Mid-Infrared Photons Alleviate Tinnitus by Activating the KCNQ2 Channel in the Auditory Cortex.

Tinnitus is a phantom auditory sensation often accompanied by hearing loss, cognitive impairments, and psychological disturbances in various populations. Dysfunction of KCNQ2 and KCNQ3 channels-voltage-dependent potassium ion channels-in the cochlear nucleus can cause tinnitus. Despite the recognized significance of KCNQ2 and KCNQ3 channels in the auditory cortex, their precise relationship and implications in the pathogenesis of tinnitus remain areas of scientific inquiry. This study aimed to elucidate the pathological roles of KCNQ2 and KCNQ3 channels within the auditory cortex in tinnitus development and examine the therapeutic potential of mid-infrared photons for tinnitus treatment. We utilized a noise-induced tinnitus model combined with immunofluorescence, electrophysiological recording, and molecular dynamic simulation to investigate the morphological and physiological alterations after inducing tinnitus. Moreover, in vivo irradiation was administered to verify the treatment effects of infrared photons. Tinnitus was verified by deficits of the gap ratio with similar prepulse inhibition ratio and auditory brainstem response threshold. We observed an important enhancement in neuronal excitability in the auditory cortex using patch-clamp recordings, which correlated with KCNQ2 and KCNQ3 channel dysfunction. After irradiation with infrared photons, excitatory neuron firing was inhibited owing to increased KCNQ2 current resulting from structural alterations in the filter region. Meanwhile, deficits of the acoustic startle response in tinnitus animals were alleviated by infrared photons. Furthermore, infrared photons reversed the abnormal hyperexcitability of excitatory neurons in the tinnitus group. This study provided a novel method for modulating neuron excitability in the auditory cortex using KCNQ2 channels through a nonthermal effect. Infrared photons effectively mitigated tinnitus-related behaviors by suppressing abnormal neural excitability, potentially laying the groundwork for innovative therapeutic approaches for tinnitus treatment.

Transcriptional profile changes caused by noise-induced tinnitus in the cochlear nucleus and inferior colliculus of the rat.

INTRODUCTION: Tinnitus is a prevalent and disabling condition characterized by the perception of sound in the absence of external acoustic stimuli. The hyperactivity of the auditory pathway is a crucial factor in the development of tinnitus. This study aims to examine genetic expression variations in the dorsal cochlear nucleus (DCN) and inferior colliculus (IC) following the onset of tinnitus using transcriptomic analysis. The goal is to investigate the relationship between hyperactivity in the DCN and IC. METHODS: To confirm the presence of tinnitus behavior, we utilized the gap pre-pulse inhibition of the acoustic startle (GPIAS) response paradigm. In addition, we conducted auditory brainstem response (ABR) tests to determine the baseline hearing thresholds, and repeated the test one week after subjecting the rats to noise exposure (8-16 kHz, 126 dBHL, 2 h). Samples of tissue were collected from the DCN and IC in both the tinnitus and non-tinnitus groups of rats. We employed RNA sequencing and quantitative PCR techniques to analyze the changes in gene expression between these two groups. This allowed us to identify any specific genes or gene pathways that may be associated with the development or maintenance of tinnitus in the DCN and IC. RESULTS: Our results demonstrated tinnitus-like behavior in rats exposed to noise, as evidenced by GPIAS measurements. We identified 61 upregulated genes and 189 downregulated genes in the DCN, along with 396 upregulated genes and 195 downregulated genes in the IC. Enrichment analysis of the DCN revealed the involvement of ion transmembrane transport regulation, synaptic transmission, and negative regulation of neuron apoptotic processes in the development of tinnitus. In the IC, the enrichment analysis indicated that glutamatergic synapses and neuroactive ligand-receptor interaction pathways may significantly contribute to the process of tinnitus development. Additionally, protein-protein interaction (PPI) networks were constructed, and 9 hub genes were selected based on their betweenness centrality rank in the DCN and IC, respectively. CONCLUSIONS: Our findings reveal enrichment of differential expressed genes (DEGs) associated with pathways linked to alterations in neuronal excitability within the DCN and IC when comparing the tinnitus group to the non-tinnitus group. This indicates an increased trend in neuronal excitability within both the DCN and IC in the tinnitus model rats. Additionally, the enriched signaling pathways within the DCN related to changes in synaptic plasticity suggest that the excitability changes may propagate to IC. NEW AND NOTEWORTHY: Our findings reveal gene expression alterations in neuronal excitability within the DCN and IC when comparing the tinnitus group to the non-tinnitus group at the transcriptome level. Additionally, the enriched signaling pathways related to changes in synaptic plasticity in the differentially expressed genes within the DCN suggest that the excitability changes may propagate to IC.

