Sparse Graph Representation Learning based on Reinforcement Learning for Personalized Mild Cognitive Impairment (MCI) Diagnosis.

Resting-state functional magnetic resonance imaging (rs-fMRI) has gained attention as a reliable technique for investigating the intrinsic function patterns of the brain. It facilitates the extraction of functional connectivity networks (FCNs) that capture synchronized activity patterns among regions of interest (ROIs). Analyzing FCNs enables the identification of distinctive connectivity patterns associated with mild cognitive impairment (MCI). For MCI diagnosis, various sparse representation techniques have been introduced, including statistical- and deep learningbased methods. However, these methods face limitations due to their reliance on supervised learning schemes, which restrict the exploration necessary for probing novel solutions. To overcome such limitation, prior work has incorporated reinforcement learning (RL) to dynamically select ROIs, but effective exploration remains challenging due to the vast search space during training. To tackle this issue, in this study, we propose an advanced RL-based framework that utilizes a divide-and-conquer approach to decompose the FCN construction task into smaller subproblems in a subject-specific manner, enabling efficient exploration under each sub-problem condition. Additionally, we leverage the learned value function to determine the sparsity level of FCNs, considering individual characteristics of FCNs. We validate the effectiveness of our proposed framework by demonstrating its superior performance in MCI diagnosis on publicly available cohort datasets.

FTMMR: Fusion Transformer for Integrating Multiple Molecular Representations.

Molecular property prediction has gained substantial attention due to its potential for various bio-chemical applications. Numerous attempts have been made to enhance the performance by combining multiple molecular representations (1D, 2D, and 3D). However, most prior works only merged a limited number of representations or tried to embed multiple representations through a single network without using representation-specific networks. Furthermore, the heterogeneous characteristics of each representation made the fusion more challenging. Addressing these challenges, we introduce the Fusion Transformer for Multiple Molecular Representations (FTMMR) framework. Our strategy employs three distinct representation-specific networks and integrates information from each network using a fusion transformer architecture to generate fused representations. Additionally, we use self-supervised learning methods to align heterogeneous representations and to effectively utilize the limited chemical data available. In particular, we adopt a combinatorial loss function to leverage the contrastive loss for all three representations. We evaluate the performance of FTMMR using seven benchmark datasets, demonstrating that our framework outperforms existing fusion and self-supervised methods.

Spectral Graph Neural Network-Based Multi-Atlas Brain Network Fusion for Major Depressive Disorder Diagnosis.

Major Depressive Disorder (MDD) imposes a substantial burden within the healthcare domain, impacting millions of individuals worldwide. Functional Magnetic Resonance Imaging (fMRI) has emerged as a promising tool for the objective diagnosis of MDD, enabling the investigation of functional connectivity patterns in the brain associated with this disorder. However, most existing methods focus on a single brain atlas, which limits their ability to capture the complex, multi-scale nature of functional brain networks. To address these limitations, we propose a novel multi-atlas fusion method that incorporates early and late fusion in a unified framework. Our method introduces the concept of the holistic Functional Connectivity Network (FCN), which captures both intra-atlas relationships within individual atlases and inter-regional relationships between atlases with different brain parcellation scales. This comprehensive representation enables the identification of potential disease-related patterns associated with MDD in the early stage of our framework. Moreover, by decoding the holistic FCN from various perspectives through multiple spectral Graph Convolutional Neural Networks and fusing their results with decision-level ensembles, we further improve the performance of MDD diagnosis. Our approach is easily implemented with minimal modifications to existing model structures and demonstrates a robust performance across different baseline models. Our method, evaluated on public resting-state fMRI datasets, surpasses the current multi-atlas fusion methods, enhancing the accuracy of MDD diagnosis. The proposed novel multi-atlas fusion framework provides a more reliable MDD diagnostic technique. Experimental results show our approach outperforms both single- and multi-atlas-based methods, demonstrating its effectiveness in advancing MDD diagnosis.

