Projects

Research projects and contracts of the PHuSe Lab.

Research · Tools · Knowledge sharing

BrainCap and BrainNet: An Affordable Integrated Solution for EEG Acquisition, Analysis, and BCI System Development

Explore the research, software and experimental insights from BrainCapNet: an affordable integrated solution for EEG acquisition, analysis and brain–computer interface development.

BrainCapNet logo
Scientific contributions

Publications

From transferable EEG representations to explainable monitoring: five contributions underpinning the project’s research.

Showing 5 publications

2026
Representation learning · Subject independence

Investigating Foundation Models, Disentanglement and Latent Alignment for Subject-Independent EEG Learning

L. Schmid, J. Burger, A. D’Amelio & R. Lanzarotti

ACM International Conference on Multimodal Interaction · ICMI 2026

A modular framework comparing foundation models, disentangled representations and alignment on motor imagery and event-related potential datasets. Latent alignment consistently improved generalisation to unseen participants.

Read publication ↗Code repository ↗
2025
Review · EEG & BCI

On Using AI for EEG-Based BCI Applications: Problems, Current Challenges and Future Trends

T. Barbera, J. Burger, A. D’Amelio, S. Zini, S. Bianco, R. Lanzarotti, P. Napoletano, G. Boccignone & J. L. Contreras-Vidal

International Journal of Human–Computer Interaction

A review of AI approaches to EEG-based interfaces, outlining challenges in signal variability, generalisation and practical deployment.

Read publication ↗
2025
Graph learning · Dementia detection

ECoGNet: An EEG-Based Effective Connectivity Graph Neural Network for Brain Disorder Detection

J. Burger, V. Cuculo, A. D’Amelio, G. Grossi & R. Lanzarotti

IJCNN 2025 · Research manuscript

A graph neural network combining estimated directed connectivity and local EEG features to distinguish Alzheimer’s disease, frontotemporal dementia and healthy controls in a public clinical dataset.

Related outputs: D2.2 · D2.4
2025*
Explainable AI · Stress assessment

EEG-Based Mental Stress Detection: A Comparative and Explainable Study Across Tasks and Subjects

F. Agnelli, G. Blandano, J. Burger, A. D’Amelio, G. Facchi, O. Ghezzi, R. Lanzarotti & L. Schmid

ICIAP 2025 Workshops · Proceedings published in 2026

An interpretable EEG classification pipeline evaluated on SAM40 and WAUC. SHAP analysis highlights the role of asymmetry and entropy features, alongside the challenge of generalisation across subjects.

Read publication ↗
2025*
Multimodal learning · Affective computing

A Study on Multimodal Foundation Models for Affective Video Prediction

F. Agnelli, M. Ditroia, G. Blandano, A. D’Amelio, O. Ghezzi, M. De Paoli & R. Lanzarotti

ICIAP 2025 · Proceedings published in 2026

Experiments on audio, video and text representations show how multimodal pretraining supports affective prediction, including when some modalities are unavailable at inference time.

Read publication ↗

* Conference edition year; the supplied Springer proceedings were published in 2026.

From research to implementation

Project outputs

Explore the hardware, computational methods and experimental integration developed across the three research work packages.

WP1

BrainCap

Acquisition tools and ergonomic adaptations designed around practical EEG recording needs.

D1.1 · EEG device review

Comparison of acquisition systems against signal quality, comfort, paediatric fit and software integration requirements.

D1.2 · BrainCap prototype

Customised supports and accessories for the Diadem and Unicorn headsets, developed through scanning, CAD modelling and 3D printing.

D1.3 · EEG & multimodal analysis

Software for synchronised acquisition, denoising, SSVEP decoding and exploratory analysis of EEG, gaze and event markers.

D1.4 · Ethics approval

Study approval and a subsequent protocol amendment permitting CE-marked non-medical EEG systems.

WP2

BrainNet

Reusable computational approaches for EEG representation learning, monitoring and BCI.

D2.1 · EEG feature embeddings

A modular framework for evaluating foundation models, disentanglement and alignment across EEG tasks.

D2.2 · Dementia & stress models

Public-dataset evaluation of connectivity-based dementia detection and explainable EEG stress assessment.

D2.3 · Subject-independent BCI

Evaluation on motor imagery and ERP benchmarks, with improvements from latent alignment.

D2.4 · Clinical EEG evaluation

Retrospective evaluation on public recordings from patients with dementia and healthy controls. No additional clinical cohort was recruited for this evaluation.

WP3

BrainTOP!

Exploratory integration of EEG and eye tracking into a game evaluated with paediatric participants.

D3.1 · Real-population BCI evaluation

Assessment of SSVEP decoding and calibration strategies using an adult control cohort as a reference.

D3.2 · Multimodal game module

Integration of EEG decoding and gaze-supported target selection, evaluated with 11 paediatric participants.

D3.3 · Monitoring module

A live EEG-derived index and reproducible offline processing. Reliable stress-state discrimination remains to be validated.

D3.4 · Clinical observations

Good overall acceptance and cooperation were observed at Fondazione TOG, alongside priorities for improving signal acquisition, individual fit and game personalisation.

These summaries describe research outputs. Exploratory findings do not establish diagnostic or rehabilitative efficacy. Public report download links can be added once dissemination versions are available.

Research dissemination

Conference contributions

Scientific outputs associated with international venues in neural networks, image analysis and multimodal interaction.

Neural networks

IJCNN 2025

ECoGNet: effective-connectivity graph learning for EEG-based brain disorder detection.

Image analysis & processing

ICIAP 2025

Contributions on explainable EEG stress detection and multimodal foundation models for affective prediction.

Multimodal interaction

ICMI 2026

Foundation models, disentanglement and latent alignment for subject-independent EEG learning.

5–9 October 2026 · Naples, Italy

Research context

Connecting computational and clinical perspectives

The project brings together EEG acquisition and computational analysis at the University of Milan with exploratory paediatric evaluation at Fondazione TOG.

Research outputs acknowledge support from ANTHEM — AdvaNced Technologies for Human-centrEd Medicine, project PNC0000003.

Scientific contact

Raffaella Lanzarotti

Department of Computer Science “Giovanni degli Antoni”
University of Milan

raffaella.lanzarotti@unimi.it ↗

For research enquiries, collaborations and information about project outputs.