Pub Date : 2025-07-01DOI: 10.1016/j.bbe.2025.08.002
Fabio Arthur Soares Araújo , Robson Luis Oliveira de Amorim , Marly Guimarães Fernandes Costa , Henrique Oliveira Martins , Cicero Ferreira Fernandes Costa Filho
Traumatic Brain Injury (TBI) remains a leading cause of morbidity and mortality worldwide, with significant disparities in outcomes influenced by regional healthcare access and infrastructure. This study evaluates the performance and generalizability of machine learning models for predicting 14-day mortality in TBI patients using datasets from two distinct Brazilian regions: São Paulo, an urban center, and Manaus, an isolated urban center with unique logistical challenges. To our knowledge, this research represents the first cross-validation of predictive models across two datasets within the same country, underscoring the critical need for localized approaches in TBI research. Our findings indicate that while convolutional neural network (CNN)-based models achieved high performance, with an area under the curve (AUC) of 0.90 in São Paulo and 0.93 in Manaus, the best model from São Paulo exhibited a strikingly low AUC when applied to the Manaus dataset. The incorporation of context-specific features, such as pandemic-related variables and time from trauma to admission, significantly enhanced model accuracy, with the Manaus model reaching an impressive AUC of 0.98. Notably, the study highlights key regional differences in predictors of mortality, with hypoxia and hypotension being more critical in Manaus, emphasizing the importance of tailoring predictive models to local contexts. These regionally important variables identified in the ML prediction model may inform quality-improvement priorities and further research in these settings. Our results indicate that CNN-based models have the potential to enhance mortality predictions for patients with traumatic brain injury (TBI).
{"title":"Evaluating the generalization of machine learning models for predicting 14-day mortality in traumatic brain injury patients","authors":"Fabio Arthur Soares Araújo , Robson Luis Oliveira de Amorim , Marly Guimarães Fernandes Costa , Henrique Oliveira Martins , Cicero Ferreira Fernandes Costa Filho","doi":"10.1016/j.bbe.2025.08.002","DOIUrl":"10.1016/j.bbe.2025.08.002","url":null,"abstract":"<div><div>Traumatic Brain Injury (TBI) remains a leading cause of morbidity and mortality worldwide, with significant disparities in outcomes influenced by regional healthcare access and infrastructure. This study evaluates the performance and generalizability of machine learning models for predicting 14-day mortality in TBI patients using datasets from two distinct Brazilian regions: São Paulo, an urban center, and Manaus, an isolated urban center with unique logistical challenges. To our knowledge, this research represents the first cross-validation of predictive models across two datasets within the same country, underscoring the critical need for localized approaches in TBI research. Our findings indicate that while convolutional neural network (CNN)-based models achieved high performance, with an area under the curve (AUC) of 0.90 in São Paulo and 0.93 in Manaus, the best model from São Paulo exhibited a strikingly low AUC when applied to the Manaus dataset. The incorporation of context-specific features, such as pandemic-related variables and time from trauma to admission, significantly enhanced model accuracy, with the Manaus model reaching an impressive AUC of 0.98. Notably, the study highlights key regional differences in predictors of mortality, with hypoxia and hypotension being more critical in Manaus, emphasizing the importance of tailoring predictive models to local contexts. These regionally important variables identified in the ML prediction model may inform quality-improvement priorities and further research in these settings. Our results indicate that CNN-based models have the potential to enhance mortality predictions for patients with traumatic brain injury (TBI).</div></div>","PeriodicalId":55381,"journal":{"name":"Biocybernetics and Biomedical Engineering","volume":"45 3","pages":"Pages 560-571"},"PeriodicalIF":6.6,"publicationDate":"2025-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144826418","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}
Decoding pre-movement intention is crucial in developing a brain-computer interface (BCI) for neuro-rehabilitation robotic systems. However, the weak amplitude and non-smooth characteristics of EEG signals lead to the inability of existing methods to achieve the accuracy for proper applications. This study proposed a novel pre-movement intention decoding network framework to improve accuracy by extracting and optimizing the deep spatio-temporal features of EEG signals.
Methods
A deep spatio-temporal neural network structure was constructed based on the brain intention generation mechanism and its movement expression. The collected multi-channel EEG data were reorganized into brain topographic distributions, after the initial extraction of the features and optimization using the coordinate attention mechanism, a 3-layer dense block with two bi-directional gated recirculation units was designed to effectively extract the deep spatial and temporal features, further decoding the pre-movement intention efficiently.
