and the ensemble classification system is established by extracting features from 64-patched fluorescence images. Importantly, the area under the curve (AUC) using this system is 0.99, which is significantly higher (p < 0.001) than that using a conventional biomarker (carcinoembryonic antigen (CEA), AUC of 0.71). The above system showed similar diagnostic performance, dealing with early-stage CRC, young CRC, and CEA-negative CRC patients.