BCINexusBCILattice
Flagship Desktop Application

BCILattice
The Professional
BCI Research App

The flagship product of the BCINexus platform. BCILattice brings professional EEG, fNIRS & EMG analysis, visual AI pipeline design, and local GPU training to your desktop, fully private, fully yours.

Data never leaves your machine
Full local GPU training
Free forever plan
Product Overview

Everything BCI research needs,
in one desktop app

BCILattice is the offline-first, research-grade desktop application that connects to the BCINexus cloud ecosystem. Run the complete BCI research workflow entirely on your machine, then publish to the community with one click.

Research-Grade

Built for neuroscientists and ML engineers. Publication-quality pipelines, full signal processing, and peer-reviewed model architectures out of the box.

AI-Native

Every pipeline is a deep learning pipeline. Visual drag-and-drop AI model design with EEGNet, Transformers, LSTM, and custom architectures running on your local GPU.

Privacy-First

Your EEG data never touches a cloud server unless you choose to publish. 100% local compute, zero telemetry, GDPR-compliant by design.

Platform Capabilities

Built for the whole team.
Every workflow. Every scale.

01

Multi-Modal Signal Support

EEG up to 256 chHD-DOT fNIRS · HbO/HbREMG · EOGLive LSL
02

Custom ML / DL Pipelines

Visual canvasCustom layersPyTorchCUDA GPU
03

Custom BCI Paradigm Design

Block sequencerMI · P300 · SSVEPOne-click Compile
04

Multi-Format Export

ONNX · TFLite · PyTorchMATLAB (.mat)CSV · HDF5 · BIDS
05

Community Pipeline Upload

One-click publishModeration340+ pipelines
06

Team System

Private workspaceRolesEmail invites
07

Shared Project Workspaces

Real-time syncMulti-machineActivity feed
08

Local & Cloud Sharing

.NeuroMatrix packageSHA-256Air-gapped
09

Private Team Data Sharing

Encrypted transitTeam-onlyGDPR compliant
Research Workflow

From raw signal
to published pipeline

BCILattice covers the complete BCI research lifecycle in six structured stages, from data import to community publication. No stitching together separate tools.

Import any BCI data format or stream live via LSL
Automated preprocessing with ICA and artefact rejection
Visual AI model design with Neural Flow canvas
Full local GPU training with real-time monitoring
Export to ONNX, TFLite, PyTorch, or MATLAB
Publish pipelines to the BCINexus research community
BCILattice · Research Pipeline
01
Import BCI Data
EDF · BDF · SNIRF · CSV · live LSL stream
Connected
02
Preprocess Signals
ICA · bandpass · epoch · artefact rejection
Running
03
Build Neural Flow
Drag-and-drop AI architecture canvas
Building
04
Train Locally
PyTorch GPU · 91.4% val accuracy · private
Epoch 47
05
Export & Version
ONNX · TFLite · PyTorch · MATLAB export
Exported
06
Publish to BCINexus
Share pipelines · community collaboration
Live
Technical Specifications

Built for serious research.
No compromises.

Signal Modalities

  • EEG (up to 256 ch)
  • fNIRS (HD-DOT)
  • EMG / EOG
  • Live LSL stream

File Formats

  • EDF / BDF / GDF
  • SNIRF / NIRS
  • CSV · HDF5 · MAT
  • BIDS-compatible

AI Architectures

  • EEGNet (CNN)
  • ShallowConvNet
  • LSTM / BiLSTM
  • Transformer · Custom

Export Targets

  • ONNX (universal)
  • TFLite (mobile)
  • PyTorch (.pt)
  • MATLAB (.mat)

Layered architecture,
modular by design

BCILattice is structured in four clean layers, Application, Processing Engine, AI Core, and Data & Storage. Each layer is independently maintainable and extensible, making BCILattice a stable platform for years of research.

Cross-platformWindows · macOS · Linux
BackendPython 3.10+ · PyQt6
ML FrameworkPyTorch (CUDA / CPU)
Cloud syncBCINexus API · optional
Application Layer
Neural Flow CanvasData ManagerML SuiteAI ChatExperiment Hub
Processing Engine
Signal PreprocessorPipeline ExecutorModel TrainerInference Runtime
AI & ML Core
PyTorch BackendEEGNet · LSTM · TransformerONNX ExporterTransfer Learning
Data & Storage
EDF/BDF/SNIRF ParserLSL Stream HandlerLocal DatabaseBCINexus Sync
Built-in AI Models

State-of-the-art BCI models,
ready to train

8 peer-reviewed architectures pre-configured for BCI classification. No setup, load your data and hit Train.

CNN

EEGNet

Compact depthwise CNN for raw EEG. Temporal + spatial convolutions, robust across paradigms.

ConvNet

ShallowConvNet

Frequency-band power features. Ideal for motor imagery and SSVEP with minimal parameters.

RNN

LSTM / BiLSTM

Bidirectional sequence modelling for temporal dependencies across EEG and fNIRS time series.

Attention

Transformer

Multi-head self-attention. State-of-the-art on BCI Competition IV and MOABB benchmark suites.

Deep CNN

DeepConvNet

Four-block deep CNN with batch norm and dropout. Top accuracy on multi-subject, high-density EEG.

Custom

Custom Neural Flow

Design your own architecture on the canvas. Connect any combination of layers, losses, and optimisers visually.

Fine-tune

Transfer Learning

Pre-trained BCI weights from large subject-general datasets. Fine-tune on your data in minutes.

Live

Real-Time Inference

Deploy any trained model to live LSL stream. Closed-loop BCI output at sub-100 ms latency.

BCINexus Ecosystem

Local power meets
global community

BCILattice is the desktop application. BCINexus is the ecosystem that connects your work to thousands of researchers worldwide. Every local pipeline can become a community resource.

Community Hub

Browse 340+ published BCI pipelines from global research groups.

Cloud Sync

Sync datasets and experiments across machines via BCINexus account.

Pipeline Sharing

Publish versioned pipelines with full reproducibility metadata.

Research Collaboration

Team workspaces, shared experiments, and peer pipeline review.

Contributors

9+

Paradigm Types

Published Pipelines

8+

AI Architectures

3

Signal Modalities

100%

Local & Private

Interface Demo

See the Neural Flow canvas
in action

Motor imagery 4-class classification pipeline, from raw 64-channel EEG to 91.4% validation accuracy.

7

Pipeline nodes

91.4%

Validation accuracy

100

Training epochs

Coming Soon

What we're building next

These features are in active development. Available on all paid plans, not included in Free.

In development

Live Recording

Stream EEG, EMG, and fNIRS directly into BCILattice. Run your full preprocessing pipeline on live data and classify signals in real time, all locally.

LSL streamingLive preprocessingOnline classificationLocal & private
Progress55%
Learn more
Early stage

Edge Devices

Deploy trained BCI models to Raspberry Pi and Jetson Nano. Stream live biosignals from edge hardware and run on-device inference for real-world applications.

Raspberry PiJetson NanoONNX / TFLiteOn-device inferenceAir-gapped
Progress30%
Learn more
Start for free · No credit card

Download BCILattice
and start your first pipeline

Windows · macOS · Linux · Free forever plan · Local GPU · Your data stays yours.