Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Video

Updated: Jun 27, 2026

A Multimodal Imaging- and Stimulation-based Method of Evaluating Connectivity-related Brain Excitability in Patients with Epilepsy
08:23

A Multimodal Imaging- and Stimulation-based Method of Evaluating Connectivity-related Brain Excitability in Patients with Epilepsy

Published on: November 13, 2016

Aggregating XAI-based explanations to identify spectral-spatial patterns in CNN-based resting-state EEG

Izabela Rejer1, Izabela Gago2, Vaidotas Marozas3

  • 1West Pomeranian University of Technology in Szczecin, Żołnierska 49, Szczecin, 71-210, Poland.

Scientific Reports
|June 25, 2026
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Controlled External Thigh Compression: A Feasible Method to Simulate Venous Hemodynamic Alterations Resembling Deep Vein Thrombosis.

Annals of biomedical engineering·2026
Same author

Lightweight Data-driven ECG Classification Approach with Explainable CAM Output.

Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference·2025
Same author

AveragedLIME for general explanations in EEG domain.

NeuroImage·2025
Same author

New insights in fluid monitoring for surgical patients. A concept study.

Frontiers in medical technology·2025
Same author

The impact of sensor application site on continuous noninvasive haemoglobin trends.

Computational and structural biotechnology journal·2025
Same author

Determinants of photoplethysmography signal quality at the wrist.

PLOS digital health·2025
This summary is machine-generated.

Explainable AI (XAI) methods applied to Convolutional Neural Networks (CNNs) can reveal general patterns in electroencephalographic (EEG) data. Structured aggregation of these explanations uncovers cross-subject patterns, aiding neuroscientific discovery in Alzheimer's disease detection.

Area of Science:

  • Neuroscience
  • Artificial Intelligence
  • Machine Learning

Background:

  • Convolutional Neural Networks (CNNs) excel in electroencephalographic (EEG) classification but lack interpretability.
  • Existing Explainable AI (XAI) methods offer instance-level insights, not global learned patterns.
  • A gap exists in understanding the overall structure learned by CNNs in EEG analysis.

Purpose of the Study:

  • To investigate if XAI can reveal general patterns learned by CNNs in EEG data.
  • To transform predictive CNNs into frameworks for discovering neuroscientific hypotheses.
  • To demonstrate structured aggregation of XAI explanations for cross-subject pattern identification.

Main Methods:

  • Employed averagedLIME, aggregating sample-level explanations into global class-level saliency maps.
Keywords:
Alzheimer’s diseaseAveragedLIMEConvolutional neural networks (CNN)Electroencephalography (EEG)Explainable artificial intelligence (XAI)Feature attributionModel interpretabilityResting-State EEGSpectral CNN

Related Experiment Videos

Last Updated: Jun 27, 2026

A Multimodal Imaging- and Stimulation-based Method of Evaluating Connectivity-related Brain Excitability in Patients with Epilepsy
08:23

A Multimodal Imaging- and Stimulation-based Method of Evaluating Connectivity-related Brain Excitability in Patients with Epilepsy

Published on: November 13, 2016

  • Compared four neural architectures, including spectral CNN, for subject-independent classification of Alzheimer's disease.
  • Utilized SHapley Additive exPlanations (SHAP) and Grad-CAM as complementary explanation techniques.

    • Main Results:

    • Spectral CNN achieved the highest performance (95.81% test accuracy).
    • AveragedLIME patterns were stable, robust, reproducible, and supported by SHAP explanations.
    • Saliency maps revealed known EEG slowing in Alzheimer's disease and novel spatial-spectral structures.

    Conclusions:

    • Structured aggregation of CNN explanations effectively extracts stable cross-subject patterns from resting-state EEG.
    • This approach can reveal candidate discriminative structures to motivate future neuroscientific hypotheses.
    • XAI, when aggregated, provides a powerful framework for interpreting CNNs in neuroimaging analysis.