Crystallization and electrochemical properties of KxV2O5 nano-ribbons obtained via a solvothermal process as a promising cathode for PIBs.

In this research, a series of K+-intercalated quasi-1D vanadium-based nano-ribbons (KxV2O5 NRs) were synthesized via a facile solvothermal method. The solvation and reductive effects of vanadium oxide precursors (V2O5 powder) on the crystallization and growth of KxV2O5 NRs were studied. Besides, post-heat treatment was performed to improve the crystallinity of KxV2O5 NRs. These KxV2O5 NRs were adopted as active cathodes for potassium-ion batteries (PIBs), whose K+ storage properties were systematically evaluated using various electrochemical methods. The relationship among the morphology, crystallinity, working voltage window and electrochemical reversible K+ storage performance of KxV2O5 NRs was studied and established. Results reveal that KxV2O5-HG, which was prepared via a solvothermal reaction involving a solvation process (using H2O2) and a proper reducing condition (proper dose of glucose) with V2O5 powder as the raw material, would be more beneficial for the reversible storage of K+ when used as the cathode for PIBs compared to other contrast samples. In addition, the enhanced crystallinity and slightly broadened working voltage window of KxV2O5-HG could hinder its long-term cycling stability upon repeated K+ insertions/extractions.

Retrospective for the Dynamic Sensorium Competition for predicting large-scale mouse primary visual cortex activity from videos.

Understanding how biological visual systems process information is challenging because of the nonlinear relationship between visual input and neuronal responses. Artificial neural networks allow computational neuroscientists to create predictive models that connect biological and machine vision. Machine learning has benefited tremendously from benchmarks that compare different model on the same task under standardized conditions. However, there was no standardized benchmark to identify state-of-the-art dynamic models of the mouse visual system. To address this gap, we established the SENSORIUM 2023 Benchmark Competition with dynamic input, featuring a new large-scale dataset from the primary visual cortex of ten mice. This dataset includes responses from 78,853 neurons to 2 hours of dynamic stimuli per neuron, together with the behavioral measurements such as running speed, pupil dilation, and eye movements. The competition ranked models in two tracks based on predictive performance for neuronal responses on a held-out test set: one focusing on predicting in-domain natural stimuli and another on out-of-distribution (OOD) stimuli to assess model generalization. As part of the NeurIPS 2023 competition track, we received more than 160 model submissions from 22 teams. Several new architectures for predictive models were proposed, and the winning teams improved the previous state-of-the-art model by 50%. Access to the dataset as well as the benchmarking infrastructure will remain online at www.sensorium-competition.net.

Analysis of network patterns and its influencing factors in Chengdu-Chongqing urban agglomeration based on multi-flow.

