Upload roi_connectivity.py

roi_connectivity.py

@@ -0,0 +1,177 @@
+# roi_connectivity.py
+
+
+import numpy as np
+import mne
+import json
+import pandas as pd
+from mne_connectivity import spectral_connectivity_epochs
+from nilearn import datasets
+
+# ───────────────────────────────────────
+# 1. Utility: ROI definitions
+# ───────────────────────────────────────
+
+def get_difumo_names():
+    try:
+        atlas = datasets.fetch_atlas_difumo(dimension=512, resolution_mm=2)
+        return atlas.labels['difumo_names'].astype(str).tolist()
+    except Exception:
+        return [f"Component_{i}" for i in range(512)]
+
+def define_motor_cognitive_regions():
+    Motor_M1 = [40, 86, 198, 268, 305, 437, 458, 465]
+    Motor_SMA_Premotor = [17, 18, 288, 291, 296, 297, 302, 305, 314, 315, 335, 375, 379, 448]
+    Motor_Medial = [101, 102, 388, 409, 498]
+    Thalamus = [70, 73, 297, 334, 414, 420] 
+    Basal_Ganglia = [30, 53, 224, 260, 405, 422, 109, 110, 315, 331, 467, 479, 55, 71, 307, 223]  
+    Cerebellum_Motor = [43, 47, 83, 84, 127, 183, 220, 221, 295, 304, 310, 311, 374, 378, 381, 403, 441, 490, 491]
+    Somatosensory = [44, 131, 210, 411, 413, 436]
+    Executive_Control = [3, 85, 104, 148, 184, 337, 377, 446, 447, 506, 507]
+    Interoception = [2, 387, 358, 389, 165, 469]
+    Error_Monitoring = [185, 219, 326, 473, 492]
+    return sorted(set(
+        Motor_M1 + Motor_SMA_Premotor + Motor_Medial + Thalamus +
+        Basal_Ganglia + Cerebellum_Motor + Somatosensory +
+        Executive_Control + Interoception + Error_Monitoring
+    ))
+
+def get_band_freqs(band_name):
+    bands = {
+        "Theta": (4, 8),
+        "Alpha": (8, 12),
+        "Low_Beta": (13, 20),
+        "High_Beta": (20, 30),
+        "Low_Gamma": (30, 60),
+        "High_Gamma": (60, 120)
+    }
+    if band_name not in bands:
+        raise ValueError(f"Unknown band: {band_name}. Options: {list(bands.keys())}")
+    return bands[band_name]
+
+# ───────────────────────────────────────
+# 2. Epoch creation functions
+# ───────────────────────────────────────
+
+def create_task_epochs(
+    data_file,
+    events_file,
+    event_id_file,
+    condition,
+    tmin=0.0,
+    tmax=1.5,
+    sfreq=500.0
+):
+    """Create epochs from event markers."""
+    data = np.load(data_file)
+    if data.shape[0] > data.shape[1]:
+        data = data.T
+
+    events = mne.read_events(events_file)
+    with open(event_id_file, 'r') as f:
+        event_id = json.load(f)
+
+    ch_names = [f'C{i}' for i in range(data.shape[0])]
+    info = mne.create_info(ch_names, sfreq=sfreq, ch_types='misc')
+    raw = mne.io.RawArray(data, info, verbose=False)
+
+    epochs = mne.Epochs(
+        raw, events, {condition: event_id[condition]},
+        tmin=tmin, tmax=tmax, baseline=None,
+        preload=True, verbose=False, event_repeated='drop'
+    )
+    return epochs
+
+def create_rest_epochs(
+    data_file,
+    duration=2.5,
+    sfreq=500.0
+):
+    """Create fixed-length epochs from continuous data."""
+    data = np.load(data_file)
+    if data.shape[0] > data.shape[1]:
+        data = data.T
+
+    ch_names = [f'C{i}' for i in range(data.shape[0])]
+    info = mne.create_info(ch_names, sfreq=sfreq, ch_types='misc')
+    raw = mne.io.RawArray(data, info, verbose=False)
+
+    events = mne.make_fixed_length_events(raw, duration=duration)
+    epochs = mne.Epochs(
+        raw, events, tmin=0, tmax=duration,
+        baseline=None, preload=True, verbose=False
+    )
+    return epochs
+
+# ───────────────────────────────────────
+# 3. Connectivity function
+# ───────────────────────────────────────
+
+def compute_roi_connectivity_matrix(
+    epochs,
+    band_name="Low_Beta",
+    method='wpli2_debiased',
+    sfreq=500.0
+):
+    """
+    Compute a single ROI × ROI connectivity matrix from MNE Epochs object.
+    
+    Parameters:
+    - epochs: mne.Epochs instance (already loaded and preprocessed)
+    - band_name: e.g., "Alpha", "Low_Beta"
+    - method: connectivity method (default: 'wpli2_debiased')
+    
+    Returns:
+    - conn_df: pandas DataFrame (n_roi × n_roi) with DiFuMo ROI names as labels
+    """
+    # Get ROI info
+    all_names = get_difumo_names()
+    selected_indices = define_motor_cognitive_regions()
+    roi_names = [all_names[i] for i in selected_indices]
+
+    # Extract data for selected ROIs
+    epoch_data = epochs.get_data()[:, selected_indices, :]  # (n_epochs, n_roi, n_times)
+
+    # Get frequency range
+    fmin, fmax = get_band_freqs(band_name)
+
+    # Compute connectivity
+    con = spectral_connectivity_epochs(
+        data=epoch_data,
+        method=method,
+        mode='multitaper',
+        sfreq=sfreq,
+        fmin=fmin,
+        fmax=fmax,
+        faverage=True,
+        verbose=False
+    )
+    matrix = con.get_data(output='dense').squeeze()
+    matrix = (matrix + matrix.T) / 2
+    np.fill_diagonal(matrix, 0)
+
+    return pd.DataFrame(matrix, index=roi_names, columns=roi_names)
+
+
+'''
+# Task
+epochs = create_task_epochs(
+    data_file=r"sub-02_task\difumo_time_courses.npy",
+    events_file=r"sub-02_task\sub-02_events_mne_binary-eve.fif",
+    event_id_file=r"sub-02_task\sub-02_event_id_binary.json",
+    condition="InPhase"
+)
+
+conn_matrix = compute_roi_connectivity_matrix(epochs, band_name="Alpha")
+conn_matrix.to_csv("sub-02_InPhase_Alpha_matrix.csv")
+
+
+# Rest
+epochs = create_rest_epochs(
+    data_file=r"sub-02_rest\difumo_time_courses.npy",
+    duration=2.5
+)
+
+conn_matrix = compute_roi_connectivity_matrix(epochs, band_name="Low_Beta")
+
+'''