Results
The experimental results showed an average accuracy of 95.51 ± 1.79 % for healthy subjects and 90.48 ± 2.90 % for stroke survivors in decoding pre-movement intention. All evaluation indexes are excellent. Pseudo-online testing showed the average TPR was 95.45 ± 3.80 % and 90.71 ± 7.77 % for healthy subjects and stroke survivors, respectively, and the latency was −1965 ± 48 ms and −1974 ± 36 ms. The results of the ablation and comparative analysis showed that the proposed framework is justified and its decoding capability outperforms other state-of-the-art algorithms.
Conclusion
The method proposed in this study has high decoding accuracy and good online performance in pre-movement intention decoding based on EEG signals, which lays the foundation for further neuro-rehabilitation robotic systems.
{"title":"Deep spatio-temporal features optimised fusion with coordinate attention mechanism for EEG lower limb pre-movement intention decoding","authors":"Runlin Dong , Xiaodong Zhang , Zhengzheng Zhou , Wenyu Zha , Aibin Zhu","doi":"10.1016/j.bbe.2025.06.004","DOIUrl":"10.1016/j.bbe.2025.06.004","url":null,"abstract":"<div><h3>Background and Objective</h3><div>Decoding pre-movement intention is crucial in developing a brain-computer interface (BCI) for neuro-rehabilitation robotic systems. However, the weak amplitude and non-smooth characteristics of EEG signals lead to the inability of existing methods to achieve the accuracy for proper applications. This study proposed a novel pre-movement intention decoding network framework to improve accuracy by extracting and optimizing the deep spatio-temporal features of EEG signals.</div></div><div><h3>Methods</h3><div>A deep spatio-temporal neural network structure was constructed based on the brain intention generation mechanism and its movement expression. The collected multi-channel EEG data were reorganized into brain topographic distributions, after the initial extraction of the features and optimization using the coordinate attention mechanism, a 3-layer dense block with two bi-directional gated recirculation units was designed to effectively extract the deep spatial and temporal features, further decoding the pre-movement intention efficiently.</div></div><div><h3>Results</h3><div>The experimental results showed an average accuracy of 95.51 ± 1.79 % for healthy subjects and 90.48 ± 2.90 % for stroke survivors in decoding pre-movement intention. All evaluation indexes are excellent. Pseudo-online testing showed the average TPR was 95.45 ± 3.80 % and 90.71 ± 7.77 % for healthy subjects and stroke survivors, respectively, and the latency was −1965 ± 48 ms and −1974 ± 36 ms. The results of the ablation and comparative analysis showed that the proposed framework is justified and its decoding capability outperforms other state-of-the-art algorithms.</div></div><div><h3>Conclusion</h3><div>The method proposed in this study has high decoding accuracy and good online performance in pre-movement intention decoding based on EEG signals, which lays the foundation for further neuro-rehabilitation robotic systems.</div></div>","PeriodicalId":55381,"journal":{"name":"Biocybernetics and Biomedical Engineering","volume":"45 3","pages":"Pages 515-527"},"PeriodicalIF":5.3,"publicationDate":"2025-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144702745","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}
Pub Date : 2025-07-01DOI: 10.1016/j.bbe.2025.08.005
Sayedmohsen Mortazavi Najafabadi , Dariusz Grzelczyk , Mohammed N. Ashtiani
Age, neurodegenerative diseases, diabetes, and sports injuries can all impair proprioception, i.e. a crucial sensory feedback system for balance control and gait. The purpose of this study was to assess the validity and reliability of a recently constructed device for measuring proprioceptive function. Forty-seven participants, comprising 26 younger healthy adults (20–40 years) and 21 older adults (> 65 years), were evaluated. The ankle’s sense of motion (SoM) sensitivity and sense of position (SoP, active/passive) acuity were measured by the device. The Intraclass Correlation Coefficient (ICC) and the Area Under the Receiver Operating Characteristic Curve (AUC-ROC) were used as indicators of reliability and validity. The results showed excellent reliability for SoM sensitivity in dorsiflexion (ICC = 0.985 for younger, 0.98 for older) and plantarflexion (ICC = 0.972 for younger, 0.982 for older). High reliability was also observed in passive SoP acuity (ICC = 0.825 – 0.989). However, the reliability of the active SoP acuity method was poor to moderate. Strong discriminative validity was demonstrated by the AUC-ROC values, with SoM sensitivity distinguishing between younger and older participants with an accuracy of over 91 %. Bland-Altman analysis revealed tighter agreement for SoM sensitivity (18 to 40 % of the device precision) than passive SoP acuity (70 to 90 % of the device precision), as well as minimal systematic bias (−0.03 to −0.01 degrees) to show interday test–retest reliability. According to these results, the device is valid for evaluating proprioceptive function, particularly SoM sensitivity, and it may be useful in clinical and research settings.