With the strengthening of the cross-regional flows of the economy, information, innovation, and population, this paper constructs a network model of multi-flow integration and analyzes the spatial pattern and influencing factors of urban networks in Chengdu-Chongqing Urban Agglomeration using social network analysis and spatial analysis technology. The main conclusions are as follows. (1) The density and efficiency are in the transition stage from the primary level to the medium level in the comprehensive network. (2) The overall pattern keeps a polyhedral pyramid structure with Chengdu ↔ Chongqing as the core axis, and the grade of each axis has been significantly raised. (3) Four groups are formed using the social network method and show a geographic proximity effect. In addition, the connections within each group are relatively close, but the connections between the groups are significantly different. (4) Location conditions, economic development level, enterprise development level, scientific research investment, scientific and technological development level, and government support have a greater impact on the formation of the comprehensive network of Chengdu-Chongqing urban agglomeration. Information application level and transportation accessibility show a small impact and human capital level has not yet produced a significant impact.

Spatial-temporal differentiation pattern and influencing factors of land economic density at the township scale in Zhejiang Province.

Based on the land economic density of 892 town units, the spatial pattern of the land economic density in Zhejiang Province is analyzed using the coefficient of variation, spatial classification, and spatial correlation methods, and the influencing factors are analyzed using a spatial regression model. The results are as follows: (1) The coefficients of variation were 2.6 and 3.1 in 2014 and 2019, respectively, indicating that the degree of imbalance of the town's industrial economy at the county level increased. (2) The distribution of the high-level agglomeration areas was characterized by one core area and two sub-core areas. The main core area was located at the junction of Hangzhou City, Shaoxing City, and Jiaxing City, and the two sub-core areas were located in Yuyao City and the main urban area of Ningbo City. In addition, several small-scale agglomeration areas composed of medium and high-level units were distributed in Wenzhou City. (3) The high-value agglomeration and low-value agglomeration distribution in the spatial correlation patterns was identified using the spatial auto-correlation method. The hot spots and sub-hot spots were distributed in Northern Zhejiang, and the cold spots formed a large-scale agglomeration in Quzhou City, Lishui City, Taizhou City, and several other cities in Southern Zhejiang. (4) Compared with the county scale, the spatial scope of the high-level areas in Northern Zhejiang shrunk significantly at the township scale, and the high-level agglomeration areas along the southeast coast changed into a cluster of several townships. (5) According to the geographically weighted regression (GWR) model, the importance of influencing factors is as follows: population density > regional area > industrial output value per capita > total population > proportion of secondary and tertiary personnel > total employees.

Most discriminative stimuli for functional cell type clustering.

Identifying cell types and understanding their functional properties is crucial for unraveling the mechanisms underlying perception and cognition. In the retina, functional types can be identified by carefully selected stimuli, but this requires expert domain knowledge and biases the procedure towards previously known cell types. In the visual cortex, it is still unknown what functional types exist and how to identify them. Thus, for unbiased identification of the functional cell types in retina and visual cortex, new approaches are needed. Here we propose an optimization-based clustering approach using deep predictive models to obtain functional clusters of neurons using Most Discriminative Stimuli (MDS). Our approach alternates between stimulus optimization with cluster reassignment akin to an expectation-maximization algorithm. The algorithm recovers functional clusters in mouse retina, marmoset retina and macaque visual area V4. This demonstrates that our approach can successfully find discriminative stimuli across species, stages of the visual system and recording techniques. The resulting most discriminative stimuli can be used to assign functional cell types fast and on the fly, without the need to train complex predictive models or show a large natural scene dataset, paving the way for experiments that were previously limited by experimental time. Crucially, MDS are interpretable: they visualize the distinctive stimulus patterns that most unambiguously identify a specific type of neuron.

The Association Between Tinnitus Sensation-Level Loudness and Sleep Quality in Patients With Subjective Consecutive Tinnitus: A Mediation Analysis.