Graph-Based Conditional Generative Adversarial Networks for Major Depressive Disorder Diagnosis With Synthetic Functional Brain Network Generation.

Major Depressive Disorder (MDD) is a pervasive disorder affecting millions of individuals, presenting a significant global health concern. Functional connectivity (FC) derived from resting-state functional Magnetic Resonance Imaging (rs-fMRI) serves as a crucial tool in revealing functional connectivity patterns associated with MDD, playing an essential role in precise diagnosis. However, the limited data availability of FC poses challenges for robust MDD diagnosis. To tackle this, some studies have employed Deep Neural Networks (DNN) architectures to construct Generative Adversarial Networks (GAN) for synthetic FC generation, but this tends to overlook the inherent topology characteristics of FC. To overcome this challenge, we propose a novel Graph Convolutional Networks (GCN)-based Conditional GAN with Class-Aware Discriminator (GC-GAN). GC-GAN utilizes GCN in both the generator and discriminator to capture intricate FC patterns among brain regions, and the class-aware discriminator ensures the diversity and quality of the generated synthetic FC. Additionally, we introduce a topology refinement technique to enhance MDD diagnosis performance by optimizing the topology using the augmented FC dataset. Our framework was evaluated on publicly available rs-fMRI datasets, and the results demonstrate that GC-GAN outperforms existing methods. This indicates the superior potential of GCN in capturing intricate topology characteristics and generating high-fidelity synthetic FC, thus contributing to a more robust MDD diagnosis.

Self-powered semitransparent WS2/LaVO3vertical-heterostructure photodetectors by employing interfacial hexagonal boron nitride.

Two-dimensional (2D) semiconductor and LaVO3materials with high absorption coefficients in the visible light region are attractive structures for high-performance photodetector (PD) applications. Insulating 2D hexagonal boron nitride (h-BN) with a large band gap and excellent transmittance is a very attractive material as an interface between 2D/semiconductor heterostructures. We first introduce WS2/h-BN/LaVO3semitransparent PD. The photo-current/dark current ratio of the device exhibits a delta-function characteristic of 4 × 105at 0 V, meaning 'self-powered'. The WS2/h-BN/LaVO3PD shows up to 0.27 A W-1responsivity (R) and 4.6 × 1010cm Hz1/2W-1detectivity (D*) at 730 nm. Especially, it was confirmed that theD* performance improved by about 5 times compared to the WS2/LaVO3device at zero bias. Additionally, it is suggested that the PD maintains 87% of its initialRfor 2000 h under the atmosphere with a temperature of 25 °C and humidity of 30%. Based on the above results, we suggest that the WS2/h-BN/LaVO3heterojunction is promising as a self-powered optoelectronic device.

Semitransparent Solar Cells Employing n-Type Graphene on LaVO3.

To effectively utilize solar energy, semitransparent solar cells are essential in various fields such as building-integrated solar power generation and portable solar chargers. We report triethylenetetramine (TETA)-doped graphene (Gr) transparent conductive electrode (TCE)-based LaVO3 semitransparent solar cells. To optimize the Gr TCE, we varied the TETA molar concentration (nD) from 0.1 to 0.3 mM. TETA-doped Gr (TETA-Gr)/LaVO3 semitransparent solar cells exhibit the highest 1.45% efficiency and 62% average visible transmittance at nD = 0.2 mM. These results indicate that the TETA-Gr/LaVO3 structure not only harvests solar energy in the ultraviolet-visible region but also exhibits translucency, thanks to the thin film. Thanks to its translucent properties, we improved the power conversion efficiency (PCE) to 1.99% by adding an Al reflective mirror to the semitransparent cells. Finally, the device's PCE loss is only within 3% for 3000 h in air, suggesting good durability.

Bridging the BCI illiteracy gap: a subject-to-subject semantic style transfer for EEG-based motor imagery classification.

