BACKGROUND:Psychogenic erectile dysfunction (pED) is a prevalent male erectile dysfunction without organic causes, and difficulties in erection attainment and post-penetration maintenance often co-occur. Although neuroimaging studies have implicated abnormalities in attentional control networks, direct behavioral evidence of how pED patients with this comorbid pattern process sexual cues is lacking. OBJECTIVES:To provide direct behavioral evidence and characterize attentional allocation to sexual cues in pED patients with the comorbid pattern versus healthy men, we conducted a study using eye-tracking and machine learning. MATERIALS AND METHODS:Thirty-seven men with pED exhibiting the comorbid pattern and 38 age-matched healthy controls completed free viewing of standardized sexual images varying in explicitness, whereas eye movements were recorded. Ninety gaze features were derived across six attentional metrics and five regions of interest. A two-stage feature-selection procedure identified a compact subset of discriminative features, which were used to train a classifier with cross-validation. Standardized self-report questionnaires assessing sexual function, sexual inhibition, anxiety, and depression were also administered. RESULTS:Eight eye-tracking features distinguished pED from controls with 94.8% accuracy under leave-one-out cross-validation. Relative to controls, men with pED showed reduced initial and sustained attention to sexually salient regions and increased attention to neutral regions, consistent with attentional avoidance. Questionnaire scores revealed lower sexual arousal and higher sexual inhibition, anxiety, and depression in pED, with gaze features indicative of attentional avoidance correlating with reduced arousal and elevated affective symptoms. DISCUSSION AND CONCLUSION:This is, to our knowledge, the first study to combine eye-tracking and machine learning to investigate pED, providing direct behavioral evidence for attentional avoidance of sexual cues. These findings provide behavioral evidence of altered attentional processing in pED patients with the comorbid pattern and underscore the potential of integrating eye-tracking with machine learning to support research in sexual dysfunction and sexual psychophysiology.

Lifelong premature ejaculation (LPE) involves altered responses to sexual cues. Neuroimaging has identified attention-related neural abnormalities in LPE, but behavioral evidence for attentional bias remains limited. Using eye tracking, we compared visual-attention patterns in 35 heterosexual men with LPE (andrology outpatient clinic, Henan Provincial People's Hospital) and 38 heterosexual healthy controls (HCs) (online advertisements on official platforms of Henan Provincial People's Hospital and Xidian University), from Nov 2023 to Jun 2024. Participants freely viewed three categories of sexual images (bikini-clad, nude, intercourse). Gaze metrics-first fixation order (FFO), fixation count (FC), and proportion of dwell time (PDT)-were extracted within predefined areas of interests (AOIs; face, chest, genital) for each stimulus category, yielding 27 candidate eye-tracking features per participant (3 categories × 3 AOIs × 3 metrics). We then applied sequential backward selection (SBS) with support vector machine (SVM) to select the most discriminative feature subset. An SVM classifier trained on the selected features distinguished LPE from HCs with 82.2% accuracy (AUC = 0.78). Compared with HCs, LPE showed earlier genital orienting (FFO: nude 4.2 ± 1.5 vs 6.0 ± 2.2, p < 0.001; intercourse 3.8 ± 1.5 vs 4.9 ± 1.9, p = 0.004) and more genital fixations (FC: bikini-clad 4.2 ± 1.6 vs 3.3 ± 1.2, p = 0.006; intercourse 7.9 ± 3.1 vs 6.1 ± 2.0, p = 0.008), indicating a genital-prioritized attentional bias. Anxiety and depression scores were higher in LPE (SAS: 34.1 ± 6.4 vs 29.4 ± 5.8, p = 0.003; SDS: 34.8 ± 7.6 vs 29.5 ± 6.8, p = 0.003), but no eye-tracking features correlated with these symptoms. Earlier genital orienting correlated with greater LPE severity (FFO vs PEDT: nude r = -0.47, p < 0.001; intercourse r = -0.36, p = 0.002). This study provides the first behavioral evidence of a genital-prioritized attentional bias in heterosexual men with LPE, offering novel mechanistic insight into its neurocognitive underpinnings.