PURPOSE: So far, there have been no in-depth analyses of the connection between tinnitus sensation-level loudness and sleep quality. Accordingly, the present study was formulated as a mediation analysis focused on exploring this relationship. METHOD: Overall, 1,255 adults with consecutive subjective tinnitus who had sought outpatient treatment were enrolled in the present study. RESULTS: Direct effects of tinnitus sensation-level loudness on sleep quality were not statistically significant (95% confidence intervals [CI] include zero), as measured by the point estimate, -0.016. However, the 95% CI for indirect effects did not include zero when assessing the Self-Rating Anxiety Scale (SAS) scores, the Self-Rating Depression Scale (SDS) scores, the visual analogue scale (VAS) scores, and self-reported tinnitus annoyance. CONCLUSIONS: These results suggest that tinnitus sensation-level loudness does not directly have an effect on sleep quality. However, it indirectly impacts sleep quality, mediated by SAS scores, SDS scores, the impact of tinnitus on life measured using the VAS, and self-reported tinnitus annoyance. As such, alleviating anxiety and depression in patients with tinnitus may result in reductions in their insomnia even if there is no reduction in tinnitus loudness. Importantly, otolaryngologists and other clinicians treating tinnitus should refer patients with tinnitus suffering from insomnia with comorbid depression or anxiety for appropriate psychological and/or psychiatric treatment.

Transcriptional-profile changes in the medial geniculate body after noise-induced tinnitus.

Tinnitus is a disturbing condition defined as the occurrence of acoustic hallucinations with no actual sound. Although the mechanisms underlying tinnitus have been explored extensively, the pathophysiology of the disease is not completely understood. Moreover, genes and potential treatment targets related to auditory hallucinations remain unknown. In this study, we examined transcriptional-profile changes in the medial geniculate body after noise-induced tinnitus in rats by performing RNA sequencing and validated differentially expressed genes via quantitative polymerase chain reaction analysis. The rat model of tinnitus was established by analyzing startle behavior based on gap-pre-pulse inhibition of acoustic startles. We identified 87 differently expressed genes, of which 40 were upregulated and 47 were downregulated. Pathway-enrichment analysis revealed that the differentially enriched genes in the tinnitus group were associated with pathway terms, such as coronavirus disease COVID-19, neuroactive ligand-receptor interaction. Protein-protein-interaction networks were established, and two hub genes (Rpl7a and AC136661.1) were identified among the selected genes. Further studies focusing on targeting and modulating these genes are required for developing potential treatments for noise-induced tinnitus in patients.

Spatial pattern of China's rural digital economy based on subjective-Objective evaluation: Evidence from 2085 counties.

The rural digital economy plays an essential role in China's industrial upgrading, transformation, and urban-rural integration. To determine the state of China's rural digital economy, we constructed a county-level evaluation system using the subjective-objective evaluation method and calculated the digital economic levels of 2085 counties. Then, we analyzed the spatial distribution characteristics, spatial autocorrelation pattern, spatial disequilibrium degree, and spatial driving force of the rural digital economy at the county level using spatial analysis technology and a self-organizing feature mapping model. The results are as follows: 1) Compared with the real economy, the agglomeration effect of the digital economy was more obvious, and the economic gradient was more significant. Specifically, the dense high-value regions formed a continuous belt on the eastern coast from the Beijing-Tianjin area to the Pearl River Delta, opposite the dense low-value regions in the west. 2) There were significant differences in the rural digital economy within cities or provinces. Intraregional differences were not necessarily linked to the overall digital economy level because central and northeastern China presented a more balanced rural digital economy. 3) Digital network performance, e-commerce level, and economic vitality were identified as the core factors influencing the rural digital economy.

Pain relief effect of angiopuncture therapy on patients with postoperative pain: A clinical trial.

BACKGROUND: The objective of this study is to study the pain relief effects of angiopuncture therapy in patients with postoperative pain. METHODS: Forty-one patients were randomly selected based on the inclusion and exclusion criteria. Doppler imaging was performed to locate the cutaneous perforator. Angiopuncture was performed on the first postoperative day. A Numerical Rating Scale was used to evaluate the degree of pain before and after angiopuncture. Utilizing the paired t test or Wilcoxon signed rank test, all pre- and post-data were examined, and further subgroup analysis based on time was performed. RESULTS: Variance analysis revealed a significant difference before and after angiopuncture (P < .05). The results of the subgroup analysis showed the pain-relieving effect of angiopuncture for postoperative pain patients at the time points of 6 hours, 12 hours, 24 hours, 48 hours, and 72 hours was apparent (P < .05). CONCLUSION: The angiopuncture therapy approach may assist in pain relief in patients with postoperative pain.