Introduction: Brain-computer interfaces (BCIs) facilitate direct interaction between the human brain and computers, enabling individuals to control external devices through cognitive processes. Despite its potential, the problem of BCI illiteracy remains one of the major challenges due to inter-subject EEG variability, which hinders many users from effectively utilizing BCI systems. In this study, we propose a subject-to-subject semantic style transfer network (SSSTN) at the feature-level to address the BCI illiteracy problem in electroencephalogram (EEG)-based motor imagery (MI) classification tasks. Methods: Our approach uses the continuous wavelet transform method to convert high-dimensional EEG data into images as input data. The SSSTN 1) trains a classifier for each subject, 2) transfers the distribution of class discrimination styles from the source subject (the best-performing subject for the classifier, i.e., BCI expert) to each subject of the target domain (the remaining subjects except the source subject, specifically BCI illiterates) through the proposed style loss, and applies a modified content loss to preserve the class-relevant semantic information of the target domain, and 3) finally merges the classifier predictions of both source and target subject using an ensemble technique. Results and discussion: We evaluate the proposed method on the BCI Competition IV-2a and IV-2b datasets and demonstrate improved classification performance over existing methods, especially for BCI illiterate users. The ablation experiments and t-SNE visualizations further highlight the effectiveness of the proposed method in achieving meaningful feature-level semantic style transfer.

Highly stable semitransparent multilayer graphene/LaVO3vertical-heterostructure photodetectors.

A heterostructure composed of a combination of semi-metallic graphene (Gr) and high-absorption LaVO3is ideal for high-performance translucent photodetector (PD) applications. Here, we present multilayer Gr/LaVO3vertical-heterostructure semitransparent PDs with various layer numbers (Ln). AtLn= 2, the PD shows the best performance with a responsivity (R) of 0.094 A W-1and a specific detectivity (D*) of 7.385 × 107cm Hz1/2W-1at 532 nm. Additionally, the average visible transmittance of the PD is 63%, i.e. it is semitransparent. We increased photocurrent (PC) by approximately 13%, from 0.564 to 0.635µA cm-2by using an Al reflector on the semitransparent PD. The PC of an unencapsulated PD maintains about 86% (from 0.571 to 0.493µA cm-2) of its initial PC value after 2000 h at 25 °C temperature/30% relative humidity, showing good stability. This behavior is superior to that of previously reported graphene-based PDs. These results show that these PDs have great potential for semitransparent optoelectronic applications.

Deep attentive spatio-temporal feature learning for automatic resting-state fMRI denoising.

Resting-state functional magnetic resonance imaging (rs-fMRI) is a non-invasive functional neuroimaging modality that has been widely used to investigate functional connectomes in the brain. Since noise and artifacts generated by non-neuronal physiological activities are predominant in raw rs-fMRI data, effective noise removal is one of the most important preprocessing steps prior to any subsequent analysis. For rs-fMRI denoising, a common trend is to decompose rs-fMRI data into multiple components and then regress out noise-related components. Therefore, various machine learning techniques have been used in such analyses with predefined procedures and manually engineered features. However, the lack of a universal definition of a noise-related source or artifact complicates manual feature engineering. Manual feature selection can result in the failure to capture unknown types of noise. Furthermore, the possibility that the hand-crafted features will only work for the broader population (e.g., healthy adults) but not for "outliers" (e.g., infants or subjects that belong to a disease cohort) is quite high. In practice, we have limited knowledge of which features should be extracted; thus, multi-classifier assembly must be implemented to improve performance, although this process is quite time-consuming. However, in real rs-fMRI applications, fast and accurate automatic identification of noise-related components on different datasets is critical. To solve this problem, we propose a novel, automatic, and end-to-end deep learning framework dedicated to noise-related component identification via a faster and more effective multi-layer feature extraction strategy that learns deeply embedded spatio-temporal features of the components. In this study, we achieved remarkable performance on various rs-fMRI datasets, including multiple adult rs-fMRI datasets from different rs-fMRI studies and an infant rs-fMRI dataset, which is quite heterogeneous and differs from that of adults. Our proposed framework also dramatically increases the noise detection speed owing to its inherent ability for deep learning (< 1s for single-component classification). It can be easily integrated into any preprocessing pipeline, even those that do not use standard procedures but depend on alternative toolboxes.

