def compute_model_ts(center_freq, sigma, spectrogram, freqs, target_times):
# generate stimulus time-series
rf = gaussian_1D(freqs, center_freq, sigma)
# make sure sigma isn't too big
# if np.any(np.round(rf[0],3) > 0):
# return np.inf
# if np.any(np.round(rf[-1],3) > 0):
# return np.inf
# create mask for speed
distance = freqs - center_freq
mask = np.zeros_like(distance, dtype='uint8')
mask[distance < (5 * sigma)] = 1
# extract the response
stim = generate_rf_timeseries_1D(spectrogram, rf, mask)
# recast the stimulus into a time-series that i can
source_times = np.linspace(0, target_times[-1], len(stim), endpoint=True)
f = interp1d(source_times, stim, kind='linear')
new_stim = f(target_times)
# hard-set the hrf_delay
hrf_delay = 0
# convolve it with the HRF
hrf = utils.double_gamma_hrf(hrf_delay, 1.0, 10)
stim_pad = np.tile(new_stim, 3)
model = fftconvolve(stim_pad, hrf, 'same')[len(new_stim):len(new_stim) * 2]
# normalize it
model = utils.zscore(model)
return model
def compute_model_ts(x, y, sigma, hrf_delay, theta, phi, cpd,
deg_x, deg_y, stim_arr, tr_length,
frames_per_tr, norm_func=utils.zscore):
# otherwise generate a prediction
ts_stim = MakeFastGaborPrediction(deg_x,
deg_y,
stim_arr,
x,
y,
sigma,
theta,
phi,
cpd)
# convolve it
hrf = utils.double_gamma_hrf(hrf_delay, tr_length, frames_per_tr)
# normalize it
model = norm_func(ss.fftconvolve(ts_stim, hrf)[0:len(ts_stim)])
# decimate it
model = ss.decimate(model, int(frames_per_tr), 1)
return model
def compute_model_ts(x, y, sigma, beta, hrf_delay,
deg_x, deg_y, stim_arr, tr_length):
"""
The objective function for GaussianFi class.
Parameters
----------
x : float
The model estimate along the horizontal dimensions of the display.
y : float
The model estimate along the vertical dimensions of the display.
sigma : float
The model estimate of the dispersion across the the display.
hrf_delay : float
The model estimate of the relative delay of the HRF. The canonical
HRF is assumed to be 5 s post-stimulus [1]_.
beta : float
The model estimate of the amplitude of the BOLD signal.
tr_length : float
The length of the repetition time in seconds.
Returns
-------
model : ndarray
The model prediction time-series.
References
----------
.. [1] Glover, GH. (1999). Deconvolution of impulse response in
event-related BOLD fMRI. NeuroImage 9: 416-429.
"""
# otherwise generate a prediction
stim = generate_og_timeseries(deg_x, deg_y, stim_arr, x, y, sigma)
# scale it
stim *= beta
# create the HRF
hrf = utils.double_gamma_hrf(hrf_delay, tr_length)
# convolve it with the stimulus
model = fftconvolve(stim, hrf)[0:len(stim)]
return model*beta
def test_double_gamma_hrf():
# set the TR length ... this affects the HRF sampling rate ...
tr_length = 1.0
hrf_0 = utils.double_gamma_hrf(-1, tr_length)
hrf_1 = utils.double_gamma_hrf(0, tr_length)
hrf_2 = utils.double_gamma_hrf(1, tr_length)
npt.assert_almost_equal(hrf_0.sum(), hrf_1.sum(), 3)
npt.assert_almost_equal(hrf_0.sum(), hrf_2.sum(), 3)
npt.assert_almost_equal(hrf_1.sum(), hrf_2.sum(), 3)
hrf_0 = utils.double_gamma_hrf(-1, tr_length, integrator=None)
hrf_1 = utils.double_gamma_hrf(0, tr_length, integrator=None)
hrf_2 = utils.double_gamma_hrf(1, tr_length, integrator=None)
npt.assert_array_less(hrf_0.sum(), hrf_1.sum())
npt.assert_array_less(hrf_1.sum(), hrf_2.sum())
npt.assert_array_less(hrf_0.sum(), hrf_2.sum())
def test_double_gamma_hrf():
