def extract_motor_components_OF(m, n_components, mask = None, resize_fact= .5, only_magnitude = False, max_iter = 1000, verbose = False):
if mask is not None:
mask = coo_matrix(np.array(mask).squeeze())
ms = [get_nonzero_subarray(mask.multiply(fr),mask) for fr in m]
ms = np.dstack(ms)
ms = cm.movie(ms.transpose([2,0,1]))
else:
ms = m
of_or = compute_optical_flow(ms,do_show=False,polar_coord=False)
of_or = np.concatenate([cm.movie(of_or[0]).resize(resize_fact,resize_fact,1)[np.newaxis,:,:,:],cm.movie(of_or[1]).resize(resize_fact,resize_fact,1)[np.newaxis,:,:,:]],axis = 0)
if only_magnitude:
of = np.sqrt(of[0]**2+of[1]**2)
else:
offset_removed = np.min(of_or)
of = of_or - offset_removed
spatial_filter_, time_trace_, norm_fact = extract_components(of,n_components=n_components,verbose = verbose ,normalize_std=False,max_iter=max_iter)
return spatial_filter_, time_trace_, of_or
def extract_motor_components_OF(m, n_components, mask=None, resize_fact=.5, only_magnitude=False, max_iter=1000,
verbose=False, method_factorization='nmf', max_iter_DL=-30):
# todo todocument
if mask is not None:
mask = coo_matrix(np.array(mask).squeeze())
ms = [get_nonzero_subarray(mask.multiply(fr), mask) for fr in m]
ms = np.dstack(ms)
ms = cm.movie(ms.transpose([2, 0, 1]))
else:
ms = m
of_or = compute_optical_flow(ms, do_show=False, polar_coord=False)
of_or = np.concatenate([cm.movie(of_or[0]).resize(resize_fact, resize_fact, 1)[np.newaxis, :, :, :],
cm.movie(of_or[1]).resize(resize_fact, resize_fact, 1)[np.newaxis, :, :, :]], axis=0)
if only_magnitude:
of = np.sqrt(of[0]**2 + of[1]**2)
else:
if method_factorization == 'nmf':
offset_removed = np.min(of_or)
of = of_or - offset_removed
else:
of = of_or
spatial_filter_, time_trace_, _ = extract_components(of, n_components=n_components, verbose=verbose,
normalize_std=False, max_iter=max_iter,
method_factorization=method_factorization,
max_iter_DL=max_iter_DL)
return spatial_filter_, time_trace_, of_or
def save():
global all_ROIs, save_ROIs, summary_images
print('Saving')
ffll = F.getSaveFileName(filter='HDF5 (*.hdf5)')
print(ffll[0])
save_ROIs = np.array(list(all_ROIs.values())).copy()
save_ROIs = np.flip(save_ROIs, axis=1)
if os.path.splitext(ffll[0])[1] == '.hdf5':
cm.movie(save_ROIs).save(ffll[0])
summary_images = summary_images.transpose([0, 2, 1])
summary_images = np.flip(summary_images, axis=1)
def local_correlations_movie_offline(file_name,
Tot_frames=None,
fr: int = 10,
window: int = 30,
stride: int = 3,
swap_dim: bool = True,
eight_neighbours: bool = True,
order_mean: int = 1,
ismulticolor: bool = False,
dview=None):
if Tot_frames is None:
Tot_frames = cm.load(file_name).shape[0]
params: List = [[
file_name,
range(j, j + window), eight_neighbours, swap_dim, order_mean,
ismulticolor
] for j in range(0, Tot_frames - window, stride)]
if dview is None:
parallel_result = list(map(local_correlations_movie_parallel, params))
else:
if 'multiprocessing' in str(type(dview)):
parallel_result = dview.map_async(
local_correlations_movie_parallel, params).get(4294967)
else:
parallel_result = dview.map_sync(local_correlations_movie_parallel,
params)
dview.results.clear()
mm = cm.movie(np.concatenate(parallel_result, axis=0), fr=fr)
return mm
def save_tif_to_mmap_online(movie_iterable,
save_base_name='YrOL_',
order='C',
add_to_movie=0,
data_type=np.uint16,
border_to_0=0):
# todo: todocument
if isinstance(movie_iterable, basestring
): # Allow specifying a filename rather than its data rep
with tifffile.TiffFile(
movie_iterable) as tf: # And load it if that happens
movie_iterable = cm.movie(tf)
count = 0
new_mov = []
dims = (len(movie_iterable), ) + movie_iterable[0].shape
fname_tot = (save_base_name + '_d1_' + str(dims[1]) + '_d2_' +
str(dims[2]) + '_d3_' +
str(1 if len(dims) == 3 else dims[3]) + '_order_' +
str(order) + '_frames_' + str(dims[0]) + '_.mmap')
big_mov = np.memmap(fname_tot,
mode='w+',
dtype=data_type,
shape=prepare_shape((np.prod(dims[1:]), dims[0])),
order=order)
for page in movie_iterable:
if count % 100 == 0:
logging.debug(count)
if 'tifffile' in str(type(movie_iterable[0])):
page = page.asarray()
img = np.array(page, dtype=data_type)
new_img = img
if save_base_name is not None:
big_mov[:, count] = np.reshape(new_img,
np.prod(dims[1:]),
order='F')
else:
new_mov.append(new_img)
if border_to_0 > 0:
img[:border_to_0, :] = 0
img[:, :border_to_0] = 0
img[:, -border_to_0:] = 0
img[-border_to_0:, :] = 0
big_mov[:, count] = np.reshape(img + add_to_movie,
np.prod(dims[1:]),
order='F')
count += 1
big_mov.flush()
del big_mov
return fname_tot
def extract_motor_components_OF(
m,
n_components,
mask=None,
resize_fact: float = .5,
only_magnitude: bool = False,
max_iter: int = 1000,
verbose: bool = False,
method_factorization: str = 'nmf',
max_iter_DL=-30) -> Tuple[np.ndarray, np.ndarray, np.ndarray]:
# todo todocument
if mask is not None:
mask = coo_matrix(np.array(mask).squeeze())
ms = [get_nonzero_subarray(mask.multiply(fr), mask) for fr in m]
ms = np.dstack(ms)
ms = cm.movie(ms.transpose([2, 0, 1]))
else:
ms = m
of_or = compute_optical_flow(ms, do_show=False, polar_coord=False)
of_or = np.concatenate([
cm.movie(of_or[0]).resize(resize_fact, resize_fact,
1)[np.newaxis, :, :, :],
cm.movie(of_or[1]).resize(resize_fact, resize_fact,
1)[np.newaxis, :, :, :]
],
axis=0)
if only_magnitude:
of = np.sqrt(of[0]**2 + of[1]**2)
else:
if method_factorization == 'nmf':
offset_removed = np.min(of_or)
of = of_or - offset_removed
else:
of = of_or
spatial_filter_, time_trace_, _ = extract_components(
of,
n_components=n_components,
verbose=verbose,
normalize_std=False,
max_iter=max_iter,
method_factorization=method_factorization,
max_iter_DL=max_iter_DL)
return spatial_filter_, time_trace_, of_or
def save_tif_to_mmap_online(movie_iterable,
save_base_name='YrOL_',
order='C',
add_to_movie=0,
border_to_0=0) -> str:
# todo: todocument
if isinstance(movie_iterable, basestring
): # Allow specifying a filename rather than its data rep
with tifffile.TiffFile(
movie_iterable) as tf: # And load it if that happens
movie_iterable = cm.movie(tf)
count = 0
new_mov = []
dims = (len(movie_iterable),
) + movie_iterable[0].shape # TODO: Don't pack length into dims
fname_tot = caiman.paths.fn_relocated(
caiman.paths.memmap_frames_filename(save_base_name, dims[1:], dims[0],
order))
big_mov = np.memmap(fname_tot,
mode='w+',
dtype=np.float32,
shape=prepare_shape((np.prod(dims[1:]), dims[0])),
order=order)
for page in movie_iterable:
if count % 100 == 0:
logging.debug(count)
if 'tifffile' in str(type(movie_iterable[0])):
page = page.asarray()
img = np.array(page, dtype=np.float32)
new_img = img
if save_base_name is not None:
big_mov[:, count] = np.reshape(new_img,
np.prod(dims[1:]),
order='F')
else:
new_mov.append(new_img)
if border_to_0 > 0:
img[:border_to_0, :] = 0
img[:, :border_to_0] = 0
img[:, -border_to_0:] = 0
img[-border_to_0:, :] = 0
big_mov[:, count] = np.reshape(img + add_to_movie,
np.prod(dims[1:]),
order='F')
count += 1
big_mov.flush()
del big_mov
return fname_tot
def write_avi(memmap_fpath, result_data_dir):
images = load_images(memmap_fpath)
# Write motion corrected video to drive
w = cm.movie(images)
mcwriter = skvideo.io.FFmpegWriter(result_data_dir + '/mc.avi',
outputdict={'-vcodec': 'rawvideo'})
for iddxx, frame in enumerate(w):
mcwriter.writeFrame(frame.astype('uint8'))
mcwriter.close()
def save_tif_to_mmap_online(movie_iterable,
save_base_name='YrOL_',
order='C',
add_to_movie=0,
border_to_0=0):
if isinstace(movie_iterable, basestring):
with tifffile.TiffFile(movie_iterable) as tf:
movie_iterable = cm.movie(tf)
count = 0
new_mov = []
dims = (len(movie_iterable), ) + movie_iterable[0].shape
fname_tot = save_base_name + '_d1_' + str(dims[1]) + '_d2_' + str(
dims[2]) + '_d3_' + str(1 if len(dims) == 3 else dims[3]
) + '_order_' + str(order) + '_frames_' + str(
dims[0]) + '_.mmap'
big_mov = np.memmap(fname_tot,
mode='w+',
dtype=np.float32,
shape=(np.prod(dims[1:]), dims[0]),
order=order)
for page in movie_iterable:
if count % 100 == 0:
print(count)
if 'tifffile' in str(type(movie_iterable[0])):
page = page.asarray()
img = np.array(page, dtype=np.float32)
new_img = img
if save_base_name is not None:
big_mov[:, count] = np.reshape(new_img,
np.prod(dims[1:]),
order='F')
else:
new_mov.append(new_img)
if border_to_0 > 0:
img[:border_to_0, :] = 0
img[:, :border_to_0] = 0
img[:, -border_to_0:] = 0
img[-border_to_0:, :] = 0
big_mov[:, count] = np.reshape(img + add_to_movie,
np.prod(dims[1:]),
order='F')
count += 1
big_mov.flush()
del big_mov
return fname_tot
def preview_parameters_cropping_cropping_points_temporal_get_in_out_movie(m, cropping_points_temporal):
clip_range = range(cropping_points_temporal[0],cropping_points_temporal[1])
text_width, text_height = cv2.getTextSize('EXCLUDE: Frame 100', fontFace=5, fontScale = 0.8, thickness=1)[0]
m_with_in_out = []
for i, frame in enumerate(m):
m_with_in_out.append(cv2.putText(frame, f'EXCLUDE: Frame {i}' if i in clip_range else f'INCLUDE: Frame {i}',((frame.shape[1] - text_width) // 2,
frame.shape[0] - (text_height + 5)), fontFace=5, fontScale=0.8, color=(255, 255, 255), thickness=1))
m_with_in_out = cm.movie(np.array(m_with_in_out))
return m_with_in_out
def mean_image(file_name,
Tot_frames=None,
fr: float = 10.,
window: int = 100,
dview=None):
"""
Efficient (parallel) computation of mean image in chunks
Args:
Y: str
path to movie file
Tot_frames: int
Number of total frames considered
fr: int (100)
Frame rate (optional)
window: int (100)
Window length in frames
dview: map object
Use it for parallel computation
Returns:
mm: cm.movie (2D).