BACKGROUND:Nonorganic erectile dysfunction (nonorganic ED) is a common subtype of ED characterized by impaired erectile function in the absence of identifiable organic causes. Although ED is increasingly recognized as a multidimensional condition involving interacting biological, psychological, relational, and sociocultural factors, how these influences are integrated within central neural systems in nonorganic ED remains incompletely understood. Objective neurobiological markers are therefore still lacking. OBJECTIVES:To investigate whether patients with nonorganic ED exhibit abnormalities in local intrinsic brain activity and to examine the association between these alterations and erectile function. MATERIALS AND METHODS:Resting-state functional magnetic resonance imaging data were acquired from 29 patients with nonorganic ED and 29 age-matched healthy controls. Regional homogeneity (ReHo) was used to assess local synchronization of spontaneous neural activity across the whole brain. Between-group differences were evaluated using voxel-wise analysis. Correlation analyses were performed to assess associations between altered ReHo and erectile function as measured by the International Index of Erectile Function (IIEF). RESULTS:Compared with healthy controls, patients with nonorganic ED exhibited significantly decreased ReHo in the right anterior insula (family-wise error corrected). Furthermore, reduced ReHo in this region was significantly associated with greater ED severity. DISCUSSION:These findings indicate that nonorganic ED is associated with disrupted local functional coherence in the anterior insula, a key region involved in integrating interoceptive, emotional, cognitive, and autonomic processes. The observed association with erectile function suggests that impaired integration of these processes may contribute to the central neural mechanisms underlying nonorganic ED. CONCLUSION:Patients with nonorganic ED exhibit reduced local functional coherence in the anterior insula, which is associated with ED severity. These findings provide preliminary evidence linking central neural dysfunction to clinical symptoms and support the further evaluation of resting-state fMRI metrics as candidate neurobiological markers for nonorganic ED.

Rapid serial visual presentation (RSVP) enables efficient electroencephalography (EEG)-based brain-computer interfaces, yet single-trial decoding remains difficult due to signal overlap and multicomponent entanglement. This work developed DisCo-Former, a Transformer-based framework incorporating three priors-guided components, including trend-periodicity disentanglement, channel-level embeddings that preserve global temporal pattern, and contrastive learning that exploits target-adjacent nontargets. Although DisCo-Former surpassed existing approaches, analysis revealed a consistent attention collapse: attention maps became nearly uniform, and value projection weights shrank toward zero. Removing the Transformer encoder yields DisCo-MLP, a purely multilayer perceptron (MLP) variant that preserves all remaining modules. Across two datasets and three evaluation regimes, DisCo-MLP matched or outperformed its Transformer-based counterpart. In within-subject decoding, mean AUCs ranged from approximately 0.94 to 0.98 across two datasets, consistently exceeding strong baselines. These results indicate that, for RSVP-EEG decoding, effectiveness stems less from architectural complexity and more from modeling the signal's structure. Simplicity motivated by paradigm-specific neurophysiological priors offers a practical path to state-of-the-art performance in EEG-based interfaces.

BACKGROUND/OBJECTIVES:Radiological expertise draws on semantic knowledge and perceptual-cognitive mechanisms that support diagnostic reasoning. Early radiological training is a formative period when key cognitive processes begin to integrate. Nevertheless, how the brain pattern of early radiological expertise reorganizes during the first weeks of clinical exposure remains unknown, as prior work has relied mainly on cross-sectional designs comparing mature experts to beginners. METHODS:We therefore conducted a longitudinal resting-state fMRI study in radiology interns (n = 43; 41 valid) scanned before and after short-term training. Behavioral performance improved significantly after training (p < 0.01). Regional homogeneity (ReHo) was computed for 246 Brainnetome ROIs for each subject. RESULTS:Using a Support Vector Machine (SVM)-based recursive feature elimination (RFE) pipeline, 14 of these 246 features were identified as most discriminative, spanning regions involved in visual, semantic, memory, attentional, and decision-making processes. An SVM trained on these features effectively differentiated pre- and post-training brain states (training set: 86.67% accuracy, AUC = 0.97; validation set: 81.82% accuracy, AUC = 0.72). CONCLUSIONS:The observed neuroplastic changes provide direct evidence that multidimensional cognitive functions reorganize early in radiological expertise development and offer neural targets to inform evidence-based curriculum design, personalized training, and brain-targeted interventions (e.g., neuromodulation or neurofeedback) in radiology education.

