Computers in biology and medicine最新文献

{"title":"Deep-ATM DL-LSTM: A novel adaptive thresholding model with dual-layer LSTM architecture for real-time driver drowsiness detection using skin conductance signals","authors":"J Robert Theivadas,&nbsp;Suresh Ponnan","doi":"10.1016/j.compbiomed.2025.110243","DOIUrl":"10.1016/j.compbiomed.2025.110243","url":null,"abstract":"<div><div>Driver drowsiness detection systems are crucial for road safety. However, existing machine learning models struggle to adjust thresholds for Skin Conductance (SC) adaptively signals due to insufficient feature extraction of tonic and phasic responses. These responses, controlled by the sympathetic nervous system, provide valuable insights into drowsiness states but are often distorted by signal noise, reducing detection accuracy. This study proposes a novel Deep learning-based Adaptive Thresholding Model with a Dual-layer Long Short-Term Memory (LSTM) architecture, called Deep-ATM DL-LSTM, to process SC signals and dynamically improve drowsiness detection. The model leverages a two-layer LSTM architecture to compute dynamic thresholds for tonic (baseline) and phasic (rapid fluctuation) responses. The first LSTM layer extracts global and regional SC features, while the second layer processes temporal differences in drowsiness states. A softmax layer classifies drowsiness levels based on feature vector differences. The proposed method effectively addresses inter-individual variability and signal distortions by integrating robust feature extraction and adaptive thresholding, ensuring accurate drowsiness detection. Experimental testing using professional drivers on highways, in urban areas, during the day and night and in rain and frost environments demonstrated a 96.4 % accuracy level and a 0.978 AUC-ROC value that surpassed standard machine learning techniques and standard LSTM models. Existing driving conditions that include rain and frost conditions do not affect model performance (94.2 % in rain, 93.5 % in frost) and the model achieves high F1-scores for drowsiness state identification in all alert states (0.978), mild states (0.954), modest states (0.938), and acute states (0.933). The Deep-ATM DL-LSTM system detects drowsiness early by 7.5 min before dangerous levels and needs minimal wearable sensors. The combination of high accuracy, adaptive features and practical deployment from this approach solves significant problems with existing driver monitoring systems, including Behavioural-based detection PERCLOS [9] and Physiological-based detection EEG and EMG signals [4], before delivering an effective tool to reduce drowsiness-caused road accidents.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"192 ","pages":"Article 110243"},"PeriodicalIF":7.0,"publicationDate":"2025-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143859196","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"BigLSTM: Recurrent neural network for the treatment of anomalous temporal signals. Application in the prediction of endotracheal obstruction in COVID-19 patients in the intensive care unit","authors":"Pablo Fernández-López ,&nbsp;Patricio García Báez ,&nbsp;Ylermi Cabrera-León ,&nbsp;Juan L. Navarro-Mesa ,&nbsp;Guillermo Pérez-Acosta ,&nbsp;José Blanco-López ,&nbsp;Carmen Paz Suárez-Araujo","doi":"10.1016/j.compbiomed.2025.110146","DOIUrl":"10.1016/j.compbiomed.2025.110146","url":null,"abstract":"<div><div>Real-world applications, particularly in the medical field, often handle irregular time signals (ITS) with non-uniform intervals between measurements. These irregularities arise due to missing data, inconsistent sampling frequencies, and multi-sensor signals from different sources. Predicting outcomes using ISMTS is complex, especially when missing data is involved.</div><div>This paper introduces the Binomial Gate LSTM (BigLSTM), a modular Recurrent Neural Network model designed to process ISMTS. Built on the LSTM network, BigLSTM integrates techniques for handling irregular time intervals and multiple sampling rates by injecting information redundancy. BigLSTM comprises five interconnected modules. Four are dedicated to information processing: Information Distribution, Central Computing, Predictive, and Time Axis Processing Modules. These modules ensure the redundancy of system, making it tolerant to missing data. The fifth module, LSTM Cells On/Off Control, manages the internal operations of the network.</div><div>BigLSTM was tested on a critical clinical problem: predicting endotracheal obstruction in COVID-19 patients in intensive care units using ventilatory signals from 96 patients. BigLSTM achieved a mean validation mean squared error (MSE) of 0.028 for patients with obstructions and 0.2 for the entire dataset.