Perforating cutaneous vessels: A key feature of acupoints - Anatomical evidence from five-Shu acupoints in the upper limbs.

Acupuncture has been proven an effective clinical treatment for numerous pathological conditions and malfunctions. However, substantial anatomical evidence for acupuncture points (APs) and meridians is still lacking, so the location of APs is relatively subjective and understanding of the biological mechanisms of acupuncture is limited. All these problems hinder the clinical applications and worldwide acceptance of acupuncture. Our long-term microsurgery experience has indicated that Perforating Cutaneous Vessels (PCVs) are highly relevant to APs but the anatomical evidence is insufficient. To address this lack, two specimens of fresh adult human upper limbs were dissected using an advanced vascular perfusion-fixation method and then examined. The results show that all 30 five-Shu APs in the upper limbs have corresponding PCVs. Both specimens showed a 100% coincidence rate between APs and PCVs, indicating that PCVs could be critical anatomical features of APs. This study also provides an anatomical basis for locating APs objectively via preliminary detection of PCVs. The findings could lead to a better theoretical understanding of mechanisms of acupuncture and the essence of meridians.

Outcomes of cochlear implantation in 75 patients with auditory neuropathy.

Background: Cochlear implantation (CI) outcomes in patients with auditory neuropathy (AN) are variable, which hampers patients' decisions on CI. Objective: This study aims to assess the outcomes of CI in individuals diagnosed with AN and to examine the various factors that may influence the effectiveness of this intervention. Methods: A total of 75 patients diagnosed with AN were included in the study. The hearing threshold, the score of categories of auditory performance (CAP), speech intelligibility rating (SIR), and speech audiometry were tested. Genetic testing was conducted by medical exome sequencing in 46 patients. Results: After CI, the average aided hearing threshold for patients with prelingual and post-lingual onset was 38.25 ± 6.63 dB and 32.58 ± 9.26 dB, respectively; CAP score improved to 5.52 ± 1.64 (p < 0.001) and 6.00 ± 0.96 (p < 0.001), respectively; SIR score increased to 3.57 ± 1.22 (p < 0.001) and 4.15 ± 0.95 (p < 0.001), respectively. Maximum speech recognition ranged from 58 to 93% for prelingual onset patients and 43 to 98% for those with post-lingual onset. Speech outcomes of CI in cases with cochlear nerve (CN) deficiency were significantly poorer (p = 0.008). Molecular etiologies, including TWIST1, ACTG1, m.A7445G, and a copy-number variant (CNV) carrying ACTB, were related to AN here. Conclusion: CI is a viable therapy option for patients with AN; CN deficiency might impact outcomes of CI.

Accuracy analysis of traditional acupoint location and the coincidence of cutaneous arterial perforators and acupoints.

Several reports have shown a coincidence relationship between perforators and acupoints. However, there have been few previous reports of objective experimental methods to verify the reliability of the accuracy of acupoint location (APL) with nearby perforators. This research aimed to determine the internal agreement of the APL of five acupuncturists and to analyze the coincidence rate of acupoints with nearby perforators. Three two healthy volunteers were recruited with the inclusion and exclusion criteria. Three TCM clinical physicians determined acupoints in areas of the lower limb of participants. Two microsurgeons sketched corresponding regions based on the most common skin flap operation sites, located bone markers, and drew the skin flap axis. Doppler ultrasound was used to mark the perforator point and the distances measured for both points. There is no significant difference in the distance between the acupoints and perforators localization in different groups, and there are significant differences between the angle formed by acupoints and penetrators in all groups. All the points located by the traditional Chinese medicine (TCM) therapists are distributed around the dot. The distance between the coordinate point (A-B) of Wenliu (LI7) localization is the largest, reaching 16.6 mm. The accuracy of the acupoint location of each physician is limited by the clinical experience of physicians, and the difference among them is significant. There is a certain correspondence between the location of acupoints and perforators, which needs further studies to confirm.

