def individual_analysis(bids_path, ID): raw_intensity = read_raw_bids(bids_path=bids_path, verbose=False) # Convert signal to haemoglobin and resample raw_od = optical_density(raw_intensity) raw_haemo = beer_lambert_law(raw_od) raw_haemo.resample(0.3) # Cut out just the short channels for creating a GLM repressor sht_chans = get_short_channels(raw_haemo) raw_haemo = get_long_channels(raw_haemo) # Create a design matrix design_matrix = make_first_level_design_matrix(raw_haemo, stim_dur=5.0) # Append short channels mean to design matrix design_matrix["ShortHbO"] = np.mean(sht_chans.copy().pick(picks="hbo").get_data(), axis=0) design_matrix["ShortHbR"] = np.mean(sht_chans.copy().pick(picks="hbr").get_data(), axis=0) # Run GLM glm_est = run_GLM(raw_haemo, design_matrix) # Define channels in each region of interest # List the channel pairs manually left = [[4, 3], [1, 3], [3, 3], [1, 2], [2, 3], [1, 1]] right = [[6, 7], [5, 7], [7, 7], [5, 6], [6, 7], [5, 5]] # Then generate the correct indices for each pair groups = dict( Left_Hemisphere=picks_pair_to_idx(raw_haemo, left, on_missing='ignore'), Right_Hemisphere=picks_pair_to_idx(raw_haemo, right, on_missing='ignore')) # Extract channel metrics cha = glm_to_tidy(raw_haemo, glm_est, design_matrix) cha["ID"] = ID # Add the participant ID to the dataframe # Compute region of interest results from channel data roi = pd.DataFrame() for idx, col in enumerate(design_matrix.columns): roi = roi.append(glm_region_of_interest(glm_est, groups, idx, col)) roi["ID"] = ID # Add the participant ID to the dataframe # Contrast left vs right tapping contrast_matrix = np.eye(design_matrix.shape[1]) basic_conts = dict([(column, contrast_matrix[i]) for i, column in enumerate(design_matrix.columns)]) contrast_LvR = basic_conts['Tapping/Left'] - basic_conts['Tapping/Right'] contrast = compute_contrast(glm_est, contrast_LvR) con = glm_to_tidy(raw_haemo, contrast, design_matrix) con["ID"] = ID # Add the participant ID to the dataframe # Convert to uM for nicer plotting below. cha["theta"] = [t * 1.e6 for t in cha["theta"]] roi["theta"] = [t * 1.e6 for t in roi["theta"]] con["effect"] = [t * 1.e6 for t in con["effect"]] return raw_haemo, roi, cha, con
def test_run_plot_GLM_topo(): raw_intensity = _load_dataset() raw_intensity.crop(450, 600) # Keep the test fast design_matrix = make_first_level_design_matrix(raw_intensity, drift_order=1, drift_model='polynomial') raw_od = mne.preprocessing.nirs.optical_density(raw_intensity) raw_haemo = mne.preprocessing.nirs.beer_lambert_law(raw_od) glm_estimates = run_GLM(raw_haemo, design_matrix) fig = plot_glm_topo(raw_haemo, glm_estimates, design_matrix) # 5 conditions (A,B,C,Drift,Constant) * two chroma + 2xcolorbar assert len(fig.axes) == 12 # Two conditions * two chroma + 2 x colorbar fig = plot_glm_topo(raw_haemo, glm_estimates, design_matrix, requested_conditions=['A', 'B']) assert len(fig.axes) == 6 # Two conditions * one chroma + 1 x colorbar fig = plot_glm_topo(raw_haemo.copy().pick(picks="hbo"), glm_estimates, design_matrix, requested_conditions=['A', 'B']) assert len(fig.axes) == 3 # One conditions * two chroma + 2 x colorbar fig = plot_glm_topo(raw_haemo, glm_estimates, design_matrix, requested_conditions=['A']) assert len(fig.axes) == 4 # One conditions * one chroma + 1 x colorbar fig = plot_glm_topo(raw_haemo.copy().pick(picks="hbo"), glm_estimates, design_matrix, requested_conditions=['A']) assert len(fig.axes) == 2 # One conditions * one chroma + 0 x colorbar fig = plot_glm_topo(raw_haemo.copy().pick(picks="hbo"), glm_estimates, design_matrix, colorbar=False, requested_conditions=['A']) assert len(fig.axes) == 1 # Ensure warning thrown if glm estimates is missing channels from raw glm_estimates_subset = { a: glm_estimates[a] for a in raw_haemo.ch_names[0:3] } with pytest.raises(RuntimeError, match="does not match regression"): plot_glm_topo(raw_haemo, glm_estimates_subset, design_matrix)