</div><div>Additionally, we analysed the prediction tendencies of the system, finding an advance trend of 3.87 days and a delay trend of 2.15 days for distant predictions (7 days), with shorter intervals for near predictions (48 h). BigLSTM provided an obstruction prediction, in the short-term, not earlier than the next 10.64 h, and not later than the next 6.8 days, with a confidence percentage of 95%, indicating its effectiveness in handling irregular time series data.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"192 ","pages":"Article 110146"},"PeriodicalIF":7.0,"publicationDate":"2025-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143863804","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Non-stationary components in Electrograms localize arrhythmogenic substrates in a 3D model of human atria","authors":"Alejandro Gómez-Echavarría ,&nbsp;Juan P. Ugarte ,&nbsp;Catalina Tobón","doi":"10.1016/j.compbiomed.2025.110126","DOIUrl":"10.1016/j.compbiomed.2025.110126","url":null,"abstract":"<div><div>Catheter ablation, as a treatment for atrial fibrillation (AF), often yields low success rates in the advanced stages of the arrhythmia. Ablation procedures are guided by atrial mapping using electrogram (EGM) signals, which reflect local electrical activations. The primary goal is to identify arrhythmogenic mechanisms, such as rotors, to serve as ablation targets. Given the chaotic nature of AF propagation, these electrical activations occur at variable rates. This work introduces a novel signal processing approach based on the fractional Fourier transform (FrFT) to characterize the non-stationary content in EGM signals. A 3D biophysical and anatomical model of human atria was used to simulate AF, and unipolar EGMs were calculated. The FrFT-based algorithm was applied to all EGM signals, estimating the optimal FrFT order to capture linear frequency modulations. Electroanatomical maps of these optimal FrFT orders were generated. Results revealed that the AF EGMs exhibit non-stationarity, which can be characterized using the FrFT. Rotors displayed a distinct pattern of non-stationarity, allowing for dynamic tracking, while transient mechanisms were identifiable through variations in the FrFT order, showing different patterns than those of rotors. As a generalization of the classical Fourier analysis, FrFT mapping offers clinically interpretable insights into the rate of change in EGM frequency content over time. This method proves valuable for characterizing AF spatiotemporal dynamics by leveraging the non-stationary information inherent in fibrillatory propagation.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"192 ","pages":"Article 110126"},"PeriodicalIF":7.0,"publicationDate":"2025-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143854513","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"FedSynthCT-Brain: A federated learning framework for multi-institutional brain MRI-to-CT synthesis","authors":"Ciro Benito Raggio ,&nbsp;Mathias Krohmer Zabaleta ,&nbsp;Nils Skupien ,&nbsp;Oliver Blanck ,&nbsp;Francesco Cicone ,&nbsp;Giuseppe Lucio Cascini ,&nbsp;Paolo Zaffino ,&nbsp;Lucia Migliorelli ,&nbsp;Maria Francesca Spadea","doi":"10.1016/j.compbiomed.2025.110160","DOIUrl":"10.1016/j.compbiomed.2025.110160","url":null,"abstract":"<div><div>The generation of Synthetic Computed Tomography (sCT) images has become a pivotal methodology in modern clinical practice, particularly in the context of Radiotherapy (RT) treatment planning. The use of sCT enables the calculation of doses, pushing towards Magnetic Resonance Imaging (MRI) guided radiotherapy treatments. Moreover, with the introduction of MRI-Positron Emission Tomography (PET) hybrid scanners, the derivation of sCT from MRI can improve the attenuation correction of PET images.</div><div>Deep learning methods for MRI-to-sCT have shown promising results, but their reliance on single-centre training dataset limits generalisation capabilities to diverse clinical settings. Moreover, creating centralised multi-centre datasets may pose privacy concerns. To address the aforementioned issues, we introduced FedSynthCT-Brain, an approach based on the Federated Learning (FL) paradigm for MRI-to-sCT in brain imaging. This is among the first applications of FL for MRI-to-sCT, employing a cross-silo horizontal FL approach that allows multiple centres to collaboratively train a U-Net-based deep learning model. We validated our method using real multicentre data from four European and American centres, simulating heterogeneous scanner types and acquisition modalities, and tested its performance on an independent dataset from a centre outside the federation.</div><div>In the case of the unseen centre, the federated model achieved a median Mean Absolute Error (MAE) of 102.0 HU across 23 patients, with an interquartile range of 96.7–110.5 HU. The median (interquartile range) for the Structural Similarity Index (SSIM) and the Peak Signal to Noise Ratio (PNSR) were 0.89 (0.86–0.89) and 26.58 (25.52–27.42), respectively.