Temporal Brain Category Graph Convolutional Network for schizophrenia diagnosis using dynamic functional connectivity from rs-fMRI data.
Temporal-BCGCN (Temporal Brain Category Graph Convolutional Network) 是一种用于精神分裂症诊断和偏侧化分析的动态脑网络分析模型。
核心特点：
性能：
动态脑网络分析模块，用于构建动态同步特征。
功能：
基于同步时间特性的图卷积方法。
优势：
首个基于深度学习的半球偏侧化测试工具。
功能：
# 模型架构示例
class TemporalBCGCN:
def __init__(self):
self.dsf_brainnet = DSFBrainNet() # 动态同步特征
self.temporal_conv = TemporalConv() # 时序图卷积
self.category_pool = CategoryPool() # 偏侧化分析
def forward(self, rs_fmri_data):
# 1. 构建动态脑网络
dynamic_features = self.dsf_brainnet(rs_fmri_data)
# 2. 时序图卷积
graph_features = self.temporal_conv(dynamic_features)
# 3. 偏侧化分析 + 分类
diagnosis, lateralization = self.category_pool(graph_features)
return diagnosis, lateralization
# 动态功能连接提取
def extract_dynamic_fc(rs_fmri, window_size=30, step=10):
"""
从 rs-fMRI 提取动态功能连接
Args:
rs_fmri: 静息态 fMRI 数据 (timepoints, regions)
window_size: 滑动窗口大小
step: 滑动步长
Returns:
动态连接矩阵序列
"""
timepoints, regions = rs_fmri.shape
n_windows = (timepoints - window_size) // step + 1
dynamic_fc = []
for i in range(n_windows):
start = i * step
end = start + window_size
window_data = rs_fmri[start:end]
fc = np.corrcoef(window_data.T)
dynamic_fc.append(fc)
return np.array(dynamic_fc)
左半球更严重受损
关键脑区
| 方法 | 准确率 (COBRE) | 准确率 (UCLA) |
|---|---|---|
| Temporal-BCGCN | 83.62% | 89.71% |
| 传统 GCN | ~75% | ~80% |
| 静态 FC 方法 | ~70% | ~75% |
# rs-fMRI 预处理流程
preprocessing_steps = [
"slice_timing", # 时间层校正
"motion_correction", # 运动校正
"normalization", # 标准化
"smoothing", # 平滑
"bandpass_filter" # 带通滤波 (0.01-0.1 Hz)
]
# 使用标准脑图谱
atlases = [
"AAL", # 自动解剖标记
"Power", # Power 网络
"Schaefer" # Schaefer 功能网络
]
# 训练配置
config = {
"learning_rate": 0.001,
"batch_size": 32,
"epochs": 200,
"dropout": 0.5,
"weight_decay": 1e-4
}
官方代码： https://github.com/[论文作者仓库]
eeg-brain-connectivity-bci - EEG 脑连接分析time-varying-brain-connectivity - 时变脑连接drl-gnn-brain-network - 图神经网络脑网络multimodal-brain-connectivity-gnn - 多模态脑连接@article{zhu2023temporal,
title={Temporal Dynamic Synchronous Functional Brain Network for Schizophrenia Diagnosis and Lateralization Analysis},
author={Zhu, Cheng and others},
journal={arXiv preprint arXiv:2304.01347},
year={2023}
}
read - Read documentation and referencesweb_search - Search for related informationweb_fetch - Fetch paper or documentationFollow these steps when applying this skill:
User: I need to apply Temporal-BCGCN Brain Network Analysis to my analysis.
Agent: I'll help you apply temporal-bcggcn-brain-network. First, let me understand your specific use case...
Context: Apply the methodology
User: Complex analysis scenario
Agent: Based on the methodology, I'll guide you through the advanced application...
User: What are the key considerations for temporal-bcggcn-brain-network?
Agent: Let me search for the latest research and best practices...