{"title":"A novel device for proprioceptive acuity measurement: Validity and reliability analysis in young and older adults","authors":"Sayedmohsen Mortazavi Najafabadi , Dariusz Grzelczyk , Mohammed N. Ashtiani","doi":"10.1016/j.bbe.2025.08.005","DOIUrl":"10.1016/j.bbe.2025.08.005","url":null,"abstract":"<div><div>Age, neurodegenerative diseases, diabetes, and sports injuries can all impair proprioception, i.e. a crucial sensory feedback system for balance control and gait. The purpose of this study was to assess the validity and reliability of a recently constructed device for measuring proprioceptive function. Forty-seven participants, comprising 26 younger healthy adults (20–40 years) and 21 older adults (> 65 years), were evaluated. The ankle’s sense of motion (SoM) sensitivity and sense of position (SoP, active/passive) acuity were measured by the device. The Intraclass Correlation Coefficient (ICC) and the Area Under the Receiver Operating Characteristic Curve (AUC-ROC) were used as indicators of reliability and validity. The results showed excellent reliability for SoM sensitivity in dorsiflexion (ICC = 0.985 for younger, 0.98 for older) and plantarflexion (ICC = 0.972 for younger, 0.982 for older). High reliability was also observed in passive SoP acuity (ICC = 0.825 – 0.989). However, the reliability of the active SoP acuity method was poor to moderate. Strong discriminative validity was demonstrated by the AUC-ROC values, with SoM sensitivity distinguishing between younger and older participants with an accuracy of over 91 %. Bland-Altman analysis revealed tighter agreement for SoM sensitivity (18 to 40 % of the device precision) than passive SoP acuity (70 to 90 % of the device precision), as well as minimal systematic bias (−0.03 to −0.01 degrees) to show interday test–retest reliability. According to these results, the device is valid for evaluating proprioceptive function, particularly SoM sensitivity, and it may be useful in clinical and research settings.</div></div>","PeriodicalId":55381,"journal":{"name":"Biocybernetics and Biomedical Engineering","volume":"45 3","pages":"Pages 580-589"},"PeriodicalIF":6.6,"publicationDate":"2025-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144903181","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}
The continuous negative external pressure (cNEP) applied on the submental surface is a method of non-surgical treatment for obstructive sleep apnea (OSA), which can effectively widen the airway in some OSA patients. However, it cannot effectively improve airway collapse in obese patients and its mechanism remains unclear. In this study, we aim to analyze the reasons for the ineffectiveness of cNEP treatment in OSA patients with obesity. Based on magnetic resonance imaging (MRI), three-dimensional models of the head and neck were constructed for a healthy subject, an OSA patient with enlarged tongue, and an OSA patient with the tongue adjusted to normal size. By performing the one step staggered fluid–structure interaction numerical simulations, we analyzed the collapse of the airway in these three models under the influence of cNEP. Restoring the tongue to normal size in the OSA patient significantly improves the airway critical closing pressure under cNEP treatment compared to the patient with enlarged tongue. The enlargement of the tongue in the OSA patient hindered the widening of the velopharyngeal airway under the action of cNEP. The numerical results reveal that cNEP treatment can effectively widen the laryngopharyngeal airway, thus providing a potential therapeutic option for OSA patients with laryngopharyngeal obstruction. Tongue enlargement in OSA patients is a critical factor influencing the efficacy of cNEP treatment. This study reveals the reasons for cNEP treatment failure in obese patients and the potential value of cNEP targeted therapy.
