AI in Neurology

EEG Seizure Detection

import torch.nn as nn

class EEGSeizureDetector(nn.Module):
    def __init__(self, n_channels=22, n_classes=2):
        super().__init__()
        self.conv_layers = nn.Sequential(
            nn.Conv1d(n_channels, 64, 11, padding=5), nn.ReLU(), nn.MaxPool1d(4),
            nn.Conv1d(64, 128, 11, padding=5), nn.ReLU(), nn.MaxPool1d(4),
            nn.Conv1d(128, 256, 11, padding=5), nn.ReLU())
        self.pool = nn.AdaptiveAvgPool1d(1)
        self.classifier = nn.Sequential(
            nn.Linear(256, 128), nn.ReLU(), nn.Dropout(0.5),
            nn.Linear(128, n_classes))

    def forward(self, x):
        return self.classifier(self.pool(self.conv_layers(x)).squeeze(-1))

Frequency Bands

Brain Tumor Segmentation

Multi-Modal MRI

class BrainTumorSegmenter(nn.Module):
    def __init__(self, in_channels=4, n_classes=4):
        super().__init__()
        self.encoder = nn.Sequential(
            nn.Conv3d(in_channels, 32, 3, padding=1), nn.BatchNorm3d(32), nn.ReLU(),
            nn.MaxPool3d(2), nn.Conv3d(32, 64, 3, padding=1),
            nn.BatchNorm3d(64), nn.ReLU())
        self.decoder = nn.Conv3d(64, n_classes, 1)

    def forward(self, x):
        return self.decoder(self.encoder(x))

Alzheimer's Progression

Evaluation

• Dice score for tumor segmentation
• Sensitivity > 95% for seizure detection
• AUC for disease progression prediction