BrainTwin-AI: A Multimodal MRI-EEG-Based Cognitive Digital Twin for Real-Time Brain Health Intelligence

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Imaging

Background:

• Brain health monitoring is crucial due to increasing cognitive load and stress.
• Neural instability necessitates advanced diagnostic tools.
• Current methods lack integrated real-time functional and structural brain assessment.

Purpose of the Study:

• To introduce BrainTwin, a cognitive digital twin for patient-specific brain health monitoring.
• To combine MRI analytics for neuro-oncological assessment with real-time EEG-based intelligence.
• To enhance brain tumor detection, progression monitoring, and functional brain state analysis.

Main Methods:

• Utilized an Enhanced Vision Transformer (ViT++) for accurate MRI-based tumor segmentation and progression modeling.
• Employed a wearable EEG skullcap with Edge AI for low-latency, secure data preprocessing and transmission.
• Integrated EEG-MRI fusion with a BiLSTM classifier for cognitive state analysis (relaxed, stress, fatigue) and risk score generation.
• Implemented Explainable AI (XAI) using Grad-CAM for clinical interpretability of structural analysis.

Main Results:

• Achieved 96% accuracy in ViT++ MRI tumor detection, outperforming baseline models.
• Demonstrated a BiLSTM macro-F1 score of 0.94 for classifying cognitive states (Relaxed 0.96, Stress 0.93, Fatigue 0.92).
• Successfully correlated functional EEG metrics with structural MRI data to generate a unified risk score.

Conclusions:

• BrainTwin offers a reliable, interpretable, and clinically useful integrated solution for brain tumor assessment.
• The digital twin provides real-time functional brain monitoring capabilities.
• BrainTwin represents a significant advancement in personalized brain health management and neuro-oncological care.