{"title":"The impact of tongue size on submental negative pressure treatment of airway obstruction revealed by fluid-structure interaction simulations","authors":"Yuhang Tian , Huahui Xiong , Hui Tong , Changjin Ji , Xiaoqing Huang , Yaqi Huang","doi":"10.1016/j.bbe.2025.08.004","DOIUrl":"10.1016/j.bbe.2025.08.004","url":null,"abstract":"<div><div>The continuous negative external pressure (cNEP) applied on the submental surface is a method of non-surgical treatment for obstructive sleep apnea (OSA), which can effectively widen the airway in some OSA patients. However, it cannot effectively improve airway collapse in obese patients and its mechanism remains unclear. In this study, we aim to analyze the reasons for the ineffectiveness of cNEP treatment in OSA patients with obesity. Based on magnetic resonance imaging (MRI), three-dimensional models of the head and neck were constructed for a healthy subject, an OSA patient with enlarged tongue, and an OSA patient with the tongue adjusted to normal size. By performing the one step staggered fluid–structure interaction numerical simulations, we analyzed the collapse of the airway in these three models under the influence of cNEP. Restoring the tongue to normal size in the OSA patient significantly improves the airway critical closing pressure under cNEP treatment compared to the patient with enlarged tongue. The enlargement of the tongue in the OSA patient hindered the widening of the velopharyngeal airway under the action of cNEP. The numerical results reveal that cNEP treatment can effectively widen the laryngopharyngeal airway, thus providing a potential therapeutic option for OSA patients with laryngopharyngeal obstruction. Tongue enlargement in OSA patients is a critical factor influencing the efficacy of cNEP treatment. This study reveals the reasons for cNEP treatment failure in obese patients and the potential value of cNEP targeted therapy.</div></div>","PeriodicalId":55381,"journal":{"name":"Biocybernetics and Biomedical Engineering","volume":"45 3","pages":"Pages 590-599"},"PeriodicalIF":6.6,"publicationDate":"2025-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144911947","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}
Pub Date : 2025-06-27DOI: 10.1016/j.bbe.2025.06.001
Marcin Sińczuk , Jacek Rogala , Piotr Bogorodzki
This study explores the importance of scanner-specific calibration measurements for Magnetic Resonance Spectroscopy Thermometry (MRST) in human brain temperature estimations. Data acquisition was conducted on a 3-T GE scanner. Calibration constants for the water-chemical shift were obtained using a temperature-controlled phantom containing an aqueous solution of N-acetyl aspartate (NAA), Creatine (Cr), and Choline (Cho), and data from three different research groups using the same metabolites. Temperatures were estimated utilizing correlation of water chemical shift with NAA, Cr and Cho. For data acquisition, commercially available single-voxel point-resolved spectroscopy (PRESS) sequences were used for calibrations and in vivo temperature estimations. Each sequence included spectras without (WU) and with (WS) water suppression. In vivo study consisted of two PRESS sequences, one before and one after extensive 30-minute fMRI task acquisition. Significant differences were found between absolute brain temperatures measured using scanner-specific calibrations and those from other researchers, varying from −0.68 °C to + 0.37 °C for NAA, −0.92 °C to 0.37 °C for Cr, and −0.78 °C to 0.7 °C for Cho. Each method reported a similar temperature decrease of −0.26 ∓ 0.03 °C between before and after fMRI measurements. These findings suggest that while absolute temperatures from non-scanner specific calibrations may be inaccurate, comparative estimates are valid.
{"title":"MRS thermometry – Importance of scanner-specific calibrations for accurate brain temperature estimations","authors":"Marcin Sińczuk , Jacek Rogala , Piotr Bogorodzki","doi":"10.1016/j.bbe.2025.06.001","DOIUrl":"10.1016/j.bbe.2025.06.001","url":null,"abstract":"<div><div>This study explores the importance of scanner-specific calibration measurements for Magnetic Resonance Spectroscopy Thermometry (MRST) in human brain temperature estimations. Data acquisition was conducted on a 3-T GE scanner. Calibration constants for the water-chemical shift were obtained using a temperature-controlled phantom containing an aqueous solution of N-acetyl aspartate (NAA), Creatine (Cr), and Choline (Cho), and data from three different research groups using the same metabolites. Temperatures were estimated utilizing correlation of water chemical shift with NAA, Cr and Cho. For data acquisition, commercially available single-voxel point-resolved spectroscopy (PRESS) sequences were used for calibrations and in vivo temperature estimations. Each sequence included spectras without (WU) and with (WS) water suppression. In vivo study consisted of two PRESS sequences, one before and one after extensive 30-minute fMRI task acquisition. Significant differences were found between absolute brain temperatures measured using scanner-specific calibrations and those from other researchers, varying from −0.68 °C to + 0.37 °C for NAA, −0.92 °C to 0.37 °C for Cr, and −0.78 °C to 0.7 °C for Cho. Each method reported a similar temperature decrease of −0.26 ∓ 0.03 °C between before and after fMRI measurements. These findings suggest that while absolute temperatures from non-scanner specific calibrations may be inaccurate, comparative estimates are valid.</div></div>","PeriodicalId":55381,"journal":{"name":"Biocybernetics and Biomedical Engineering","volume":"45 3","pages":"Pages 451-456"},"PeriodicalIF":5.3,"publicationDate":"2025-06-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144489482","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}
Pub Date : 2025-06-27DOI: 10.1016/j.bbe.2025.06.002
Natalia Kowalczyk, Jacek Rumiński, Magdalena Mazur-Milecka
The COVID-19 pandemic has underscored the importance of wearing facial masks and monitoring respiratory health to prevent the spread of the virus. In this study, we developed a model for segmenting facial masks in thermal images. We applied the model to segment face masks in different conditions, including a person walking toward the observing camera. The segmented regions were further processed using different erosion masks to analyze the influence of the selected sources on the quality of the estimated respiratory signals. The Signal-to-Noise Ratio (SNR) was used as a quality measure. Additionally, the extracted respiratory signals were compared with two reference signals: binary signals generated by participants who signaled the inhalation phase and pressure signals measured with a respiratory belt. Our findings show a high level of concordance between the respiratory signals derived from the segmented mask region and those from the respiratory belt, validating the effectiveness of thermal imaging for capturing respiratory patterns. Notably, the signal-to-noise ratio (SNR) was higher for the segmented mask than the detection methods used in previous works. Specifically, for the mask segmentation task, the mean SNR improved by 4.3 compared to facial mask detection. The segmentation model achieved a mean Average Precision (mAP) of 0.992 for segmentation tasks and 0.857 mAP at the 50–95 % threshold using the Yolov8 “nano” architecture. This study underscores the potential of thermal imaging for non-invasive respiratory monitoring and highlights the explainability and accuracy of selecting the facial mask region for signal extraction.