Significance:Traditional exposure therapy or cognitive training requires repeated presentation of unwanted stimuli, whereas localizationist neuromodulation overlooks individual variation. We propose a closed-loop neuromodulation approach termed functional near-infrared spectroscopy-decoded neurofeedback training, designed to modify prefrontal hemoglobin dynamics and neural activity patterns. Aim:We aim to enhance interference control without interfering stimuli using a data-driven, individualized, time-resolved decoded neurofeedback, potentially offering a balanced compromise and an alternative to traditional approaches. Approach:We employed a randomized, double-blind, between-group design. Both the decoded neurofeedback group (DecNef, n = 20 ) and the Sham group (Sham, n = 25 ) developed individualized decoders with a 1-s temporal resolution following the color-word Stroop task (CWST) before training. Both groups received decoded neurofeedback training sessions lasting 25 min daily for three consecutive days, but there was a gap in their decoding accuracy due to differences in sample size. Interference control was assessed via CWST at three timepoints: pre-training (pre-test), post-training (post-test), and 1-week follow-up. Results:There was no significant difference in feedback scores between groups, but the Stroop effect of reaction time (RT) in the DecNef group showed a significant reduction compared with the Sham group, both at post-test ( t = 3.056 , p = 0.004 ) and follow-up test ( t = 2.180 , p = 0.035 ). The difference wave amplitude (incongruent minus congruent trials) for hemodynamic response functions significantly decreased at post-test in the DecNef group (within a continuous period of 7 to 12 s, p < 0.05 ), but not in the Sham group. Multivariate pattern analysis (MVPA) revealed significantly higher classification accuracy in the DecNef group compared with the Sham group ( t = 2.370 , p = 0.024 ); furthermore, this classification accuracy showed a significant negative correlation with changes in the RT Stroop effect ( r = - 0.36 , p = 0.015 ). Conclusions:We proposed a closed-loop neuromodulation approach designed to modify prefrontal neural dynamics, with its core innovation lying in time-resolved individualized decoding. This method can significantly improve cognitive function such as interference control while avoiding exposure to unwanted stimuli and has potential for cognitive enhancement and the treatment of psychological disorders such as phobias and post-traumatic stress disorder.

Visual expertise-the ability to discriminate highly similar exemplars quickly and accurately within a category-supports skilled performance across real-world domains and is supported by distributed neural systems. We focus on non-face expertise to test cross-domain convergence in acquired real-world visual skills, treating faces separately because socially embedded, sensitive-period-constrained processing could blur this inference. It remains unresolved whether non-face expertise across heterogeneous domains converges on a shared, domain-general whole-brain architecture, or instead recruits domain-contingent neural configurations that vary with task and stimulus demands. We conducted a coordinate‑based meta‑analysis of 22 task‑fMRI studies spanning 11 real‑world non‑face expertise domains (579 participants, 210 peak‑activation foci). Primary analysis revealed a robust, right‑lateralized parieto‑temporo‑occipital circuit centered on the middle occipital gyrus, middle temporal gyrus, angular gyrus and adjoining inferior parietal lobule. We propose that this circuit constitutes a domain‑general neural core that integrates fine-grained visual features with semantic associations while supporting attention‑guided recognition of visually similar objects in expert performance. Subgroup and meta‑regression analyses uncovered a complementary adaptive component, the engagement of which varied systematically with representational and contextual factors. Pictorial stimuli and expert-novice contrasts reliably strengthened recruitment of the right-hemisphere core, whereas symbolic stimuli engagement toward left temporal regions while selectively re-engaging right-core nodes. In addition, male-skewed samples showed attenuated left-hemisphere activation. Together, these findings delineate a stable right-hemisphere neural scaffold underlying non-face visual expertise, flexibly supplemented by left-hemisphere systems as a function of stimulus format, task demands, and demographic context, providing a whole-brain reference framework for future studies. ### Competing Interest Statement The authors have declared no competing interest. the National Key R&D Program of China, Grant No.2022YFF1202400