</div><div>The analysis of the results showed acceptable performances of the federated approach, thus highlighting the potential of FL to enhance MRI-to-sCT to improve generalisability and advancing safe and equitable clinical applications while fostering collaboration and preserving data privacy.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"192 ","pages":"Article 110160"},"PeriodicalIF":7.0,"publicationDate":"2025-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143859194","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Prognostication of zooplankton-driven cholera pathoepidemiological Dynamics: Novel Bayesian-regularized deep NARX neuroarchitecture","authors":"Muhammad Junaid Ali Asif Raja ,&nbsp;Adil Sultan ,&nbsp;Chuan-Yu Chang ,&nbsp;Chi-Min Shu ,&nbsp;Adiqa Kausar Kiani ,&nbsp;Muhammad Shoaib ,&nbsp;Muhammad Asif Zahoor Raja","doi":"10.1016/j.compbiomed.2025.110197","DOIUrl":"10.1016/j.compbiomed.2025.110197","url":null,"abstract":"<div><h3>Background</h3><div>Cholera outbreaks pose significant health concerns, particularly through freshwater contamination through zooplankton serving as reservoirs for <em>Vibrio Cholerae</em>. Understanding these complex interactions within the aquatic ecosystem through mathematical modeling regimes may help us predict and prevent the spread of Cholera disease spread in affected regions.</div></div><div><h3>Method</h3><div>In this study, an innovative <u>B</u>ayesian <u>r</u>egularized <u>d</u>eep <u>n</u>onlinear <u>a</u>utoregressive e<u>x</u>ogenous (BRDNARX) neural networks are employed to model the intricate dynamics of <u>Z</u>ooplankton-<u>D</u>riven <u>C</u>holera <u>D</u>isease <u>T</u>ransmission (ZDCDT) system. The cholera epidemic propagation through freshwater contamination is uncovered with analysis on densities of phytoplankton, vibrio cholerae carrying zooplankton, human population vector and microbial pathogen vector populous in the marine biosphere. Synthetic data for the ZDCDT is presented for diverse simulations using a modified Adams-Bashforth-Moulton predictor corrector numerical scheme. Subsequently, these temporal data sequences are preprocessed for the novel BRDNARX computing paradigm with an exhaustive assessment on mean square error iterative convergence plots, error histogram charts, regression index reports, input-error crosscorrelation charts, error autocorrelation charts, and time-series response dynamics.</div></div><div><h3>Results and conclusions</h3><div>Comparative absolute error analysis with reference numerical solution adheres to diminutive disparities of range 10⁻³ to 10⁻⁹. Finally, BRDNARX neurostructures are reconfigured for predictive analysis of ZDCDT system in terms of single and multi-step ahead predictors with mean square error outcomes that range from 10⁻⁹ to 10⁻¹¹. This establishes the efficacy of BRDNARX in correctly adhering to the intricacies of the zooplankton-driven cholera pathoepidemiological dynamics with precise forward prognostication.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"192 ","pages":"Article 110197"},"PeriodicalIF":7.0,"publicationDate":"2025-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143859195","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Point cloud-guided ultrasound robotic scanning path planning for the kidney based on anatomical positioning","authors":"Chunyi Wu ,&nbsp;Hu Lan ,&nbsp;Jijie Ma ,&nbsp;Xinhui Li ,&nbsp;Jianming Wen","doi":"10.1016/j.compbiomed.2025.110191","DOIUrl":"10.1016/j.compbiomed.2025.110191","url":null,"abstract":"<div><div>In order to tackle the path planning difficulties faced by ultrasound robots during renal ultrasonography procedures, this research integrates anatomical positioning with point cloud processing technology, proposing a specialized path planning algorithm tailored for renal ultrasonography. The study employs a depth camera to capture and preprocess three-dimensional point cloud data from the surface of the body. The orientation of the human model is enhanced through the automated identification of the vertebral line and the narrowest section of the waist, while the costovertebral angle is utilized to formulate an accurate scanning trajectory aimed at imaging the kidneys. To evaluate the efficacy of the algorithm, this study validates its path planning capabilities through simulation experiments and enables the robot to perform automatic scans on real human subjects. The experimental results indicate that the algorithm can effectively plan the desired scanning path across different positions and poses on standard human models, allowing the ultrasound robot to successfully acquire ultrasound images of the kidneys from volunteers under real physiological conditions. The success rate is 93.33 % and the average scanning time is 4.28 s. Experimental results demonstrate that the proposed path planning method, by incorporating anatomical features, accelerates and simplifies kidney localization in 2D ultrasound imaging. Utilizing point cloud technology, it achieves low-cost, fully automated scanning, rapidly and accurately detecting human feature lines, reducing dependence on external markers, and effectively addressing the challenges posed by different poses through pose correction. These advancements provide a strong foundation for the autonomous operation of ultrasound robots.