{"title":"Improving the quality of respiratory signals extracted from the segmented mask area","authors":"Natalia Kowalczyk, Jacek Rumiński, Magdalena Mazur-Milecka","doi":"10.1016/j.bbe.2025.06.002","DOIUrl":"10.1016/j.bbe.2025.06.002","url":null,"abstract":"<div><div>The COVID-19 pandemic has underscored the importance of wearing facial masks and monitoring respiratory health to prevent the spread of the virus. In this study, we developed a model for segmenting facial masks in thermal images. We applied the model to segment face masks in different conditions, including a person walking toward the observing camera. The segmented regions were further processed using different erosion masks to analyze the influence of the selected sources on the quality of the estimated respiratory signals. The Signal-to-Noise Ratio (SNR) was used as a quality measure. Additionally, the extracted respiratory signals were compared with two reference signals: binary signals generated by participants who signaled the inhalation phase and pressure signals measured with a respiratory belt. Our findings show a high level of concordance between the respiratory signals derived from the segmented mask region and those from the respiratory belt, validating the effectiveness of thermal imaging for capturing respiratory patterns. Notably, the signal-to-noise ratio (SNR) was higher for the segmented mask than the detection methods used in previous works. Specifically, for the mask segmentation task, the mean SNR improved by 4.3 compared to facial mask detection. The segmentation model achieved a mean Average Precision (mAP) of 0.992 for segmentation tasks and 0.857 mAP at the 50–95 % threshold using the Yolov8 “nano” architecture. This study underscores the potential of thermal imaging for non-invasive respiratory monitoring and highlights the explainability and accuracy of selecting the facial mask region for signal extraction.</div></div>","PeriodicalId":55381,"journal":{"name":"Biocybernetics and Biomedical Engineering","volume":"45 3","pages":"Pages 457-468"},"PeriodicalIF":5.3,"publicationDate":"2025-06-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144489483","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}
Pub Date : 2025-06-14DOI: 10.1016/j.bbe.2025.05.011
Víctor Gutiérrez-de Pablo , María Herrero-Tudela , Marina Sandonís-Fernández , Jesús Poza , Aarón Maturana-Candelas , Víctor Rodríguez-González , Miguel Ángel Tola-Arribas , Mónica Cano , Hideyuki Hoshi , Yoshihito Shigihara , Roberto Hornero , Carlos Gómez
Dementia and mild cognitive impairment (MCI) due to Alzheimer’s disease (AD) are neurological pathologies associated with disruptions in brain electromagnetic activity, typically studied using magnetoencephalography (MEG) and electroencephalography (EEG). To quantify diverse brain properties, different families of parameters can be computed from MEG and EEG (i.e., spectral, non-linear, morphological, functional connectivity, or network structure and organisation). However, studying these characteristics separately overlooks the complex nature of brain activity. Integrative frameworks can be useful to unveil the intricate neurophysiological fingerprint, as well as to characterise pathological conditions comprehensively. To that purpose, data fusion methodologies are crucial, despite their interpretational challenges. In this study, Machine Learning (ML) models were trained to discriminate between groups of severity, whereas the SHapley Additive eXplanations (SHAP) algorithm was afterwards utilised to assess the relevance of the input characteristics into the output classification. Three databases were analysed: MEG (55 healthy controls, HC, 42 MCI patients, and 86 AD patients), EEG1 (51 HC, 52 MCI, and 100 AD), and EEG2 (45 HC, 69 MCI, and 82 AD). The best results for the three-class classification problem were obtained by Gradient Boosting for the MEG database: 3-class Cohen’s kappa coefficient of 0.5452 and accuracy of 72.63 %. Afterwards, using SHAP on Gradient Boosting, it has been shown that spectral features were identified as highly relevant across all databases. Furthermore, morphology measures presented high relevance for the MEG database, whereas EEG1 and EEG2 databases showed functional connectivity and multiplex organisation measures, respectively, as relevant subgroups of parameters. Finally, commonly relevant features across databases were selected using SHAP to generate the neurophysiological fingerprints of AD and MCI. This study highlights the relevance of different MEG and EEG parameters in characterising neurological pathologies. The proposed framework, based on MEG and EEG, can be used to generate interpretable, robust, and accurate neurophysiological fingerprints of AD and MCI.