Healthcare systems require the efficient development of expert performance. Several studies have explored the cognitive foundations of medical expert performance, especially in radiology. Studying at the brain level could provide further insight into specific mechanisms mediating medical expert performance. Researchers have recently begun to systematically employ neuroimaging in this field. Most studies focus on specific specializations rather than identifying shared neural substrates across disciplines. This systematic review and activation likelihood estimation (ALE) meta-analysis followed the PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) guidelines. A total of 297 studies examining neural correlates were identified by comparing expert and novice medical performance. After screening, 22 studies were included in the final analysis. For studies reporting three-dimensional coordinates, ALE meta-analysis revealed consistent involvement of the medial frontal lobe, including the superior frontal gyrus, dorsomedial and ventromedial prefrontal cortex, and inferior frontal and fusiform gyri. Radiology-specific analyses highlighted activation in the ventromedial prefrontal cortex, the left pre-supplementary motor area (pre-SMA), along with the fusiform and opercular inferior frontal gyri. Internal medicine-based studies highlighted involvement of the SMA, inferior frontal gyrus, and dorsomedial prefrontal cortex. Our results revealed involvement, at different levels, of the medial frontal cortex, including the SMA and superior and inferior frontal gyri, which is part of the network relevant for inhibitory control and decision-making. The development of decision-making during the diagnostic process is relevant for the training of future professionals.