mean image
"""
if Tot_frames is None:
_, Tot_frames = get_file_size(file_name)
params: List = [[file_name, range(j * window, (j + 1) * window)]
for j in range(int(Tot_frames / window))]
remain_frames = Tot_frames - int(Tot_frames / window) * window
if remain_frames > 0:
params.append([file_name, range(int(Tot_frames / window) * window, Tot_frames)])
if dview is None:
parallel_result = list(map(mean_image_parallel, params))
else:
if 'multiprocessing' in str(type(dview)):
parallel_result = dview.map_async(mean_image_parallel, params).get(4294967)
else:
parallel_result = dview.map_sync(mean_image_parallel, params)
dview.results.clear()
mm = cm.movie(np.concatenate(parallel_result, axis=0), fr=fr/len(parallel_result))
if remain_frames > 0:
mean_image = (mm[:-1].sum(axis=0) + (remain_frames / window) * mm[-1]) / (len(mm) - 1 + remain_frames / window)
else:
mean_image = mm.mean(axis=0)
return mean_image
def save_tif_to_mmap_online(movie_iterable,save_base_name='YrOL_', order = 'C',add_to_movie=0,border_to_0=0):
if isinstace(movie_iterable,basestring):
with tifffile.TiffFile(movie_iterable) as tf:
movie_iterable = cm.movie(tf)
count=0
new_mov=[]
dims=(len(movie_iterable),)+movie_iterable[0].shape
fname_tot = save_base_name + '_d1_' + str(dims[1]) + '_d2_' + str(dims[2]) + '_d3_' + str(
1 if len(dims) == 3 else dims[3]) + '_order_' + str(order) + '_frames_' + str(dims[0]) + '_.mmap'
big_mov = np.memmap(fname_tot, mode='w+', dtype=np.float32,
shape=(np.prod(dims[1:]), dims[0]), order=order)
for page in movie_iterable:
if count%100 == 0:
print(count)
if 'tifffile' in str(type(movie_iterable[0])):
page=page.asarray()
img=np.array(page,dtype=np.float32)
new_img = img
if save_base_name is not None:
big_mov[:,count] = np.reshape(new_img,np.prod(dims[1:]),order='F')
else:
new_mov.append(new_img)
if border_to_0 > 0:
img[:border_to_0,:]=0
img[:,:border_to_0]=0
img[:,-border_to_0:]=0
img[-border_to_0:,:]=0
big_mov[:,count] = np.reshape(img+add_to_movie,np.prod(dims[1:]),order='F')
count+=1
big_mov.flush()
del big_mov
return fname_tot
def correlation_image(images):
'''Compute the correlation image for the input video
Args:
images:
The input video
shape: (frames, dims, dims)
Returns:
Cn:
The correlation image
'''
Cn = cm.movie(images).local_correlations(swap_dim=False)
return Cn
def pre_process_handle(args):
# todo: todocument
from scipy.ndimage import filters as ft
import logging
fil, resize_factors, diameter_bilateral_blur, median_filter_size = args
name_log = fil[:-4] + '_LOG'
logger = logging.getLogger(name_log)
hdlr = logging.FileHandler(name_log)
formatter = logging.Formatter('%(asctime)s %(levelname)s %(message)s')
hdlr.setFormatter(formatter)
logger.addHandler(hdlr)
logger.setLevel(logging.INFO)
logger.info('START')
logger.info(fil)
mov = cm.load(fil, fr=30)
logger.info('Read file')
mov = mov.resize(1, 1, resize_factors[0])
logger.info('Resize')
mov = mov.bilateral_blur_2D(diameter=diameter_bilateral_blur)
logger.info('Bilateral')
mov1 = cm.movie(ft.median_filter(mov, median_filter_size), fr=30)
logger.info('Median filter')
mov1 = mov1.resize(1, 1, resize_factors[1])
logger.info('Resize 2')
mov1 = mov1 - cm.utils.stats.mode_robust(mov1, 0)
logger.info('Mode')
mov = mov.resize(1, 1, resize_factors[1])
logger.info('Resize')
mov.save(fil[:-4] + '_compress_.tif')
logger.info('Save 1')
mov1.save(fil[:-4] + '_BL_compress_.tif')
logger.info('Save 2')
return 1
def pre_process_handle(args):
# todo: todocument
from scipy.ndimage import filters as ft
import logging
fil, resize_factors, diameter_bilateral_blur, median_filter_size = args
name_log = fil[:-4] + '_LOG'
logger = logging.getLogger(name_log)
hdlr = logging.FileHandler(name_log)
formatter = logging.Formatter('%(asctime)s %(levelname)s %(message)s')
hdlr.setFormatter(formatter)
logger.addHandler(hdlr)
logger.setLevel(logging.INFO)
logger.info('START')
logger.info(fil)
mov = cm.load(fil, fr=30)
logger.info('Read file')
mov = mov.resize(1, 1, resize_factors[0])
logger.info('Resize')
mov = mov.bilateral_blur_2D(diameter=diameter_bilateral_blur)
logger.info('Bilateral')
mov1 = cm.movie(ft.median_filter(mov, median_filter_size), fr=30)
logger.info('Median filter')
mov1 = mov1.resize(1, 1, resize_factors[1])
logger.info('Resize 2')
mov1 = mov1 - cm.utils.stats.mode_robust(mov1, 0)
logger.info('Mode')
mov = mov.resize(1, 1, resize_factors[1])
logger.info('Resize')
mov.save(fil[:-4] + '_compress_.tif')
logger.info('Save 1')
mov1.save(fil[:-4] + '_BL_compress_.tif')
logger.info('Save 2')
return 1
def memmap_file(fname, video, dview):
'''This is for changing data format into caiman data and saving into several memmap files.