{"title":"Integrative and interpretable framework to unveil the neurophysiological fingerprint of Alzheimer’s disease and mild cognitive impairment: A machine learning-SHAP approach","authors":"Víctor Gutiérrez-de Pablo , María Herrero-Tudela , Marina Sandonís-Fernández , Jesús Poza , Aarón Maturana-Candelas , Víctor Rodríguez-González , Miguel Ángel Tola-Arribas , Mónica Cano , Hideyuki Hoshi , Yoshihito Shigihara , Roberto Hornero , Carlos Gómez","doi":"10.1016/j.bbe.2025.05.011","DOIUrl":"10.1016/j.bbe.2025.05.011","url":null,"abstract":"<div><div>Dementia and mild cognitive impairment (MCI) due to Alzheimer’s disease (AD) are neurological pathologies associated with disruptions in brain electromagnetic activity, typically studied using magnetoencephalography (MEG) and electroencephalography (EEG). To quantify diverse brain properties, different families of parameters can be computed from MEG and EEG (i.e., spectral, non-linear, morphological, functional connectivity, or network structure and organisation). However, studying these characteristics separately overlooks the complex nature of brain activity. Integrative frameworks can be useful to unveil the intricate neurophysiological fingerprint, as well as to characterise pathological conditions comprehensively. To that purpose, data fusion methodologies are crucial, despite their interpretational challenges. In this study, Machine Learning (ML) models were trained to discriminate between groups of severity, whereas the SHapley Additive eXplanations (SHAP) algorithm was afterwards utilised to assess the relevance of the input characteristics into the output classification. Three databases were analysed: MEG (55 healthy controls, HC, 42 MCI patients, and 86 AD patients), EEG1 (51 HC, 52 MCI, and 100 AD), and EEG2 (45 HC, 69 MCI, and 82 AD). The best results for the three-class classification problem were obtained by Gradient Boosting for the MEG database: 3-class Cohen’s kappa coefficient of 0.5452 and accuracy of 72.63 %. Afterwards, using SHAP on Gradient Boosting, it has been shown that spectral features were identified as highly relevant across all databases. Furthermore, morphology measures presented high relevance for the MEG database, whereas EEG1 and EEG2 databases showed functional connectivity and multiplex organisation measures, respectively, as relevant subgroups of parameters. Finally, commonly relevant features across databases were selected using SHAP to generate the neurophysiological fingerprints of AD and MCI. This study highlights the relevance of different MEG and EEG parameters in characterising neurological pathologies. The proposed framework, based on MEG and EEG, can be used to generate interpretable, robust, and accurate neurophysiological fingerprints of AD and MCI.</div></div>","PeriodicalId":55381,"journal":{"name":"Biocybernetics and Biomedical Engineering","volume":"45 3","pages":"Pages 438-450"},"PeriodicalIF":5.3,"publicationDate":"2025-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144279921","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}
Diffusion tensor imaging (DTI) was used to observe degeneration processes at the microstructural and functional levels in the brains of patients with Parkinson’s disease (PD). Two tensor-based unscented Kalman filter (UKF) was used for analyses of eight regions: the substantia nigra, putamen, caudate nucleus, globus pallidus, primary motor cortex, preprimary motor cortex, supplementary motor area (SMA), presupplementary motor area (pre-SMA), and whole brain of patients with PD (n = 14) and controls (n = 12). We analyzed eight DTI metrics in the entire brain and eight brain regions separately for each hemisphere using univariate and multivariate statistical analysis and their correlation with the clinical parameters. The most affected brain regions in patients with PD were the substantia nigra, pre-SMA, globus pallidus, and caudate nucleus. These results suggest that DTI is an adequate tool for evaluating structural and functional alterations, including inflammation, reduced fiber length, changes in neurite density, axonal growth, demyelination, and axonal damage or loss, in the studied brain regions of patients with PD. The results also revealed a generalized brain degeneration process. In conclusion, DTI can be applied for in vivo studies of the degenerative process and could be considered a complementary method in future studies to improve the accuracy of PD diagnosis.