The Dynamic Sensorium competition for predicting large-scale mouse visual cortex activity from videos.

Understanding how biological visual systems process information is challenging due to the complex nonlinear relationship between neuronal responses and high-dimensional visual input. Artificial neural networks have already improved our understanding of this system by allowing computational neuroscientists to create predictive models and bridge biological and machine vision. During the Sensorium 2022 competition, we introduced benchmarks for vision models with static input (i.e. images). However, animals operate and excel in dynamic environments, making it crucial to study and understand how the brain functions under these conditions. Moreover, many biological theories, such as predictive coding, suggest that previous input is crucial for current input processing. Currently, there is no standardized benchmark to identify state-of-the-art dynamic models of the mouse visual system. To address this gap, we propose the Sensorium 2023 Benchmark Competition with dynamic input (https://www.sensorium-competition.net/). This competition includes the collection of a new large-scale dataset from the primary visual cortex of five mice, containing responses from over 38,000 neurons to over 2 hours of dynamic stimuli per neuron. Participants in the main benchmark track will compete to identify the best predictive models of neuronal responses for dynamic input (i.e. video). We will also host a bonus track in which submission performance will be evaluated on out-of-domain input, using withheld neuronal responses to dynamic input stimuli whose statistics differ from the training set. Both tracks will offer behavioral data along with video stimuli. As before, we will provide code, tutorials, and strong pre-trained baseline models to encourage participation. We hope this competition will continue to strengthen the accompanying Sensorium benchmarks collection as a standard tool to measure progress in large-scale neural system identification models of the entire mouse visual hierarchy and beyond.

Prognostic factor analysis in patients with early-stage nasopharyngeal carcinoma in the USA.

Aim: To evaluate independent risk factors specific for early-stage nasopharyngeal carcinoma (NPC). Methods: A total of 566 patients with early-stage NPC from 2004 to 2019 were identified using the Surveillance, Epidemiology and End Results database. Results: Older ages (70-79 and >80 years) were independent risk factors, with hazard ratios of 1.961 and 5.011, respectively. The hazard ratio for early-stage NPC in Asian and Pacific Islander residents (0.475) was lower than that for White residents. A tumor size <3 cm was a protective factor for overall and cancer-specific survival in the current study. Conclusion: In patients with early-stage NPC, age >70 years, race and tumor size were independent prognosticators for cancer-specific survival.

Transcriptional Profile Changes after Noise-Induced Tinnitus in Rats.

Tinnitus is an unpleasant symptom characterized by detective hearing without the actual sound input. Despite numerous studies elucidating a variety of pathomechanisms inducing tinnitus, the pathophysiology of tinnitus is not fully understood. The genes that are closely associated with this subtype of the auditory hallucination that could be utilized as potential treatment targets are still unknown. In this study, we explored the transcriptional profile changes of the auditory cortex after noise-induced tinnitus in rats using high throughput sequencing and verification of the detected genes using quantitative PCR (qPCR). Tinnitus models were established by analyzing startle behaviors through gap pre-pulse inhibition (PPI) of the acoustic startle. Two hundred and fifty-nine differential genes were identified, of which 162 genes were up-regulated and 97 genes were down-regulated. Analysis of the pathway enrichment indicated that the tinnitus group exhibited increased gene expression related to neurodegenerative disorders such as Huntington's disease and Amyotrophic lateral sclerosis. Based on the identified genes, networks of protein-protein interaction were established and five hub genes were identified through degree rank, including Fos, Nr4a1, Nr4a3, Egr2, and Egr3. Therein, the Fos gene ranked first with the highest degree after noise exposure, and may be a potential target for the modulation of noise-induced tinnitus.