Args:
fname:
Name of the video (This is only used for working with caiman support)
video:
The input video
shape: (frames, dims, dims)
dview:
Direct View object for parallelization pruposes when using ipyparallel
Returns:
Yr:
Memory mapped variable
dims:
Dimension of one frame
shape: (521,521)
T:
Number of frames
'''
video = np.squeeze(video)
loaded = cm.movie(video,
fr=30,
start_time=0,
file_name=fname[-1],
meta_data=None)
add_to_movie = -np.min(loaded).astype(float)
add_to_movie = np.maximum(add_to_movie, 0)
base_name = 'Yr'
name_new = csp.save_memmap_each(video,
fname,
dview=dview,
base_name=base_name,
add_to_movie=add_to_movie)
name_new.sort()
fname_new = cm.save_memmap_join(name_new, base_name='Yr', dview=dview)
Yr, dims, T = cm.load_memmap(fname_new)
return Yr, dims, T
def main():
mmat = loadmat('mov_AG051514-01-060914 C.mat')['mov']
m = cm.movie(mmat.transpose((2, 0, 1)), fr=120)
mask = select_roi(m[0])
if 1:
mov_tot = compute_optical_flow(m[:3000], mask)
else:
mov_tot = compute_optical_flow(m[:3000], mask, polar_coord=False)
sp_filt, t_trace, norm_fact = extract_components(mov_tot)
plot_components(sp_filt, t_trace)
id_comp = 1
pl.plot(
old_div(
np.sum(np.reshape(sp_filt[id_comp] > 1, [d1, d2]) * mov_tot[1],
axis=(1, 2)), np.sum(sp_filt[id_comp] > 1)))
pl.plot(t_trace[id_comp][:, 1])
def reconstructed_movie(estimates, fnames, idx, scope, flip_signal):
""" Create reconstructed movie in VolPy. The movie has three panels:
motion corrected movie on the left panel, movie removed from the baseline
on the mid panel and reconstructed movie on the right panel.
Args:
estimates: dict
estimates dictionary contain results of VolPy
fnames: list
motion corrected movie in F-order memory mapping format
idx: list
index of selected neurons
scope: list
scope of number of frames in reconstructed movie
flip_signal: boolean
if True the signal will be flipped (for voltron)
Return:
mv_all: 3-D array
motion corrected movie, movie removed from baseline, reconstructed movie
concatenated into one matrix
"""
# motion corrected movie and movie removed from baseline
mv = cm.load(fnames, fr=400)[scope[0]:scope[1]]
dims = (mv.shape[1], mv.shape[2])
mv_bl = mv.computeDFF(secsWindow=0.1)[0]
mv = (mv - mv.min()) / (mv.max() - mv.min())
if flip_signal:
mv_bl = -mv_bl
mv_bl[mv_bl < np.percentile(mv_bl, 3)] = np.percentile(mv_bl, 3)
mv_bl[mv_bl > np.percentile(mv_bl, 98)] = np.percentile(mv_bl, 98)
mv_bl = (mv_bl - mv_bl.min()) / (mv_bl.max() - mv_bl.min())
# reconstructed movie
estimates['weights'][estimates['weights'] < 0] = 0
A = estimates['weights'][idx].transpose([1, 2, 0]).reshape((-1, len(idx)))
C = estimates['t_rec'][idx, scope[0]:scope[1]]
mv_rec = np.dot(A, C).reshape(
(dims[0], dims[1], scope[1] - scope[0])).transpose((2, 0, 1))
mv_rec = cm.movie(mv_rec, fr=400)
mv_rec = (mv_rec - mv_rec.min()) / (mv_rec.max() - mv_rec.min())
mv_all = cm.concatenate((mv, mv_bl, mv_rec), axis=2)
return mv_all
def denoise_mov(movie, ncomp=1, batch_size=1000):
"""
Denoises movie using PCA.
"""
num_frames, h, w = movie.shape
frame_size = h * w
frame_samples = np.reshape(movie, (num_frames, frame_size)).T
ipca_f = sklearn.decomposition.IncrementalPCA(n_components=ncomp,
batch_size=batch_size)
ipca_f.fit(frame_samples)
proj_frame_vectors = ipca_f.inverse_transform(
ipca_f.transform(frame_samples))
eigenseries = ipca_f.components_.T
eigenframes = np.dot(proj_frame_vectors, eigenseries)
movie2 = caiman.movie(
np.reshape(np.float32(proj_frame_vectors.T), movie.shape))
return movie2, eigenframes
def main():
#%
mmat=loadmat('mov_AG051514-01-060914 C.mat')['mov']
m=cm.movie(mmat.transpose((2,0,1)),fr=120)
mask=select_roi(m[0])
if 1:
mov_tot=compute_optical_flow(m[:3000],mask)
else:
mov_tot=compute_optical_flow(m[:3000],mask,polar_coord=False)
sp_filt,t_trace,norm_fact=extract_components(mov_tot)
new_t_trace,coor_1,coor_2 = normalize_components(t_trace,sp_filt)
plot_components(sp_filt,t_trace)
#%
id_comp=1
pl.plot(old_div(np.sum(np.reshape(sp_filt[id_comp]>1,[d1,d2])*mov_tot[1],axis=(1,2)),np.sum(sp_filt[id_comp]>1)))
pl.plot(t_trace[id_comp][:,1])
def make_init_mmap(self):
logger.debug('Making init memmap...')
self.init_dir.mkdir(exist_ok=True, parents=True)
self._validate_tiffs()
mov = tiffs2array(movie_list=self.files,
x_slice=self.xslice,
y_slice=self.yslice,
t_slice=self.tslice)
self.frame_start = mov.shape[0] + 1
self.t = mov.shape[0] + 1
self.params.change_params(dict(init_batch=mov.shape[0]))
m = cm.movie(mov.astype('float32'))
save_path = f'initplane{self.plane}.mmap'
init_mmap = m.save(save_path, order='C')
logger.debug(f'Init mmap saved to {init_mmap}.')