{"title":"Diffusion tensor imaging technique for studying brain microstructural changes in Parkinson’s disease patients","authors":"Beata Toczylowska , Małgorzata Michałowska , Malgorzata Chalimoniuk , Piotr Ladyzynski , Leszek Krolicki , Urszula Fiszer","doi":"10.1016/j.bbe.2025.05.010","DOIUrl":"10.1016/j.bbe.2025.05.010","url":null,"abstract":"<div><div>Diffusion tensor imaging (DTI) was used to observe degeneration processes at the microstructural and functional levels in the brains of patients with Parkinson’s disease (PD). Two tensor-based unscented Kalman filter (UKF) was used for analyses of eight regions: the substantia nigra, putamen, caudate nucleus, globus pallidus, primary motor cortex, preprimary motor cortex, supplementary motor area (SMA), presupplementary motor area (pre-SMA), and whole brain of patients with PD (n = 14) and controls (n = 12). We analyzed eight DTI metrics in the entire brain and eight brain regions separately for each hemisphere using univariate and multivariate statistical analysis and their correlation with the clinical parameters. The most affected brain regions in patients with PD were the substantia nigra, pre-SMA, globus pallidus, and caudate nucleus. These results suggest that DTI is an adequate tool for evaluating structural and functional alterations, including inflammation, reduced fiber length, changes in neurite density, axonal growth, demyelination, and axonal damage or loss, in the studied brain regions of patients with PD. The results also revealed a generalized brain degeneration process. In conclusion, DTI can be applied for in vivo studies of the degenerative process and could be considered a complementary method in future studies to improve the accuracy of PD diagnosis.</div></div>","PeriodicalId":55381,"journal":{"name":"Biocybernetics and Biomedical Engineering","volume":"45 3","pages":"Pages 426-437"},"PeriodicalIF":5.3,"publicationDate":"2025-05-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144177851","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}
Pub Date : 2025-05-28DOI: 10.1016/j.bbe.2025.05.005
Fatemeh Ghorbani, Andry Rakotonirainy, Mohammed Elhenawy
This study examined the temporal dynamics of face perception using event-related potentials (ERPs) to investigate how familiarity and repetition influence early and late stages of face processing. A generalised linear mixed-effects (GLME) model was employed to assess the amplitude and latency of the P100, N170, and N250 ERP components across three stimulus types (famous, non-famous, and scrambled faces), three repetition conditions (first presentation, immediate repeat, delayed repeat), and two brain hemispheres. The P100 component, associated with early visual processing, showed no significant modulation by stimulus familiarity or repetition, suggesting stable perceptual encoding across conditions. In contrast, N170 and N250 amplitudes were significantly affected by repetition, indicating enhanced neural responses during repeated exposure, particularly in the right hemisphere. Latency analyses revealed that N250 component was also sensitive to repetition timing, with delayed repetitions eliciting shorter response time, implying shifts in processing efficiency and memory engagement. Multivariate time-series decoding further demonstrated higher discriminability between scrambled and familiar faces compared to non-famous faces, particularly during first and delayed repeat conditions. Notably, decoding performance declined for immediate repeats, suggesting reduced neural differentiation during short-interval repetition. These findings provide new insights into how repetition and familiarity modulate the neural underpinnings of face perception, emphasizing the role of temporal dynamics and hemispheric specialization in face processing.