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"192 ","pages":"Article 110191"},"PeriodicalIF":7.0,"publicationDate":"2025-04-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143852124","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Deep transfer learning-based decoder calibration for intracortical brain-machine interfaces","authors":"Xiao Li ,&nbsp;Xianxin Dong ,&nbsp;Jun Wang ,&nbsp;Haodong Mao ,&nbsp;Xikai Tu ,&nbsp;Wei Li ,&nbsp;Jiping He ,&nbsp;Qiang Li ,&nbsp;Peng Zhang","doi":"10.1016/j.compbiomed.2025.110231","DOIUrl":"10.1016/j.compbiomed.2025.110231","url":null,"abstract":"<div><div>Intracortical brain-machine interfaces (iBMIs) aim to establish a communication path between the brain and external devices. However, in the daily use of iBMIs, the non-stationarity of recorded neural signals necessitates frequent recalibration of the iBMI decoder to maintain decoding performance, which requires collecting and labeling a large amount of new data. To address this challenge and minimize the time needed for decoder recalibration, we proposed an active learning domain adversarial neural network (AL-DANN). This model leveraged a substantial volume of historical data alongside a small amount of current data (four samples per category) to calibrate the decoder. By incorporating domain adversarial and active learning strategies, the model effectively transferred knowledge from historical data to new data, reducing the demand for new samples. We validated the proposed method using neural signals recorded from three monkeys performing different movements in a classification task or a regression task. The results showed that the AL-DANN outperformed existing state-of-the-art methods. Impressively, it required only four new samples per category for decoder recalibration, leading to a recalibration time reduction of over 80 %. To our knowledge, this is the first study to incorporate deep transfer learning into iBMI decoder calibration, highlighting the significant potential of applying deep learning technologies in iBMIs.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"192 ","pages":"Article 110231"},"PeriodicalIF":7.0,"publicationDate":"2025-04-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143854512","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Intraoperative measurements in stapedotomy using 3D stereo imaging for optimal prosthesis length selection","authors":"Eric L. Wisotzky ,&nbsp;Felix Lausch ,&nbsp;Linus L. Kienle ,&nbsp;Sara van Bonn-Ytrehus ,&nbsp;Anna Hilsmann ,&nbsp;Peter Eisert ,&nbsp;Sebastian P. Schraven ,&nbsp;Robert Mlynski","doi":"10.1016/j.compbiomed.2025.110233","DOIUrl":"10.1016/j.compbiomed.2025.110233","url":null,"abstract":"<div><div>The application of 3D stereoscopic imaging in surgery represents an innovative approach for anatomical measurements without the use of ionizing radiation or tools contacting the anatomy. Achieving precise intraoperative measurements is crucial in microsurgery, yet conventional methods often lack accuracy due to technical limitations in microscopic zoom lens systems. This study investigates 3D imaging within a digital microscope for stapedotomy, focusing on its accuracy and clinical applicability in selecting optimal prosthesis lengths. We present an optimized calibration scheme for stereoscopic zoom-focus systems across all focus settings, particularly at high magnification levels. A cohort of 23 patients underwent stapedotomy with stereo imaging for landmark annotation and prosthesis measurement. Our calibration method ensured sub-millimeter accuracy, achieving an average deviation of 0.2097 ± 0.1598 mm. The findings demonstrated a significant correlation between insertion depth and postoperative audiological outcomes. Audiological evaluations revealed a mean air-bone gap improvement of 19.1 dB [HL], validating the method's clinical efficacy. This technique offers a radiation-free, efficient alternative to conventional methods, integrating seamlessly into surgical workflows. The study highlights the potential of this imaging modality to enhance surgical precision and patient outcomes, setting the stage for future advancements in automated real-time measurements and broader clinical validation.