return init_mmap
def local_correlations_movie_offline(file_name, Tot_frames = None, fr = 10, window=30, stride = 3, swap_dim=True, eight_neighbours=True, order_mean = 1, ismulticolor = False, dview = None):
import caiman as cm
if Tot_frames is None:
Tot_frames = cm.load(file_name).shape[0]
params = [[file_name,range(j,j + window), eight_neighbours, swap_dim, order_mean, ismulticolor] for j in range(0,Tot_frames - window,stride)]
if dview is None:
# parallel_result = [self[j:j + window, :, :].local_correlations(
# eight_neighbours=True,swap_dim=swap_dim, order_mean=order_mean)[np.newaxis, :, :] for j in range(T - window)]
parallel_result = list(map(local_correlations_movie_parallel,params))
else:
if 'multiprocessing' in str(type(dview)):
parallel_result = dview.map_async(
local_correlations_movie_parallel, params).get(4294967)
else:
parallel_result = dview.map_sync(
local_correlations_movie_parallel, params)
dview.results.clear()
mm = cm.movie(np.concatenate(parallel_result, axis=0),fr=fr)
return mm
def parameters_test_gSig(path, figname, gSig_filt_list=None):
m_orig = cm.load(path)
if gSig_filt_list == None:
gSig_filt_list = [(2, 2), (4, 4), (6, 6), (8, 8), (10, 10), (20, 20),
(30, 30)]
m_filt_list = []
for i, gSig_filt in enumerate(gSig_filt_list):
m_filt_list.append(
cm.movie(
np.array(
[high_pass_filter_space(m_, gSig_filt) for m_ in m_orig])))
import matplotlib.pyplot as plt
for i, mov in enumerate(m_filt_list):
plt.figure()
plt.imshow(mov[0], cmap='gray')
gSig_size = gSig_filt_list[i]
plt.title(f'{figname} \n gSig_size = {gSig_size}')
plt.savefig(
f'data/motion_correction/png/{figname}_gSig_experiment_{gSig_size}.png'
)
return
def process_movie_parallel(arg_in):
fname,fr,margins_out,template,max_shift_w, max_shift_h,remove_blanks,apply_smooth,save_hdf5=arg_in
if template is not None:
if isinstance(template,basestring):
if os.path.exists(template):
template=cm.load(template,fr=1)
else:
raise Exception('Path to template does not exist:'+template)
# with open(fname[:-4]+'.stout', "a") as log:
# print fname
# sys.stdout = log
# import pdb
# pdb.set_trace()
type_input = str(type(fname))
if 'movie' in type_input:
print((type(fname)))
Yr=fname
elif ('ndarray' in type_input):
Yr=cm.movie(np.array(fname,dtype=np.float32),fr=fr)
elif isinstance(fname,basestring):
Yr=cm.load(fname,fr=fr)
else:
raise Exception('Unkown input type:' + type_input)
if Yr.ndim>1:
print('loaded')
if apply_smooth:
print('applying smoothing')
Yr=Yr.bilateral_blur_2D(diameter=10,sigmaColor=10000,sigmaSpace=0)
# bl_yr=np.float32(np.percentile(Yr,8))
# Yr=Yr-bl_yr # needed to remove baseline
print('Remove BL')
if margins_out!=0:
Yr=Yr[:,margins_out:-margins_out,margins_out:-margins_out] # borders create troubles
print('motion correcting')
Yr,shifts,xcorrs,template=Yr.motion_correct(max_shift_w=max_shift_w, max_shift_h=max_shift_h, method='opencv',template=template,remove_blanks=remove_blanks)
# Yr = Yr + bl_yr
if ('movie' in type_input) or ('ndarray' in type_input):
print('Returning Values')
return Yr, shifts, xcorrs, template
else:
print('median computing')
template=Yr.bin_median()
print('saving')
idx_dot=len(fname.split('.')[-1])
if save_hdf5:
Yr.save(fname[:-idx_dot]+'hdf5')
print('saving 2')
np.savez(fname[:-idx_dot]+'npz',shifts=shifts,xcorrs=xcorrs,template=template)
print('deleting')
del Yr
print('done!')
return fname[:-idx_dot]
#sys.stdout = sys.__stdout__
else:
return None
Y = np.reshape(Yr, dims + (T,), order='F')
m_orig = np.array(images)
patch_size = 50
half_crop = np.minimum(gSig[0]*4+1,patch_size)//2
idx_included = np.arange(A_gt.shape[-1])
#idx_included = idx_components_cnn[np.arange(50)] # neurons that are displayed
all_pos_crops = []
all_neg_crops = []
all_dubious_crops = []
base_name = fname_new.split('/')[5]
dims = (d1,d2)
idx_included = np.array([x for x in idx_included if x is not None])
base_name = base_name + '_' + str(idx_included[0]) + '_' + str(idx_included[-1])
print(base_name)
idx_excluded = np.setdiff1d(np.arange(A_gt.shape[-1]),idx_included)
m_res = m_orig - cm.movie(np.reshape(A_gt.tocsc()[:,idx_excluded].dot(C_gt[idx_excluded]) + b_gt.dot(f_gt),dims+(-1,),order = 'F').transpose([2,0,1]))
mean_proj = np.mean(m_res,0)
std_mov = mean_proj.std()
min_mov = np.median(mean_proj)
#%%
full_fov = False # will use either the overfeat like (full FOV, train_net_cifar_edge_cutter_FOV.py) network or the single patch one (train_net_cifar_edge_cutter.py)
count_start = 30
dims = np.array(dims)
#bin_ = 10
cms_total = [np.array(scipy.ndimage.center_of_mass(np.reshape(a.toarray(),dims,order = 'F'))).astype(np.int) for a in A_gt.tocsc()[:,idx_included].T]
cms_total = np.maximum(cms_total ,half_crop)
cms_total = np.array([np.minimum(cms,dims-half_crop) for cms in cms_total ]).astype(np.int)
for count in range(count_start,T):
if count%1000==0:
print(count)
print(count)
def save_memmap(filenames,
base_name='Yr',
resize_fact=(1, 1, 1),
remove_init=0,
idx_xy=None,
order='F',
xy_shifts=None,
is_3D=False,
add_to_movie=0,
border_to_0=0,
dview=None,
n_chunks=100,
slices=None):
""" Efficiently write data from a list of tif files into a memory mappable file
Args:
filenames: list
list of tif files or list of numpy arrays
base_name: str
the base used to build the file name. IT MUST NOT CONTAIN "_"
resize_fact: tuple
x,y, and z downsampling factors (0.5 means downsampled by a factor 2)
remove_init: int
number of frames to remove at the begining of each tif file
(used for resonant scanning images if laser in rutned on trial by trial)
idx_xy: tuple size 2 [or 3 for 3D data]
for selecting slices of the original FOV, for instance
idx_xy = (slice(150,350,None), slice(150,350,None))
order: string
whether to save the file in 'C' or 'F' order
xy_shifts: list
x and y shifts computed by a motion correction algorithm to be applied before memory mapping
is_3D: boolean
whether it is 3D data
add_to_movie: floating-point
value to add to each image point, typically to keep negative values out.
