EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Decoding basic emotional states through integration of an fNIRS-based brain–computer interface with supervised learning algorithms

Ayşenur Eser iD, Sinem Burcu Erdoğan iD

PLOS One · 2025

DOI: 10.1371/journal.pone.0325850 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PLOS One, 2025. Verfasst von Ayşenur Eser und 1 weiteren Autor:innen. Beteiligte Einrichtung: Faculty of Engineering and Natural Sciences, Department of Biomedical Engineering, Acıbadem Mehmet Ali Aydınlar University, İstanbul, Türkiye. Hauptthemen laut MEDLINE: Brain-Computer Interfaces, Emotions, Supervised Machine Learning. Die Arbeit nennt 54 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Automated detection of emotional states through brain-computer interfaces (BCIs) offers significant potential for enhancing user experiences and personalizing services across domains such as mental health, adaptive learning and interactive entertainment. Within this advancing field, the aim of this study was to test the feasibility of a functional near-infrared spectroscopy (fNIRS)-based BCI system for accurate prediction and objective identification of three fundamental emotional states that involved positive, negative and neutral conditions. Consequently, the efficacy of fNIRS signals in predicting the valence of standardized stimuli from the International Affective Picture System (IAPS) was assessed. fNIRS data were collected from twenty healthy participants while images from the IAPS database were presented. The images varied in both valence (i.e., positive, neutral, negative) and arousal (i.e., high, low) level. Hemodynamic responses of prefrontal cortical (PFC) regions were recorded with a twenty-two channel system. Twenty fNIRS derived time domain features were extracted from HbO time traces of each channel corresponding to each stimulus period. Classification performances of three machine learning algorithms, namely the k-Nearest Neighbors (kNN), Ensemble (Subspace kNN) and Support Vector Machines (SVM), in two class and three class classification of positive, neutral and negative states were evaluated with ten runs of a tenfold cross-validation procedure through splitting the data into test, train and validation groups at each run. Three class classification performances of all algorithms were above 90% in terms of accuracy, sensitivity, specificity, F-1 score and precision metrics while two class accuracy performances of all algorithms were above 93% in terms of each performance metric. The high-performance classification results highlight the potential of fNIRS-based BCI systems for real-time, objective detection of basic emotional states for daily life and clinical applications. fNIRSbased BCIs may show promise for future developments in personalized user experiences and clinical applications due to their practicality and low computational complexity.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Brain-Computer InterfacesEmotionsSupervised Machine LearningHumansSpectroscopy, Near-InfraredMaleFemaleAdultYoung AdultAlgorithmsPrefrontal CortexSupport Vector Machine

Autor:innen und Institutionen

  • Faculty of Engineering and Natural Sciences, Department of Biomedical Engineering, Acıbadem Mehmet Ali Aydınlar University, İstanbul, Türkiye.

ORCID-Profile: Ayşenur Eser, Sinem Burcu Erdoğan

Zitieren

APA

Eser, A., & Erdoğan, S. B. (2025). Decoding basic emotional states through integration of an fNIRS-based brain–computer interface with supervised learning algorithms. PLOS One. https://doi.org/10.1371/journal.pone.0325850

Vancouver

Eser A, Erdoğan SB. Decoding basic emotional states through integration of an fNIRS-based brain–computer interface with supervised learning algorithms. PLOS One. 2025. doi:10.1371/journal.pone.0325850

BibTeX
@article{Eser2025,
  author = {Eser, Ayşenur and Erdoğan, Sinem Burcu},
  title = {Decoding basic emotional states through integration of an fNIRS-based brain–computer interface with supervised learning algorithms},
  journal = {PLOS One},
  year = {2025},
  doi = {10.1371/journal.pone.0325850}
}
Kostenlos starten →