Enhanced Flexibility and Stability in Perovskite Photodiode-Solar Cell Nanosystem Using MoS2 Electron-Transport Layer.

Hybrid organic-inorganic perovskites and MoS2 are highly attractive as emerging materials for various kinds of optoelectronic devices. Here, we first report perovskite photodiode-solar cell nanosystems (PPSNs) by employing bilayer (BL) MoS2 and triethylenetetramine-doped graphene (TETA-GR) as the electron-transport layer (ETL) and transparent conductive electrode (TCE), respectively. The rigid/flexible PPSNs exhibit 0.42/0.40 AW-1 responsivity (R), 37.2/80.1 pW Hz-1/2 noise equivalent power, 1.1 × 1010/5.0 × 109 cm Hz1/2 W-1 specific detectivity at a zero-bias photodiode mode (i.e., self-power operation), similar to or even greater than those of previous reports, and 14.27/12.12% power conversion efficiency at a photovoltaic mode. The PPSNs show high long-term stabilities by maintaining more than 78% of the initial R for 30 days. The flexible PPSNs maintain about 80% of the original R during 1000 bending tests at 4 mm radius of curvature, indicating excellent mechanical properties. These high performances result from the enhanced TCE properties, well-matched band offsets at the cathode/ETL/active layer interfaces, and the reduced carrier recombination/charge-transfer resistance by the use of TETA-GR TCE and BL-MoS2 ETL.

High-performance and -stability graphene quantum dots-mixed conducting polymer/porous Si hybrid solar cells with titanium oxide passivation layer.

Recently, conducting polymer/Si hybrid solar cells (HSCs) based on simple fabrication processes have become highly attractive due to their low cost, but low conductivity of the polymer, high reflection index of Si, and large recombination loss on the Si back contact are major drawbacks that should be solved for the practical applications. Here, we first report HSCs composed of graphene quantum dots (GQDs)-mixed poly (3,4-ethylenedioxythiophene) (PEDOT:GQDs)/ porous Si (PSi)/n-Si/titanium oxide (TiO x , back passivation layer). Maximum power conversion efficiency (PCE) of 10.49% is obtained from the HSCs at an active area of 5 mm2, resulting from the enhanced conductivity of the PEDOT:GQDs, the reduced reflectivity of Si (the increased absorption) by the formation of PSi, and the prevented recombination loss at the Si backside due to the passivation. In addition, the HSCs of 16 mm2 active area maintain ∼78% (absolutely from 8.03% to 6.28%) of the initial PCE even while kept under ambient conditions for 15 d.

New Pyrene Derivatives Having Rigid Imidazole Moiety for Blue Emitters.

Pyrene, imidazole and dibenzofuran were used to synthesize new blue emitters of 1-(4-(dibenzo[b,d]furan-4-yl)phenyl)-2-(pyren-1-yl)-1H-phenanthro[9,10-d]imidazole (BFP-PI) and 1-(4-(dibenzo[b,d]furan-4-yl)phenyl)-4,5-diphenyl-2-(pyren-1-yl)-1H-imidazole (BFP-DPI). In the film state, BFP-PI and BFP-DPI show photoluminescence (PL) maximum values of 462 nm and 459 nm. The relative PL quantum efficiency (PLQY) of BFP-PI and BFP-DPI is 89.16% and 79.2% by using reference compound of 9,10-diphenylanthracene. The device using BFP-PI in the non-doped state as emitting material showed current efficiency (C.E.) of 3 cd/A and external quantum efficiency (E.Q.E.) of 2.15%, and the device using BFP-DPI as emitting material exhibited C.E. of 2.64 cd/A and E.Q.E. of 1.6%.

Graphene-Based Semiconductor Heterostructures for Photodetectors.