border_to_0: (undocumented)
dview: (undocumented)
n_chunks: (undocumented)
slices: slice object or list of slice objects
slice can be used to select portion of the movies in time and x,y
directions. For instance
slices = [slice(0,200),slice(0,100),slice(0,100)] will take
the first 200 frames and the 100 pixels along x and y dimensions.
Returns:
fname_new: the name of the mapped file, the format is such that
the name will contain the frame dimensions and the number of frames
"""
if type(filenames) is not list:
raise Exception('input should be a list of filenames')
if slices is not None:
slices = [slice(0, None) if sl is None else sl for sl in slices]
if len(filenames) > 1:
recompute_each_memmap = False
for file__ in filenames:
if ('order_' + order not in file__) or ('.mmap' not in file__):
recompute_each_memmap = True
if recompute_each_memmap or (remove_init>0) or (idx_xy is not None)\
or (xy_shifts is not None) or (add_to_movie != 0) or (border_to_0>0)\
or slices is not None:
logging.debug('Distributing memory map over many files')
# Here we make a bunch of memmap files in the right order. Same parameters
fname_new = cm.save_memmap_each(filenames,
base_name=base_name,
order=order,
border_to_0=border_to_0,
dview=dview,
resize_fact=resize_fact,
remove_init=remove_init,
idx_xy=idx_xy,
xy_shifts=xy_shifts,
slices=slices,
add_to_movie=add_to_movie)
else:
fname_new = filenames
# The goal is to make a single large memmap file, which we do here
if order == 'F':
raise exception(
'You cannot merge files in F order, they must be in C order for CaImAn'
)
fname_new = cm.save_memmap_join(fname_new,
base_name=base_name,
dview=dview,
n_chunks=n_chunks)
else:
# TODO: can be done online
Ttot = 0
for idx, f in enumerate(filenames):
if isinstance(f, str): # Might not always be filenames.
logging.debug(f)
if is_3D:
Yr = f if not (isinstance(f,
basestring)) else tifffile.imread(f)
if slices is not None:
Yr = Yr[slices]
else:
if idx_xy is None: #todo remove if not used, superceded by the slices parameter
Yr = Yr[remove_init:]
elif len(
idx_xy
) == 2: #todo remove if not used, superceded by the slices parameter
Yr = Yr[remove_init:, idx_xy[0], idx_xy[1]]
else: #todo remove if not used, superceded by the slices parameter
Yr = Yr[remove_init:, idx_xy[0], idx_xy[1], idx_xy[2]]
else:
Yr = cm.load(f, fr=1, in_memory=True) if (isinstance(
f, basestring) or isinstance(f, list)) else cm.movie(
f) # TODO: Rewrite more legibly
if xy_shifts is not None:
Yr = Yr.apply_shifts(xy_shifts,
interpolation='cubic',
remove_blanks=False)
if slices is not None:
Yr = Yr[slices]
else:
if idx_xy is None:
if remove_init > 0:
Yr = Yr[remove_init:]
elif len(idx_xy) == 2:
Yr = Yr[remove_init:, idx_xy[0], idx_xy[1]]
else:
raise Exception(
'You need to set is_3D=True for 3D data)')
Yr = np.array(Yr)[remove_init:, idx_xy[0], idx_xy[1],
idx_xy[2]]
if border_to_0 > 0:
if slices is not None:
if type(slices) is list:
raise Exception(
'You cannot slice in x and y and then use add_to_movie: if you only want to slice in time do not pass in a list but just a slice object'
)
min_mov = Yr.calc_min()
Yr[:, :border_to_0, :] = min_mov
Yr[:, :, :border_to_0] = min_mov
Yr[:, :, -border_to_0:] = min_mov
Yr[:, -border_to_0:, :] = min_mov
fx, fy, fz = resize_fact
if fx != 1 or fy != 1 or fz != 1:
if 'movie' not in str(type(Yr)):
Yr = cm.movie(Yr, fr=1)
Yr = Yr.resize(fx=fx, fy=fy, fz=fz)
T, dims = Yr.shape[0], Yr.shape[1:]
Yr = np.transpose(Yr, list(range(1, len(dims) + 1)) + [0])
Yr = np.reshape(Yr, (np.prod(dims), T), order='F')
Yr = np.ascontiguousarray(Yr, dtype=np.float32) + np.float32(
0.0001) + np.float32(add_to_movie)
if idx == 0:
fname_tot = base_name + '_d1_' + str(dims[0]) + '_d2_' + str(
dims[1]) + '_d3_' + str(
1 if len(dims) == 2 else dims[2]) + '_order_' + str(
order) # TODO: Rewrite more legibly
if isinstance(f, str):
fname_tot = os.path.join(os.path.split(f)[0], fname_tot)
if len(filenames) > 1:
big_mov = np.memmap(fname_tot,
mode='w+',
dtype=np.float32,
shape=prepare_shape(
(np.prod(dims), T)),
order=order)
big_mov[:, Ttot:Ttot + T] = Yr
del big_mov
else:
logging.debug('SAVING WITH numpy.tofile()')
Yr.tofile(fname_tot)
else:
big_mov = np.memmap(fname_tot,
dtype=np.float32,
mode='r+',
shape=prepare_shape(
(np.prod(dims), Ttot + T)),
order=order)
big_mov[:, Ttot:Ttot + T] = Yr
del big_mov
sys.stdout.flush()
Ttot = Ttot + T
fname_new = fname_tot + '_frames_' + str(Ttot) + '_.mmap'
try:
# need to explicitly remove destination on windows
os.unlink(fname_new)
except OSError:
pass
os.rename(fname_tot, fname_new)
return fname_new
mc = MotionCorrect(fname, min_mov,
dview=dview, max_shifts=max_shifts, niter_rig=niter_rig, splits_rig=splits_rig,
num_splits_to_process_rig=num_splits_to_process_rig,
strides=strides, overlaps=overlaps, splits_els=splits_els,
num_splits_to_process_els=num_splits_to_process_els,
upsample_factor_grid=upsample_factor_grid, max_deviation_rigid=max_deviation_rigid,
shifts_opencv=True, nonneg_movie=True)
#%%
mc.motion_correct_rigid(save_movie=True)
# load motion corrected movie
#%%
pl.imshow(mc.total_template_rig, cmap='gray')
#%% visualize templates
cm.movie(np.array(mc.templates_rig)).play(
fr=10, gain=5, magnification=2, offset=offset_mov)
#%% plot rigid shifts
pl.close()
pl.plot(mc.shifts_rig)
pl.legend(['x shifts', 'y shifts'])
pl.xlabel('frames')
pl.ylabel('pixels')
#%% inspect movie
bord_px_rig = np.ceil(np.max(mc.shifts_rig)).astype(np.int)
downsample_ratio = .2
m_rig.resize(1, 1, downsample_ratio).play(
gain=10, offset=offset_mov * .25, fr=30, magnification=2, bord_px=bord_px_rig)
#%%
mc.motion_correct_pwrigid(
save_movie=True, template=mc.total_template_rig, show_template=True)
m_els = cm.load(mc.fname_tot_els)
'thr_plot': 0.8
}
# TODO: do find&replace on those parameters and delete this paragrph
# @params fname name of the movie
fname_new = params_movie['fname'][0]
# %% RUN ANALYSIS
c, dview, n_processes = cm.cluster.setup_cluster(
backend='local', n_processes=None, single_thread=False)
# %% LOAD MEMMAP FILE
# fname_new='Yr_d1_501_d2_398_d3_1_order_F_frames_369_.mmap'
Yr, dims, T = cm.load_memmap(fname_new)
d1, d2 = dims
images = np.reshape(Yr.T, [T] + list(dims), order='F')
Y = np.reshape(Yr, dims + (T,), order='F')
m_images = cm.movie(images)
# %% correlation image
if m_images.shape[0] < 10000:
Cn = m_images.local_correlations(
swap_dim=params_movie['swap_dim'], frames_per_chunk=1500)
Cn[np.isnan(Cn)] = 0
else:
Cn = np.array(cm.load(('/'.join(params_movie['gtname'][0].split('/')[:-2] + [
'projections', 'correlation_image_better.tif'])))).squeeze()
pl.imshow(Cn, cmap='gray', vmax=.95)
# TODO: show screenshot 11
#%%
import cv2
if not '.mat' in params_movie['seed_name'][0]:
roi_cons = np.load(params_movie['seed_name'][0])
else:
fname_new = params_movie['fname']
# %% RUN ANALYSIS
c, dview, n_processes = cm.cluster.setup_cluster(
backend='local', n_processes=None, single_thread=False)
# %% LOAD MEMMAP FILE
# fname_new='Yr_d1_501_d2_398_d3_1_order_F_frames_369_.mmap'
Yr, dims, T = cm.load_memmap(fname_new)
d1, d2 = dims
images = np.reshape(Yr.T, [T] + list(dims), order='F')
# TODO: needinfo
Y = np.reshape(Yr, dims + (T,), order='F')
m_images = cm.movie(images)
# TODO: show screenshot 10
# %% correlation image
if m_images.shape[0] < 10000:
Cn = m_images.local_correlations(
swap_dim=params_movie['swap_dim'], frames_per_chunk=1500)
Cn[np.isnan(Cn)] = 0
else:
Cn = np.array(cm.load(('/'.join(fname_new.split('/')
[:-3] + ['projections', 'correlation_image_better.tif'])))).squeeze()
pl.imshow(Cn, cmap='gray', vmax=.95)
# %% some parameter settings
# order of the autoregressive fit to calcium imaging in general one (slow gcamps) or two (fast gcamps fast scanning)
p = params_movie['p']
# pl.figure();
pl.subplot(1, 3, 1)
crd = plot_contours(A.tocsc()[:, idx_components], Cn, thr=0.9)
pl.subplot(1, 3, 2)
crd = plot_contours(A.tocsc()[:, idx_blobs], Cn, thr=0.9)
pl.subplot(1, 3, 3)
crd = plot_contours(A.tocsc()[:, idx_components_bad], Cn, thr=0.9)
#%%
#idx_very_nice=[2, 19, 23, 27,32,43,45,49,51,94,100]
# idx_very_nice=np.array(idx_very_nice)[np.array([3,4,8,10])]
# idx_very_nice=idx_blobs[idx_very_nice]
idx_very_nice = idx_blobs
view_patches_bar(Yr, scipy.sparse.coo_matrix(A.tocsc()[:, idx_very_nice]), C[
idx_very_nice, :], b, f, dims[0], dims[1], YrA=YrA[idx_very_nice, :], img=Cn)
#%%
new_m = cm.movie(np.reshape(A.tocsc()[
:, idx_blobs] * C[idx_blobs] + b.dot(f), dims + (-1,), order='F').transpose([2, 0, 1]))
new_m.play(fr=30, backend='opencv', gain=7., magnification=3.)
#%%
new_m = cm.movie(np.reshape(A.tocsc()[:, idx_blobs] * C[idx_blobs] +
b * np.median(f), dims + (-1,), order='F').transpose([2, 0, 1]))
new_m.play(fr=30, backend='opencv', gain=7., magnification=3.)
#%%
new_m = cm.movie(np.reshape(A.tocsc()[
:, idx_blobs] * C[idx_blobs], dims + (-1,), order='F').transpose([2, 0, 1]))
new_m.play(fr=30, backend='opencv', gain=30., magnification=3.)