{"title":"The effect of familiarity and repetition on neural activity during visual face perception","authors":"Fatemeh Ghorbani, Andry Rakotonirainy, Mohammed Elhenawy","doi":"10.1016/j.bbe.2025.05.005","DOIUrl":"10.1016/j.bbe.2025.05.005","url":null,"abstract":"<div><div>This study examined the temporal dynamics of face perception using event-related potentials (ERPs) to investigate how familiarity and repetition influence early and late stages of face processing. A generalised linear mixed-effects (GLME) model was employed to assess the amplitude and latency of the P100, N170, and N250 ERP components across three stimulus types (famous, non-famous, and scrambled faces), three repetition conditions (first presentation, immediate repeat, delayed repeat), and two brain hemispheres. The P100 component, associated with early visual processing, showed no significant modulation by stimulus familiarity or repetition, suggesting stable perceptual encoding across conditions. In contrast, N170 and N250 amplitudes were significantly affected by repetition, indicating enhanced neural responses during repeated exposure, particularly in the right hemisphere. Latency analyses revealed that N250 component was also sensitive to repetition timing, with delayed repetitions eliciting shorter response time, implying shifts in processing efficiency and memory engagement. Multivariate time-series decoding further demonstrated higher discriminability between scrambled and familiar faces compared to non-famous faces, particularly during first and delayed repeat conditions. Notably, decoding performance declined for immediate repeats, suggesting reduced neural differentiation during short-interval repetition. These findings provide new insights into how repetition and familiarity modulate the neural underpinnings of face perception, emphasizing the role of temporal dynamics and hemispheric specialization in face processing.</div></div>","PeriodicalId":55381,"journal":{"name":"Biocybernetics and Biomedical Engineering","volume":"45 3","pages":"Pages 399-413"},"PeriodicalIF":5.3,"publicationDate":"2025-05-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144154937","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}
Pub Date : 2025-05-28DOI: 10.1016/j.bbe.2025.04.004
Antonina Pater , Lukasz Roszkowiak , Krzysztof Siemion , Jakub Zak , Karol Deptuch , Anna Korzynska
Population screening in the form of cervical smears was introduced to reduce cervical cancer morbidity. However, the manual evaluation of cervical smears is a labour-intensive and meticulous task. This challenge has led to the development of various computer-aided cell identification methods aimed at improving the quality of evaluations and reducing the time required for slide analysis. These supportive tools for pathologists should adhere to the Bethesda classification system for cell types, facilitating integration with established clinical practices. Despite advances, the automatic classification of clustered cells in cervical smears remains a significant challenge for both automated and semiautomated methods that utilize classical image processing and deep learning techniques.
This study introduces a novel method for fragmenting clusters to improve the classification of clustered cells in digital images of Papanicolaou smears. The proposed method integrates explainable AI and marker-guided watershed segmentation ensuring both interpretability and versatility of an overall pipeline for diagnostician support in smear evaluation. Validation of the method was performed on a recently developed Papanicolaou cytology Bialystok dataset, as well as the publicly available CRIC dataset, demonstrating the method’s generalizability across different datasets.
The cell classification pipeline, enhanced by the proposed declustering method, improved the overall harmonic mean of recall and precision (F1 score) by 13.27 percentage points compared with the results obtained without this additional processing. The improvement in classifying the most critical cell type according to the Bethesda system (HSIL cell class) was even more significant, with an increase of 35.72 percentage points compared with classifying the entire cluster.
{"title":"Cytoplasm and nuclei as a basis for Bethesda cell cluster classification in cervical smears","authors":"Antonina Pater , Lukasz Roszkowiak , Krzysztof Siemion , Jakub Zak , Karol Deptuch , Anna Korzynska","doi":"10.1016/j.bbe.2025.04.004","DOIUrl":"10.1016/j.bbe.2025.04.004","url":null,"abstract":"<div><div>Population screening in the form of cervical smears was introduced to reduce cervical cancer morbidity. However, the manual evaluation of cervical smears is a labour-intensive and meticulous task. This challenge has led to the development of various computer-aided cell identification methods aimed at improving the quality of evaluations and reducing the time required for slide analysis. These supportive tools for pathologists should adhere to the Bethesda classification system for cell types, facilitating integration with established clinical practices. Despite advances, the automatic classification of clustered cells in cervical smears remains a significant challenge for both automated and semiautomated methods that utilize classical image processing and deep learning techniques.</div><div>This study introduces a novel method for fragmenting clusters to improve the classification of clustered cells in digital images of Papanicolaou smears. The proposed method integrates explainable AI and marker-guided watershed segmentation ensuring both interpretability and versatility of an overall pipeline for diagnostician support in smear evaluation. Validation of the method was performed on a recently developed Papanicolaou cytology Bialystok dataset, as well as the publicly available CRIC dataset, demonstrating the method’s generalizability across different datasets.</div><div>The cell classification pipeline, enhanced by the proposed declustering method, improved the overall harmonic mean of recall and precision (F1 score) by 13.27 percentage points compared with the results obtained without this additional processing. The improvement in classifying the most critical cell type according to the Bethesda system (HSIL cell class) was even more significant, with an increase of 35.72 percentage points compared with classifying the entire cluster.</div></div>","PeriodicalId":55381,"journal":{"name":"Biocybernetics and Biomedical Engineering","volume":"45 3","pages":"Pages 414-425"},"PeriodicalIF":5.3,"publicationDate":"2025-05-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144166979","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}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号