Graphene transparent conductive electrodes are highly attractive for photodetector (PD) applications due to their excellent electrical and optical properties. The emergence of graphene/semiconductor hybrid heterostructures provides a platform useful for fabricating high-performance optoelectronic devices, thereby overcoming the inherent limitations of graphene. Here, we review the studies of PDs based on graphene/semiconductor hybrid heterostructures, including device physics/design, performance, and process technologies for the optimization of PDs. In the last section, existing technologies and future challenges for PD applications of graphene/semiconductor hybrid heterostructures are discussed.

Highly-flexible perovskite photodiodes employing doped multilayer-graphene transparent conductive electrodes.

We first report highly-flexible perovskite photodiodes, using AuCl3-doped multilayer-graphene transparent conducting electrodes. The doping effect of the AuCl3 is more effective when the number of layers (L n ) = 1 and 2 rather than 3 and 4, as analyzed by Raman scattering and sheet resistance. The photodiodes optimized at L n  = 2 exhibit a 105 photo-/dark-current ratio, 0.4 AW-1 responsivity, 80% external quantum efficiency, 5.3 × 1010 cm Hz1/2/W detectivity, 90 dB linear dynamic range, and ∼1.1 µs response time. In addition, the photodiodes show excellent bending stabilities, maintaining a responsivity at about 70% of its initial value, even after 1000 bending cycles at a bending curvature of 4 mm.

High-Performance Next-Generation Perovskite Nanocrystal Scintillator for Nondestructive X-Ray Imaging.

Readily commercializable and cost-effective next-generation CsPbBr3 perovskite nanocrystals (PNCs) based X-ray detectors are demonstrated. The PNCs-based X-ray detector exhibits higher spatial resolution (9.8 lp mm-1 at modulation transfer function (MTF) = 0.2 and 12.5-8.9 lp mm-1 for a linear line chart), faster response time (≈200 ns), and comparable stability (>40 Gyair s-1 of X-ray exposure) compared with the commercialized terbium-doped gadolinium oxysulfide (GOS)-based detectors (spatial resolution = 6.2 lp mm-1 at MTF = 0.2 and 6.3 lp mm-1 for a linear line chart, response time = ≈1200 ns) because the PNCs-based scintillator has ≈5.6-fold faster average photoluminescence lifetime and stronger emission than the GOS-based one.

Efficient Organic-Inorganic Hybrid Flexible Perovskite Solar Cells Prepared by Lamination of Polytriarylamine/CH3NH3PbI3/Anodized Ti Metal Substrate and Graphene/PDMS Transparent Electrode Substrate.

Flexible Ti metal substrate-based efficient planar-type CH3NH3PbI3 (MAPbI3) organic-inorganic hybrid perovskite solar cells are fabricated by lamination of the flexible Ti metal substrate/dense TiO2 electron-transporting layer formed by anodization/MAPbI3/polytriarylamine and the graphene/polydimethylsiloxane (PDMS) transparent electrode substrate. By adjusting the anodization reaction time of the polished Ti metal substrate and the number of graphene layers in the graphene/PDMS electrode, we can demonstrate the planar-type MAPbI3 flexible solar cells with a power conversion efficiency of 15.0% (mask area = 1 cm2) under 1 sun condition.

A452, an HDAC6-selective inhibitor, synergistically enhances the anticancer activity of chemotherapeutic agents in colorectal cancer cells.

Although histone deacetylase inhibitors (HDACi) alone could be clinically useful, these are most recently used in combination with other anticancer agents in clinical trials for cancer treatment. Recently, we reported the anticancer activity of an HDAC6-selective inhibitor A452 toward various cancer cell types. This study aims to present a potent synergistic antiproliferative effect of A452/anticancer agent treatment in colorectal cancer cells (CRC) cells, independently of the p53 status. A452 in combination with irinotecan, or SAHA is more potent than either drug alone in the apoptotic pathway as evidenced by activated caspase-3 and PARP, increased Bak and pp38, decreased Bcl-xL, pERK, and pAKT, and induced apoptotic cells. Furthermore, A452 enhances DNA damage induced by anticancer agents as indicated by the increased accumulation of γH2AX and the activation of the checkpoint kinase Chk2. The silencing of HDAC6 enhances the cell growth inhibition and cell death caused by anticancer agents. In addition, A452 induces the synergistic suppression of cell migration and invasion. This study suggests a mechanism by which HDAC6-selective inhibition can enhance the efficacy of specific anticancer agents in CRC cells.