Biology-Informed Recurrent Neural Network for Pandemic Prediction Using Multimodal Data.

In the biomedical field, the time interval from infection to medical diagnosis is a random variable that obeys the log-normal distribution in general. Inspired by this biological law, we propose a novel back-projection infected-susceptible-infected-based long short-term memory (BPISI-LSTM) neural network for pandemic prediction. The multimodal data, including disease-related data and migration information, are used to model the impact of social contact on disease transmission. The proposed model not only predicts the number of confirmed cases, but also estimates the number of infected cases. We evaluate the proposed model on the COVID-19 datasets from India, Austria, and Indonesia. In terms of predicting the number of confirmed cases, our model outperforms the latest epidemiological modeling methods, such as vSIR, and intelligent algorithms, such as LSTM, for both short-term and long-term predictions, which shows the superiority of bio-inspired intelligent algorithms. In general, the use of mobility information improves the prediction accuracy of the model. Moreover, the number of infected cases in these three countries is also estimated, which is an unobservable but crucial indicator for the control of the pandemic.

Study on tinnitus-related electroencephalogram microstates in patients with vestibular schwannomas.

Tinnitus is closely associated with cognition functioning. In order to clarify the central reorganization of tinnitus in patients with vestibular schwannoma (VS), this study explored the aberrant dynamics of electroencephalogram (EEG) microstates and their correlations with tinnitus features in VS patients. Clinical and EEG data were collected from 98 VS patients, including 76 with tinnitus and 22 without tinnitus. Microstates were clustered into four categories. Our EEG microstate analysis revealed that VS patients with tinnitus exhibited an increased frequency of microstate C compared to those without tinnitus. Furthermore, correlation analysis demonstrated that the Tinnitus Handicap Inventory (THI) score was negatively associated with the duration of microstate A and positively associated with the frequency of microstate C. These findings suggest that the time series and syntax characteristics of EEG microstates differ significantly between VS patients with and without tinnitus, potentially reflecting abnormal allocation of neural resources and transition of functional brain activity. Our results provide a foundation for developing diverse treatments for tinnitus in VS patients.

Bipartite invariance in mouse primary visual cortex.

A defining characteristic of intelligent systems, whether natural or artificial, is the ability to generalize and infer behaviorally relevant latent causes from high-dimensional sensory input, despite significant variations in the environment. To understand how brains achieve generalization, it is crucial to identify the features to which neurons respond selectively and invariantly. However, the high-dimensional nature of visual inputs, the non-linearity of information processing in the brain, and limited experimental time make it challenging to systematically characterize neuronal tuning and invariances, especially for natural stimuli. Here, we extended "inception loops" - a paradigm that iterates between large-scale recordings, neural predictive models, and in silico experiments followed by in vivo verification - to systematically characterize single neuron invariances in the mouse primary visual cortex. Using the predictive model we synthesized Diverse Exciting Inputs (DEIs), a set of inputs that differ substantially from each other while each driving a target neuron strongly, and verified these DEIs' efficacy in vivo. We discovered a novel bipartite invariance: one portion of the receptive field encoded phase-invariant texture-like patterns, while the other portion encoded a fixed spatial pattern. Our analysis revealed that the division between the fixed and invariant portions of the receptive fields aligns with object boundaries defined by spatial frequency differences present in highly activating natural images. These findings suggest that bipartite invariance might play a role in segmentation by detecting texture-defined object boundaries, independent of the phase of the texture. We also replicated these bipartite DEIs in the functional connectomics MICrONs data set, which opens the way towards a circuit-level mechanistic understanding of this novel type of invariance. Our study demonstrates the power of using a data-driven deep learning approach to systematically characterize neuronal invariances. By applying this method across the visual hierarchy, cell types, and sensory modalities, we can decipher how latent variables are robustly extracted from natural scenes, leading to a deeper understanding of generalization.