def test_GLM_system_test(): fnirs_data_folder = mne.datasets.fnirs_motor.data_path() fnirs_raw_dir = os.path.join(fnirs_data_folder, 'Participant-1') raw_intensity = mne.io.read_raw_nirx(fnirs_raw_dir).load_data() raw_intensity.resample(1.0) new_des = [des for des in raw_intensity.annotations.description] new_des = ['Control' if x == "1.0" else x for x in new_des] new_des = ['Tapping/Left' if x == "2.0" else x for x in new_des] new_des = ['Tapping/Right' if x == "3.0" else x for x in new_des] annot = mne.Annotations(raw_intensity.annotations.onset, raw_intensity.annotations.duration, new_des) raw_intensity.set_annotations(annot) raw_intensity.annotations.crop(35, 2967) raw_od = mne.preprocessing.nirs.optical_density(raw_intensity) raw_haemo = mne.preprocessing.nirs.beer_lambert_law(raw_od) short_chs = get_short_channels(raw_haemo) raw_haemo = get_long_channels(raw_haemo) design_matrix = make_first_level_design_matrix(raw_intensity, hrf_model='spm', stim_dur=5.0, drift_order=3, drift_model='polynomial') design_matrix["ShortHbO"] = np.mean( short_chs.copy().pick(picks="hbo").get_data(), axis=0) design_matrix["ShortHbR"] = np.mean( short_chs.copy().pick(picks="hbr").get_data(), axis=0) glm_est = run_GLM(raw_haemo, design_matrix) df = glm_to_tidy(raw_haemo, glm_est, design_matrix) df = _tidy_long_to_wide(df) a = (df.query('condition in ["Control"]').groupby(['condition', 'Chroma' ]).agg(['mean'])) # Make sure false positive rate is less than 5% assert a["Significant"].values[0] < 0.05 assert a["Significant"].values[1] < 0.05 a = (df.query('condition in ["Tapping/Left", "Tapping/Right"]').groupby( ['condition', 'Chroma']).agg(['mean'])) # Fairly arbitrary cutoff here, but its more than 5% assert a["Significant"].values[0] > 0.7 assert a["Significant"].values[1] > 0.7 assert a["Significant"].values[2] > 0.7 assert a["Significant"].values[3] > 0.7 left = [[1, 1], [1, 2], [1, 3], [2, 1], [2, 3], [2, 4], [3, 2], [3, 3], [4, 3], [4, 4]] right = [[5, 5], [5, 6], [5, 7], [6, 5], [6, 7], [6, 8], [7, 6], [7, 7], [8, 7], [8, 8]] groups = dict(Left_ROI=picks_pair_to_idx(raw_haemo, left), Right_ROI=picks_pair_to_idx(raw_haemo, right)) df = pd.DataFrame() for idx, col in enumerate(design_matrix.columns[:3]): df = df.append(glm_region_of_interest(glm_est, groups, idx, col)) assert df.shape == (12, 8)
def test_io(): num_chans = 6 fnirs_data_folder = mne.datasets.fnirs_motor.data_path() fnirs_raw_dir = os.path.join(fnirs_data_folder, 'Participant-1') raw_intensity = mne.io.read_raw_nirx(fnirs_raw_dir).load_data() raw_intensity.resample(0.2) raw_od = mne.preprocessing.nirs.optical_density(raw_intensity) raw_haemo = mne.preprocessing.nirs.beer_lambert_law(raw_od) raw_haemo = mne_nirs.channels.get_long_channels(raw_haemo) raw_haemo.pick(picks=range(num_chans)) design_matrix = make_first_level_design_matrix(raw_intensity, hrf_model='spm', stim_dur=5.0, drift_order=3, drift_model='polynomial') glm_est = run_GLM(raw_haemo, design_matrix) df = glm_to_tidy(raw_haemo, glm_est, design_matrix) df = _tidy_long_to_wide(df) assert df.shape == (48, 11) assert set(df.columns) == { 'ch_name', 'condition', 'df', 'mse', 'p_value', 't', 'theta', 'Source', 'Detector', 'Chroma', 'Significant' } num_conds = 8 # triggers (1, 2, 3, 15) + 3 drifts + constant assert df.shape[0] == num_chans * num_conds contrast_matrix = np.eye(design_matrix.shape[1]) basic_conts = dict([(column, contrast_matrix[i]) for i, column in enumerate(design_matrix.columns)]) contrast_LvR = basic_conts['2.0'] - basic_conts['3.0'] contrast = mne_nirs.statistics.compute_contrast(glm_est, contrast_LvR) df = glm_to_tidy(raw_haemo, contrast, design_matrix) df = _tidy_long_to_wide(df) assert df.shape == (6, 10) assert set(df.columns) == { 'ch_name', 'ContrastType', 'z_score', 'stat', 'p_value', 'effect', 'Source', 'Detector', 'Chroma', 'Significant' } contrast = mne_nirs.statistics.compute_contrast(glm_est, contrast_LvR, contrast_type='F') df = glm_to_tidy(raw_haemo, contrast, design_matrix) df = _tidy_long_to_wide(df) assert df.shape == (6, 10) assert set(df.columns) == { 'ch_name', 'ContrastType', 'z_score', 'stat', 'p_value', 'effect', 'Source', 'Detector', 'Chroma', 'Significant' }
def test_run_GLM(): raw = simulate_nirs_raw(sig_dur=200, stim_dur=5.) design_matrix = make_first_level_design_matrix(raw, stim_dur=5., drift_order=1, drift_model='polynomial') glm_estimates = run_GLM(raw, design_matrix) assert len(glm_estimates) == len(raw.ch_names) # Check the estimate is correct within 10% error assert abs(glm_estimates["Simulated"].theta[0] - 1.e-6) < 0.1e-6 # ensure we return the same type as nilearn to encourage compatibility _, ni_est = nilearn.glm.first_level.run_glm( raw.get_data(0).T, design_matrix.values) assert type(ni_est) == type(glm_estimates)
def test_run_plot_GLM_contrast_topo(): raw_intensity = _load_dataset() raw_intensity.crop(450, 600) # Keep the test fast design_matrix = make_first_level_design_matrix(raw_intensity, drift_order=1, drift_model='polynomial') raw_od = mne.preprocessing.nirs.optical_density(raw_intensity) raw_haemo = mne.preprocessing.nirs.beer_lambert_law(raw_od) glm_est = run_GLM(raw_haemo, design_matrix) contrast_matrix = np.eye(design_matrix.shape[1]) basic_conts = dict([(column, contrast_matrix[i]) for i, column in enumerate(design_matrix.columns)]) contrast_LvR = basic_conts['A'] - basic_conts['B'] contrast = mne_nirs.statistics.compute_contrast(glm_est, contrast_LvR) fig = mne_nirs.visualisation.plot_glm_contrast_topo(raw_haemo, contrast) assert len(fig.axes) == 3
def test_simulate_NIRS(): raw = simulate_nirs_raw(sfreq=3., amplitude=1., sig_dur=300., stim_dur=5., isi_min=15., isi_max=45.) assert 'hbo' in raw assert raw.info['sfreq'] == 3. assert raw.get_data().shape == (1, 900) assert np.max(raw.get_data()) < 1.2 * 1.e-6 assert raw.annotations.description[0] == 'A' assert raw.annotations.duration[0] == 5 assert np.min(np.diff(raw.annotations.onset)) > 15. + 5. assert np.max(np.diff(raw.annotations.onset)) < 45. + 5. with pytest.raises(AssertionError, match='Same number of'): raw = simulate_nirs_raw(sfreq=3., amplitude=[1., 2.], sig_dur=300., stim_dur=5., isi_min=15., isi_max=45.) raw = simulate_nirs_raw(sfreq=3., amplitude=[0., 2., 4.], annot_desc=['Control', 'Cond_A', 'Cond_B'], stim_dur=[5, 5, 5], sig_dur=900., isi_min=15., isi_max=45.) design_matrix = make_first_level_design_matrix(raw, stim_dur=5.0, drift_order=1, drift_model='polynomial') glm_est = run_GLM(raw, design_matrix) df = glm_to_tidy(raw, glm_est, design_matrix) df = _tidy_long_to_wide(df) assert df.query("condition in ['Control']")['theta'].values[0] == \ pytest.approx(0) assert df.query("condition in ['Cond_A']")['theta'].values[0] == \ pytest.approx(2e-6) assert df.query("condition in ['Cond_B']")['theta'].values[0] == \ pytest.approx(4e-6)
def test_run_plot_GLM_topo(): raw_intensity = _load_dataset() raw_intensity.crop(450, 600) # Keep the test fast design_matrix = make_first_level_design_matrix(raw_intensity, drift_order=1, drift_model='polynomial') raw_od = mne.preprocessing.nirs.optical_density(raw_intensity) raw_haemo = mne.preprocessing.nirs.beer_lambert_law(raw_od) glm_estimates = run_GLM(raw_haemo, design_matrix) fig = plot_glm_topo(raw_haemo, glm_estimates, design_matrix) # 5 conditions (A,B,C,Drift,Constant) * two chroma + 2xcolorbar assert len(fig.axes) == 12 fig = plot_glm_topo(raw_haemo, glm_estimates, design_matrix, requested_conditions=['A', 'B']) # Two conditions * two chroma + 2xcolorbar assert len(fig.axes) == 6