To the Editor: Depression is a common psychiatric disorder, affecting over 260 million people of all ages globally.[1] Prior studies investigating the association between antidepressant use and stroke risk have yielded inconsistent results.[2,3] Consequently, it remains unclear which of the various antidepressant categories may affect stroke. Thus, the rational use of antidepressants is important for reducing stroke risk and recurrence, while offering candidate therapeutic targets. Drug-target Mendelian randomization (MR) analysis, which uses genetic variants located in or near the region of drug target genes as proxies for drug effects, is a promising tool for identifying causal links between drug targets and diseases. This study aimed to evaluate the causal associations between antidepressant target genes and stroke and its various subtypes (including any stroke [AS], ischemic stroke [IS], large artery atherosclerosis stroke [LAA], cardioembolic stroke [CES], and small vessel stroke [SVS]) using drug-target MR analysis. Various antidepressants were identified from the World Health Organization Collaborating Centre for Drug Statistics Methodology, and were classified by the Anatomical Therapeutic Chemical classification system. The DrugBank (https://go.drugbank.com/) and ChEMBL (https://www.ebi.ac.uk/chembl/) databases were used to determine the genes encoding the targets of antidepressants. To identify genetic variants as proxies for the effect of drug target genes, blood cis-expression quantitative trait loci (eQTL) data from the eQTLGen Consortium (n = 31,684) were used. The cis-eQTL located within 1 Mb downstream or upstream of the region of the drug target genes with a false discovery rate (FDR) <0.05 and F-statistic (calculated by the formula: F-statistic = beta2/se2) >10 were screened. Independent genetic variants without linkage disequilibrium (r2 <0.1) were used as the instrumental variables (IVs). Genome-wide association studies (GWAS) summary data for stroke and its subtypes were from GIGASTROKE consortium. Our study included only individuals of European ancestry, comprising AS (73,652 cases and 1,234,808 controls), IS (62,100 cases), LAA (6399 cases), CES (10,804 cases), and SVS (6811 cases). All participants enrolled in this study were of European ancestry, with no sample overlap with the exposure dataset in the main analysis. Detailed information of the different data sources is provided in Supplementary Table 1, https://links.lww.com/CM9/C261. All MR analyses were performed using TwoSampleMR R package in R software (v.4.0.3, R Development Core Team, Vienna, Austria), while the inverse variance weighted method was used to estimate the causal effects. The FDR method was applied for multiple testing, with an FDR <0.05 indicating statistical significance. Sensitivity analyses, including heterogeneity and pleiotropy tests, were performed using Cochrane's Q test, Rucker's Q test, MR-Egger intercept test, MR pleiotropy residual sum and outlier global test, and leave-one-out analysis. Colocalization analysis was conducted between the significant drug target genes identified in the primary MR analysis and stroke outcomes. A posterior probability of hypothesis 4 (PPH4) >0.8 was used to characterize significant evidence for colocalization. Further, we assessed the causal relationship between the candidate target genes and cerebrovascular risk factors. For drug target genes causally linked to both stroke and risk factors, a two-step mediation MR analysis was conducted to evaluate the effects of drug target genes (exposure) on stroke (outcomes) via the cerebrovascular risk factors (mediators). To determine whether the observed associations between antidepressant target gene expression and stroke risk are likely mediated by major depressive disorder (MDD) or independent of MDD, MR analysis was also conducted to evaluate the associations between MDD and stroke. Further details of this analysis are provided in the Supplementary Methods, https://links.lww.com/CM9/C261. A flow diagram of the study is presented in Supplementary Figure 1, https://links.lww.com/CM9/C261. A total of 111 drug targets encoding proteins have previously been experimentally shown to be modified by one or more antidepressants. After selecting the IVs for the antidepressant target genes, 29 of the 111 genes were identified in the outcome datasets [Supplementary Tables 2–4, https://links.lww.com/CM9/C261]. The associations between genetically predicted antidepressant target genes and stroke are presented in Figure 1, Supplementary Figures 2 and 3, and Supplementary Tables 5–11, https://links.lww.com/CM9/C261. Following FDR adjustment, we identified five drug target genes significantly associated with AS risk: KCNH2 (odds ratio [OR] = 1.057, 95% confidence interval [CI] 1.017–1.098, FDR = 0.027), MPO (OR = 1.071, 95% CI = 1.050–1.093, FDR = 7.81E−10), SIGMAR1 (OR = 0.952, 95% CI = 0.934–0.971, FDR = 1.12E−05), WARS (OR = 0.982, 95% CI = 0.973–0.992, FDR = 0.003), WARS2 (OR = 0.981, 95% CI = 0.970–0.993, FDR = 0.010). Moreover, four genetically predicted drug target genes were found to be significantly associated with IS risk: MPO (OR = 1.078, 95% CI = 1.052–1.105, FDR = 5.55E−08), SIGMAR1 (OR = 0.946, 95% CI = 0.927–0.965, FDR = 1.05E−06), SLC18A2 (OR = 0.942, 95% CI = 0.902–0.983, FDR = 0.043), WARS (OR = 0.980, 95% CI = 0.970–0.991, FDR = 0.002). Genetically predicted GRIN2D (LAA: OR = 0.465, 95% CI = 0.293–0.739, FDR = 0.034), KCNH2 (CES: OR = 1.222, 95% CI = 1.128–1.325, FDR = 2.95E−05), and WARS2 (SVS: OR = 0.938, 95% CI = 0.905–0.972, FDR = 0.011) levels were also found to be significantly associated with LAA, CES and SVS, respectively. Colocalization analysis indicated that MPO and IS, as well as GRIN2D and LAA, probably shared a causal single nucleotide polymorphism in the gene locus (MPO: PPH4 = 0.884; GRIN2D: PPH4 = 0.824; Supplementary Figure 4 and Supplementary Table 12, https://links.lww.com/CM9/C261).Figure 1: MR analysis of significant drug target genes with stroke risk. Five antidepressant targets (KCNH2, MPO, SIGMAR1, WARS, and WARS2) were significantly associated with AS risk after FDR adjustment. Additionally, four targets (MPO, SIGMAR1, SLC18A2, and WARS) showed significant associations with IS risk. Genetically predicted GRIN2D, KCNH2, and WARS2 were significantly linked to LAA, CES, and SVS, respectively. AS: Any stroke; CES: Cardioembolic stroke; CI: Confidence interval; FDR: False discovery rate; IS: Ischemic stroke; LAA: Large artery atherosclerosis stroke; MR: Mendelian randomization; OR: Odds ratio; SVS: Small-vessel stroke.The associations between MPO and GRIN2D with 14 cerebrovascular risk factors were also investigated [Supplementary Figure 5 and Supplementary Tables 13–15, https://links.lww.com/CM9/C261]. Genetically predicted MPO levels were significantly associated with atrial fibrillation (AF; OR = 1.043, 95% CI = 1.018–1.068, FDR = 0.003), heart failure (HF; OR = 1.048, 95% CI = 1.023–1.075, FDR = 0.002), and systolic blood pressure (SBP; OR = 1.256, 95% CI = 1.104–1.428, FDR = 0.003). MR analysis further revealed the causal effects of genetically predicted GRIN2D on AF (OR = 0.819, 95% CI = 0.732–0.917, FDR = 0.008) and triglyceride levels (OR = 0.829, 95% CI = 0.724–0.948, FDR = 0.045). A two-step mediation MR analysis was applied to evaluate the effects of MPO on stroke outcomes (AS and IS) via risk factors (AF, HF, and SBP). The proportions of the mediation effect of MPO on AS and IS via AF were 9.7% and 9.4%, respectively, while the corresponding values via SBP were 8.5% and 8.0%, respectively. The indirect effect of MPO on the risk of AS and IS via HF accounted for 29.7% and 30.2% of the total effect, respectively [Supplementary Figure 6 and Supplementary Table 16, https://links.lww.com/CM9/C261]. We found no evidence to support an association between genetically estimated MDD and AS, IS, LAA, CES, or SVS (all P values >0.05; stroke GWAS from GIGASTROKE or MEGASTROKE; Supplementary Figure 7 and Supplementary Tables 17 and 18, https://links.lww.com/CM9/C261). This indicates that the observed association of target genes with stroke is unlikely to be solely caused by MDD, and indicates that this association is likely independent of the association with MDD. The present study identified associations between antidepressant targets and stroke and its subtypes through drug-target MR analysis. In addition, we identified two candidate antidepressant target genes for IS and LAA (MPO and GRIN2D, respectively). Myeloperoxidase (MPO), a key inflammatory factor in the myeloid system, is highly expressed in activated human neutrophils, and plays an important role in inflammation and oxidative stress responses. Prior studies have shown that inhibition of MPO activity can reduce inflammation and enhance cellular protection against IS.[4]GRIN2D encodes the glutamate ionotropic receptor N-Methyl-D-Aspartate (NMDA) type subunit 2D (GluN2D), which is a subunit of the NMDA Receptor (NMDAR) and is involved in learning, memory, and synaptic functioning.[5] There is currently limited evidence linking GRIN2D with LAA or atherosclerosis, highlighting the need for further investigation. This study has several strengths. First, this MR study integrated the latest and largest GWAS and eQTL datasets to investigate causality and reduce confounding factors and reverse causation. Second, we systematically examined various antidepressant targets and stroke subtypes, and performed several sensitivity analyses to support our findings. Third, MR analysis of multiple cerebrovascular risk factors was performed to identify potential side effects and alternative indications crucial for future clinical applications. However, this study has several limitations, as follows. First, all participants included in the GWAS used in the present study were of European ancestry; therefore, our findings require validation in other races. Although our MR analysis indicated potential causal relationships, these associations should not be interpreted as direct evidence to indicate that antidepressants targeting these proteins would have causal effects on stroke risk. Inferring the actual pharmacological effects from genetic analyses is associated with complexities owing to variations in drug mechanisms, timing, magnitude, and duration of exposure. Although our colocalization analysis provided strong evidence to support the existence of shared causal variants in MPO and GRIN2D, the lack of colocalization evidence for other genes with MR evidence indicates that these relationships may require further investigation using larger datasets or complementary methods. Further studies are thus required to determine the effects of antidepressants on the risk of stroke. Our findings also require validation using independent datasets to ensure their robustness and broader applicability. Future research should thus explore downstream biomarkers to gain a more comprehensive understanding of the effects of antidepressant targets on stroke risk. As larger protein quantitative trait locus datasets become available, the investigation of drug-target relationships should be enhanced. Moreover, there is the potential for survivor bias because the GWAS primarily recruited survivors, possibly missing the genetic risk profiles of those who did not survive severe strokes. Finally, we identified a robust causal relationship between MPO and HF, with HF mediating the association between MPO and IS risk. Further research in non-HF patients is required to minimize potential pleiotropic effects. In conclusion, our drug target MR analysis provides insights into the associations between antidepressant targets and stroke, guiding the selection of antidepressants for individuals at risk of stroke, and identifying MPO and GRIN2D as promising stroke drug targets. However, further research is required to verify the long-term effects of antidepressants on stroke risk.