# set the TR length ... this affects the HRF sampling rate ...
tr_length = 1.0
hrf_0 = utils.double_gamma_hrf(-1, tr_length)
hrf_1 = utils.double_gamma_hrf(0, tr_length)
hrf_2 = utils.double_gamma_hrf(1, tr_length)
npt.assert_almost_equal(hrf_0.sum(), hrf_1.sum(), 3)
npt.assert_almost_equal(hrf_0.sum(), hrf_2.sum(), 3)
npt.assert_almost_equal(hrf_1.sum(), hrf_2.sum(), 3)
hrf_0 = utils.double_gamma_hrf(-1, tr_length, integrator=None)
hrf_1 = utils.double_gamma_hrf(0, tr_length, integrator=None)
hrf_2 = utils.double_gamma_hrf(1, tr_length, integrator=None)
npt.assert_array_less(hrf_0.sum(),hrf_1.sum())
npt.assert_array_less(hrf_1.sum(),hrf_2.sum())
npt.assert_array_less(hrf_0.sum(),hrf_2.sum())
def compute_model_ts(center_freq, sigma,
spectrogram, freqs, target_times):
# generate stimulus time-series
rf = gaussian_1D(freqs, center_freq, sigma)
# make sure sigma isn't too big
# if np.any(np.round(rf[0],3) > 0):
# return np.inf
# if np.any(np.round(rf[-1],3) > 0):
# return np.inf
# create mask for speed
distance = freqs - center_freq
mask = np.zeros_like(distance, dtype='uint8')
mask[distance < (5*sigma)] = 1
# extract the response
stim = generate_rf_timeseries_1D(spectrogram,rf,mask)
# recast the stimulus into a time-series that i can
source_times = np.linspace(0,target_times[-1],len(stim),endpoint=True)
f = interp1d(source_times,stim,kind='linear')
new_stim = f(target_times)
# hard-set the hrf_delay
hrf_delay = 0
# convolve it with the HRF
hrf = utils.double_gamma_hrf(hrf_delay, 1.0, 10)
stim_pad = np.tile(new_stim,3)
model = fftconvolve(stim_pad, hrf,'same')[len(new_stim):len(new_stim)*2]
# normalize it
model = utils.zscore(model)
return model
def hemodynamic_response(self):
return utils.double_gamma_hrf(self.hrf_delay, self.tr_length)
def compute_model_ts(x, y, sigma_center, sigma_surround, beta_center, beta_surround, hrf_delay,
deg_x, deg_y, stim_arr, tr_length):
"""
The objective function for GaussianFi class.
Parameters
----------
x : float
The model estimate along the horizontal dimensions of the display.
y : float
The model estimate along the vertical dimensions of the display.
sigma_center : float
The model estimate of the dispersion of the excitatory center.
sigma_surround : float
The model estimate of the dispersion of the inhibitory surround.
beta_center : float
The amplitude of the excitatory center.
beta_surround : float
The amplitude of the inhibitory surround.
hrf_delay : float
The model estimate of the relative delay of the HRF. The canonical
HRF is assumed to be 5 s post-stimulus [1]_.
tr_length : float
The length of the repetition time in seconds.
Returns
-------
model : ndarray
The model prediction time-series.
References
----------
.. [1] Glover, GH. (1999). Deconvolution of impulse response in
event-related BOLD fMRI. NeuroImage 9: 416-429.
"""
# limiting cases for a center-surround receptive field
if sigma_center >= sigma_surround:
return np.inf
if beta_center <= beta_surround:
return np.inf
# METHOD 1
# time-series for the center and surround
stim_center, stim_surround = generate_dog_timeseries(deg_x, deg_y, stim_arr, x, y, sigma_center, sigma_surround)
# scale them by betas
stim_center *= beta_center
stim_surround *= beta_surround
# differnce them
stim_dog = stim_center - stim_surround
# # METHOD 2
# # scale the RFs, combine, then extract ts
# rf_center = generate_og_receptive_field(deg_x, deg_y, x, y, sigma_center)
# rf_center /= rf_center.max()
# rf_center *= beta_center
#
# rf_surround = generate_og_receptive_field(deg_x, deg_y, x, y, sigma_surround)
# rf_surround /= rf_surround.max()
# rf_surround *= beta_surround
#
# rf_dog = rf_center - rf_surround
# stim_dog = np.sum(np.sum(stim_arr * rf_dog[:,:,np.newaxis],axis=0),axis=0)
# generate the hrf
hrf = utils.double_gamma_hrf(hrf_delay, tr_length)
# convolve it with the stimulus timeseries
model = fftconvolve(stim_dog, hrf)[0:len(stim_dog)]
return model