def test_run_GLM(): raw = simulate_nirs_raw(sig_dur=200, stim_dur=5.) design_matrix = make_first_level_design_matrix(raw, stim_dur=5., drift_order=1, drift_model='polynomial') glm_estimates = run_glm(raw, design_matrix) # Test backwards compatibility with pytest.deprecated_call(match='more comprehensive'): old_res = run_GLM(raw, design_matrix) assert old_res.keys() == glm_estimates.data.keys() assert (old_res["Simulated"].theta == glm_estimates.data["Simulated"].theta ).all() assert len(glm_estimates) == len(raw.ch_names) # Check the estimate is correct within 10% error assert abs(glm_estimates.pick("Simulated").theta()[0][0] - 1.e-6) < 0.1e-6 # ensure we return the same type as nilearn to encourage compatibility _, ni_est = nilearn.glm.first_level.run_glm( raw.get_data(0).T, design_matrix.values) assert isinstance(glm_estimates._data, type(ni_est))
drift_model='polynomial') fig, ax1 = plt.subplots(figsize=(10, 6), nrows=1, ncols=1) fig = plot_design_matrix(design_matrix, ax=ax1) ############################################################################### # Estimate response on clean data # ------------------------------- # # Here we run the GLM analysis on the clean data. # The design matrix had three columns, so we get an estimate for our simulated # event, the first order drift, and the constant. # We see that the estimate of the first component is 4e-6 (4 uM), # which was the amplitude we used in the simulation. # We also see that the mean square error of the model fit is close to zero. glm_est = run_GLM(raw, design_matrix) print("Estimate:", glm_est['Simulated'].theta[0], " MSE:", glm_est['Simulated'].MSE, " Error (uM):", 1e6 * (glm_est['Simulated'].theta[0] - amp * 1e-6)) ############################################################################### # Simulate noisy NIRS data (white) # -------------------------------- # # Real data has noise. Here we add white noise, this noise is not realistic # but serves as a reference point for evaluating the estimation process. # We run the GLM analysis exactly as in the previous section # and plot the noisy data and the GLM fitted model. # We print the response estimate and see that is close, but not exactly correct, # we observe the mean square error is similar to the added noise.
# # Fit GLM to subset of data and estimate response for each experimental condition # ------------------------------------------------------------------------------- # # .. sidebar:: Relevant literature # # Huppert TJ. Commentary on the statistical properties of noise and its # implication on general linear models in functional near-infrared # spectroscopy. Neurophotonics. 2016;3(1) # # We run a GLM fit for the data and experiment matrix. # First we analyse just the first two channels which correspond HbO and HbR # of a single source detector pair. data_subset = raw_haemo.copy().pick(picks=range(2)) glm_est = run_GLM(data_subset, design_matrix) ############################################################################### # # We then display the results. Note that the control condition sits # around zero. # And that the HbO is positive and larger than the HbR, this is to be expected. # Further, we note that for this channel the response to tapping on the # right hand is larger than the left. And the values are similar to what # is seen in the epoching tutorial. plt.scatter(design_matrix.columns[:3], glm_est['S1_D1 hbo'].theta[:3] * 1e6) plt.scatter(design_matrix.columns[:3], glm_est['S1_D1 hbr'].theta[:3] * 1e6) plt.xlabel("Experiment Condition") plt.ylabel("Haemoglobin (μM)") plt.legend(["Oxyhaemoglobin", "Deoxyhaemoglobin"])