Radiological expertise develops through extensive experience in specific imaging modalities. While previous research has focused on long-term learning and neural mechanisms of expertise, the effects of short-term radiological training on resting-state neural networks remain underexplored. This study investigates the impact of four weeks of radiological interpretation training on resting-state neural networks in 32 radiology interns. Using behavioral assessments and resting-state fMRI data, a Recursive Feature Elimination Support Vector Machine (RFE-SVM) model achieved 82% accuracy in classifying data from the pre- and post-training phases. Key brain regions linked to attention, decision-making, working memory, and visual processing were identified, providing insights into how short-term training reshapes intrinsic brain networks and facilitates rapid adaptation to new skills. These findings also lay a theoretical foundation for designing more effective training programs.

Background Oxidative stress and microglial activation are critical pathomechanisms in ischemic white matter injury. Microglia, as resident immune cells in the brain, are the main cells undergoing oxidative stress response. However, the role and molecular mechanism of oxidative stress in microglia have not been clearly elucidated during white matter ischemia. Methods Extensive histological analysis of the corpus callosum was performed in BCAS mice at different time points to assess white matter injury, oxidative stress and microglial activation. Flow cytometric sorting and transcriptomic sequencing were combined to explore the underlying mechanisms regulating microglial oxidative stress and functional phenotypes. The expression of critical molecule in microglia was regulated using Cx3cr1CreER mice and clinical-stage drugs to assess its effect on white matter injury and cognitive function. Results Our study identified nuclear factor erythroid-2 related factor 2 (Nrf2) as a key transcription factor regulating oxidative stress and functional phenotype in microglia. Interestingly, we found that the sustained decrease in transiently upregulated expression of Nrf2 following chronic cerebral hypoperfusion resulted in abnormal microglial activation and white matter injury. In addition, high loads of myelin debris promoted lipid peroxidation and ferroptosis in microglia with diminished antioxidant function. Microglia with pharmacologically or genetically stimulated Nrf2 expression exhibited enhanced resistance to ferroptosis and pro-regenerative properties to myelination due to lipid and iron metabolism reprogramming. Conclusion Weakened Nrf2-mediated antioxidant responses in microglia induced metabolic disturbances and ferroptosis during chronic cerebral hypoperfusion. Targeted enhancement of Nrf2 expression in microglia may be a potential therapeutic strategy for ischemic white matter injury.

Harnessing the complementarity between the brain and artificial neural networks has shown great promise for the development of novel brain-machine fusion systems that may rival the robustness and flexibility of the human visual system. Nonetheless, due to the requirement of human involvement, brain-machine fusion models are challenging to apply in a brain-in-the-loop manner to handle task demands that involve long duration, high intensity and complex operating environments. In this paper, we propose a brain-machine fusion approach to achieve the brain-in-the-loop modeling and brain-out-of-the-loop application. The similarities and differences between the image representations of brain responses and image features computed by deep convolutional neural network (DCNN) are firstly analyzed using a rhesus monkey dataset, and the results lay the foundation for the feasibility of brain-machine fusion. A brain-machine fusion model is then developed and a multimodal supervised contrastive learning method is proposed to jointly learn the image representations for brain responses and DCNNs. The fusion model can be applied in a brain-out-of-the-loop manner, effectively addressing the challenges encountered by human-involved approaches. Extensive experiments on a self-built human vehicle detection dataset demonstrate the effectiveness of the proposed method in improving the generalization ability of the downstream image classifiers, both in cross-modal learning or multimodal fusion settings.

Previous efforts to boost the performance of brain-computer interfaces (BCIs) have predominantly focused on optimizing algorithms for decoding brain signals. However, the untapped potential of leveraging brain plasticity for optimization remains underexplored. In this study, we enhanced the temporal resolution of the human brain in discriminating visual stimuli by eliminating the attentional blink (AB) through color-salient cognitive training, and we confirmed that the mechanism was an attention-based improvement. Using the rapid serial visual presentation (RSVP)-based BCI, we evaluated the behavioral and electroencephalogram (EEG) decoding performance of subjects before and after cognitive training in high target percentage (with AB) and low target percentage (without AB) surveillance tasks, respectively. The results consistently demonstrated significant improvements in the trained subjects. Further analysis indicated that this improvement was attributed to the cognitively trained brain producing more discriminative EEG. Our work highlights the feasibility of cognitive training as a means of brain enhancement to boost BCI performance.