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"191 ","pages":"Article 110233"},"PeriodicalIF":7.0,"publicationDate":"2025-04-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143850801","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Sharper insights: Adaptive ellipse-template for robust fovea localization in challenging retinal landscapes","authors":"Jyoti Prakash Medhi ,&nbsp;Nirmala S.R. ,&nbsp;Kuntala Borah ,&nbsp;Debasish Bhattacharjee ,&nbsp;Samarendra Dandapat","doi":"10.1016/j.compbiomed.2025.110125","DOIUrl":"10.1016/j.compbiomed.2025.110125","url":null,"abstract":"<div><div>Automated identification of retinal landmarks, particularly the fovea is crucial for diagnosing diabetic retinopathy and other ocular diseases. But accurate identification is challenging due to varying contrast, color irregularities, anatomical structure and the presence of lesions near the macula in fundus images. Existing methods often struggle to maintain accuracy in these complex conditions, particularly when lesions obscure vital regions. To overcome these limitations, this paper introduces a novel adaptive ellipse-template-based approach for fovea localization, leveraging mathematical modeling of blood vessel (BV) trajectories and optic disc (OD) positioning. Unlike traditional fixed-template model, our method dynamically adjusts the ellipse parameters based on OD diameter, ensuring a generalized and adaptable template. This flexibility enables consistent detection performance, even in challenging images with significant lesion interference. Extensive validation on ten publicly available databases, including MESSIDOR, DRIVE, DIARETDB0, DIARETDB1, HRF, IDRiD, HEIMED, ROC, GEI, and NETRALAYA, demonstrates a superior detection efficiency of 99.5%. Additionally, the method achieves a low mean Euclidean distance of 13.48 pixels with a standard deviation of 15.5 pixels between the actual and detected fovea locations, highlighting its precision and reliability. The proposed approach significantly outperforms conventional template-based and deep learning methods, particularly in lesion-rich and low-contrast conditions. It is computationally efficient, interpretable, and robust, making it a valuable tool for automated retinal image analysis in clinical settings.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"191 ","pages":"Article 110125"},"PeriodicalIF":7.0,"publicationDate":"2025-04-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143850803","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Neural mass modelling of brain stimulation to Alleviate Schizophrenia biomarkers in brain rhythms","authors":"Swapna Sasi ,&nbsp;Basabdatta Sen Bhattacharya ,&nbsp;Vanteemar S. Sreeraj ,&nbsp;Ganesan Venkatasubramanian","doi":"10.1016/j.compbiomed.2025.110190","DOIUrl":"10.1016/j.compbiomed.2025.110190","url":null,"abstract":"<div><div>We present a neural mass model (NMM) of the brain thalamo-cortico-thalamic (TCT) network to understand the effectiveness of non-invasive treatment with transcranial Direct Current Stimulation (tDCS) in reversing the anomalous electroencephalogram (EEG) oscillations in Schizophrenia. Our TCT NMM consists of twelve neural populations representing the thalamus and cortex modules of the visual pathway connected in a closed loop; the synaptic pathways are modelled with a 3-state kinetic framework allowing the inclusion of the slow excitatory N-methyl-D-aspartate-receptors (NMDAR). Indeed, a popular hypothesis in Schizophrenia is the hypofunction of the Glutamatergic neurotransmitter receptors, NMDAR, associated with the inhibitory Gamma-amino-butyric-acid (GABA-)ergic populations in the cortex, leading to anomalous brain oscillations. Experimental studies simulate the EEG conditions in Schizophrenia by administering sub-anesthetic dosage of Ketamine, which blocks NMDAR channels at the Magnesium binding sites. We could simulate the Ketamine-induced NMDAR channel blocking by varying the Magnesium concentration in the 3-state synaptic models of appropriate pathways. Our results show Ketamine-induced increased excitatory behaviour in the model output; the changes in the <span><math><mi>γ</mi></math></span> and <span><math><mi>σ</mi></math></span> band oscillations conform to experimental studies. A model to factor in the neuroplasticity effects of applying tDCS (after (Riedinger and Hutt, 2022)) is interfaced with the TCT NMM. Informed by experimental literature, the simulated extrinsic current induced by tDCS is set to affect the plasticity in selected pathways. With appropriate parameterisation, we could simulate the reversal of the Ketamine-induced altered EEG oscillations. Overall, our <em>in silico</em> study emphasises the potential of NMM in predicting protocols for tDCS towards effective personalised treatment of Schizophrenia.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"192 ","pages":"Article 110190"},"PeriodicalIF":7.0,"publicationDate":"2025-04-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143852123","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}