Semitransparent Flexible Organic Solar Cells Employing Doped-Graphene Layers as Anode and Cathode Electrodes.

Semitransparent flexible photovoltaic cells are advantageous for effective use of solar energy in many areas such as building-integrated solar-power generation and portable photovoltaic chargers. We report semitransparent and flexible organic solar cells (FOSCs) with high aperture, composed of doped graphene layers, ZnO, P3HT:PCBM, and PEDOT:PSS as anode/cathode transparent conductive electrodes (TCEs), electron transport layer, photoactive layer, and hole transport layer, respectively, fabricated based on simple solution processing. The FOSCs do not only harvest solar energy from ultraviolet-visible region but are also less sensitive to near-infrared photons, indicating semitransparency. For the anode/cathode TCEs, graphene is doped with bis(trifluoromethanesulfonyl)-amide or triethylene tetramine, respectively. Power conversion efficiency (PCE) of 3.12% is obtained from the fundamental FOSC structure, and the PCE is further enhanced to 4.23% by adding an Al reflective mirror on the top or bottom side of the FOSCs. The FOSCs also exhibit remarkable mechanical flexibilities through bending tests for various curvature radii.

Strong enhancement of emission efficiency in GaN light-emitting diodes by plasmon-coupled light amplification of graphene.

Recently, we have demonstrated that excitation of plasmon-polaritons in a mechanically-derived graphene sheet on the top of a ZnO semiconductor considerably enhances its light emission efficiency. If this scheme is also applied to device structures, it is then expected that the energy efficiency of light-emitting diodes (LEDs) increases substantially and the commercial potential will be enormous. Here, we report that the plasmon-induced light coupling amplifies emitted light by ∼1.6 times in doped large-area chemical-vapor-deposition-grown graphene, which is useful for practical applications. This coupling behavior also appears in GaN-based LEDs. With AuCl3-doped graphene on Ga-doped ZnO films that is used as transparent conducting electrodes for the LEDs, the average electroluminescence intensity is 1.2-1.7 times enhanced depending on the injection current. The chemical doping of graphene may produce the inhomogeneity in charge densities (i.e., electron/hole puddles) or roughness, which can play a role as grating couplers, resulting in such strong plasmon-enhanced light amplification. Based on theoretical calculations, the plasmon-coupled behavior is rigorously explained and a method of controlling its resonance condition is proposed.

A potent hydroxamic acid-based, small-molecule inhibitor A452 preferentially inhibits HDAC6 activity and induces cytotoxicity toward cancer cells irrespective of p53 status.

HDAC6-selective inhibitors are novel epigenetic anticancer agents. However, their precise mechanisms of action are incompletely understood. We investigated the anticancer mechanisms of the novel potent and selective HDAC6 inhibitor A452 compared with current clinically tested HDAC6 inhibitor ACY-1215. We demonstrate that A452 effectively inhibits the cell growth and viability of various cancer cell types, irrespective of p53 status. A452-induced apoptosis as evidenced by activated caspase 3 and PARP, increased Bak and Bax and decreased Bcl-xL. Moreover, A452 shifted cells away from antiapoptotic (AKT and ERK) pathways and toward proapoptotic (p38) pathways. A452 triggered DNA damage via increased γH2AX and activation of the checkpoint kinase Chk2. A452 induced the suppression of cell migration and invasion. Interestingly, A452 upregulated the expression of PD-L1, which regulates the PD-1 inhibitory pathway in T cells. Overall, our results suggest that A452 is more effective as an anticancer agent than ACY-1215. Therefore, therapeutically targeting HDAC6 may represent a novel strategy for cancer treatment irrespective of the p53 mutation status.