Significance:fNIRS-based neuroenhancement depends on the feasible detection of hemodynamic responses in target brain regions. Using the lateral occipital complex (LOC) and the fusiform face area (FFA) in the ventral visual pathway as neurofeedback targets boosts performance in visual recognition. However, the feasibility of utilizing fNIRS to detect LOC and FFA activity in adults remains to be validated as the depth of these regions may exceed the detection limit of fNIRS.Aim:This study aims to investigate the feasibility of using fNIRS to measure hemodynamic responses in the ventral visual pathway, specifically in the LOC and FFA, in adults.Approach:We recorded the hemodynamic activities of the LOC and FFA regions in 35 subjects using a portable eight-channel fNIRS instrument. A standard one-back object and face recognition task was employed to elicit selective brain responses in the LOC and FFA regions. The placement of fNIRS optodes for LOC and FFA detection was guided by our group's transcranial brain atlas (TBA).Results:Our findings revealed selective activation of the LOC target channel (CH2) in response to objects, whereas the FFA target channel (CH7) did not exhibit selective activation in response to faces.Conclusions:Our findings indicate that, although fNIRS detection has limitations in capturing FFA activity, the LOC region emerges as a viable target for fNIRS-based detection. Furthermore, our results advocate for the adoption of the TBA-based method for setting the LOC target channel, offering a promising solution for optrode placement. This feasibility study stands as the inaugural validation of fNIRS for detecting cortical activity in the ventral visual pathway, underscoring its ecological validity. We suggest that our findings establish a pivotal technical groundwork for prospective real-life applications of fNIRS-based research.

Visual expertise reflects accumulated experience in reviewing domain-specific images and has been shown to modulate brain function in task-specific functional magnetic resonance imaging studies. However, little is known about how visual experience modulates resting-state brain network dynamics. To explore this, we recruited 22 radiology interns and 22 matched healthy controls and used resting-state functional magnetic resonance imaging (rs-fMRI) and the degree centrality (DC) method to investigate changes in brain network dynamics. Our results revealed significant differences in DC between the RI and control group in brain regions associated with visual processing, decision making, memory, attention control, and working memory. Using a recursive feature elimination-support vector machine algorithm, we achieved a classification accuracy of 88.64%. Our findings suggest that visual experience modulates resting-state brain network dynamics in radiologists and provide new insights into the neural mechanisms of visual expertise.

相比于基于比特数据的信息处理及通信技术，人类通过语义处理和传递信息的方式，在面对智能体间传递处理海量信息这一问题时显得更为高效和自然.然而由于目前缺乏关于语义度量和刻画的数学描述，涉及语义的应用无法兼顾可解释性和泛化性，无法发挥语义的高效自然的优势.本文围绕语义的度量和刻画，首先依据信息科学和神经科学相关结论，讨论了语义的内涵，并指出语义具有模块化、多模态、层级化的特点；接着提出了一种多模态信号的语义刻画和度量的数学描述；然后为了验证所提信号语义的刻画和度量的可行性和有效性，在MNIST（Mixed National Institute of Standards and Technology database）手写数字识别和水声目标识别两个应用中进行了实验，获得比传统深度学习更好的性能；最后将语义用于视频编码，实现了远超传统方法的压缩比，展现了语义在通信领域的实用价值.这为未来建立以语义为基础的新型信息处理与通信技术奠定了理论和实践基础.

Extracting objects of interest from remote sensing imagery is an essential part in various practical applications. The objects that people pay attention to in the remote sensing scene mainly include buildings, roads, vehicles, etc. In this article, extracting the aforementioned objects are collectively referred to as the target extraction task. Arising from object scale variation, appearance similarity between adjacent patches, diversity of imaging orientation, and complexity of background, it is difficult to extract complete objects from cluttered backgrounds. Deep neural network has made great achievement in dense prediction for target extraction. However, most of the previous works are still faced with a formidable challenge in discriminative context feature representation to extract targets of various categories and correctly classify pixels around the boundary. In this article, we propose a target extraction neural network, named discriminative context-aware network, to focus on discriminative high-level context features and preserve spatial information. First, a discriminative context-aware feature module is designed to generate the feature maps in the top layer, which not only captures the rich image context information but also aggregates the contrasted local information at multiple scales. Second, a refine decoder module is adopted to preserve spatial information from low-level layers and enhance the feature representation, leading to precise segmentation results. We conducted extensive experiments on building and road extraction benchmarks, including WHU building dataset and Massachusetts road dataset, together with a self-constructed dataset for vehicle extraction in SAR images. Our method achieves state-of-the-art results with fewer parameters and faster inference.