#%%
# idx_to_show=[0,1,5,8,14,17,18,23,24,25,26,28,29,31,32,33,34,36,43,45,47,51,53,54,57,60,61,62,63,64,65,66,67,71,72,74,75,78,79,80,81,91,95,96,97,99,102]
#cm.view_patches_bar(Yr,scipy.sparse.coo_matrix(A.tocsc()[:,sure_in_idx[idx_to_show]]),C[sure_in_idx[idx_to_show],:],b,f, dims[0],dims[1], YrA=YrA[sure_in_idx[idx_to_show],:],img=np.mean(Y,-1))
#%%
# idx_to_show=[0,1,5,8,14,17,18,23,24,25,26,28,29,31,32,33,34,36,43,45,47,51,53,54,57,60,61,62,63,64,65,66,67,71,72,74,75,78,79,80,81,91,95,96,97,99,102]
# idx_to_show=np.array(idx_to_show)[[2,19,23,26,34]]
tr_BL = scipy.ndimage.zoom(np.array(tr_BL, dtype=np.float32), [
downsampfact, 1], order=3, mode='constant', cval=0.0, prefilter=True)
if padafter == 0:
traces_gt -= tr_BL.T
else:
traces_gt -= tr_BL[padbefore:-padafter].T
#traces_gt = scipy.signal.savgol_filter(traces_gt,5,2)
#%
fitness, exceptionality, sd_r, md = cm.components_evaluation.compute_event_exceptionality(
traces_gt[idx_included], robust_std=False, N=5, use_mode_fast=False)
fitness_d, exceptionality_d, sd_r_d, md_d = cm.components_evaluation.compute_event_exceptionality(
np.diff(traces_gt[idx_included], axis=1), robust_std=False, N=3, use_mode_fast=False)
#%
m_res = m_orig - cm.movie(np.reshape(A_gt.tocsc()[:, idx_excluded].dot(
C_gt[idx_excluded]) + b_gt.dot(f_gt), dims + (-1,), order='F').transpose([2, 0, 1]))
#%
# mean_proj = np.mean(m_res,0)
#%
# max_mov = mean_proj.max()
#%%
# min_mov = mean_proj.min()
#%%
# m_res = (m_res-min_mov)/(max_mov-min_mov)
#%%
# m_res.play()
#%%
# cnn = False
# if cnn:
# import json as simplejson
# from keras.models import model_from_json
#a = cm.load('/mnt/ceph/neuro/labeling/yuste.Single_150u/images/tifs/Single_150um_024.tif')
#a = cm.load('/Users/agiovann/example_movies_ALL/quietBlock_2_ds_2_2.hdf5')
a = cm.load('/mnt/ceph/neuro/labeling/yuste.Single_150u/images/tifs/Yr_d1_200_d2_256_d3_1_order_C_frames_3000_.mmap')
all_els = []
for it in range(1):
print(it)
# a = cm.movie(np.random.randn(*mns.shape).astype(np.float32))
Yr = remove_baseline_fast(np.array(cm.movie.to_2D(a)).T).T
#%
# norm = lambda(x): np.exp(-x**2/2)/np.sqrt(2*np.pi)
fitness, res, sd_r, md = compute_event_exceptionality(Yr.T)
Yr_c = -np.log(norm.sf(np.array((Yr - md) /
sd_r, dtype=np.float)))
mns = cm.movie(scipy.ndimage.convolve(np.reshape(
Yr_c, [-1, a.shape[1], a.shape[2]], order='F'), np.ones([5, 3, 3])))
mns[mns < (38 * np.log10((mns.shape[0])))] = 0
all_els.append(np.sum(mns > 0)/Yr.size)
print(all_els)
#%%
#m1 = cm.movie((np.array((Yr-md)/sd_r)).reshape([-1,60,80],order = 'F'))*(scipy.ndimage.convolve(mns>0,np.ones([5,3,3])))
m1 = cm.movie((np.array((Yr - md) / sd_r)
).reshape([-1, a.shape[1],a.shape[2]], order='F')) * (mns > 0)
#%%
m2 = cm.movie((np.array((Yr-md)/sd_r)).reshape(m1.shape,order = 'F'))*(scipy.ndimage.convolve(mns>0,np.ones([5,3,3])))
#%%
if False:
b1 = np.percentile(Yr,20,axis=0).reshape([a.shape[1],a.shape[2]], order='F')
#%%
m_rig = cm.load(fname_tot_rig)
#%%
add_to_movie = - np.min(total_template_rig) + 1
print(add_to_movie)
#%% visualize movies
m_rig.resize(1, 1, .2).play(
fr=20, gain=5, magnification=1, offset=add_to_movie)
#%%
downs = .2
cm.concatenate([m_rig.resize(1, 1, downs), m_orig.resize(1, 1, downs)], axis=1).play(
fr=30, gain=2, magnification=1, offset=add_to_movie)
#%% visualize templates
cm.movie(np.array(templates_rig)).play(
fr=10, gain=2, magnification=1, offset=add_to_movie)
#%% PIECEWISE RIGID MOTION CORRECTION
t1 = time.time()
new_templ = total_template_rig.copy()
strides = params_movie['strides']
overlaps = params_movie['overlaps']
shifts_opencv = False
save_movie = True
splits = params_movie['splits_els']
num_splits_to_process_list = params_movie['num_splits_to_process_els']
upsample_factor_grid = params_movie['upsample_factor_grid']
max_deviation_rigid = params_movie['max_deviation_rigid']
add_to_movie = - np.min(total_template_rig) + 1
num_iter = 1
for num_splits_to_process in num_splits_to_process_list[-1:]:
fname_tot_els, total_template_wls, templates_els, x_shifts_els, y_shifts_els, coord_shifts_els = cm.motion_correction.motion_correct_batch_pwrigid(fname, max_shifts, strides, overlaps, add_to_movie, newoverlaps=None, newstrides=None,
mc_list.append(mc)
t_a = time.time() - t1
print(time.time() - t1)
# %%
# load motion corrected movie
if isscreen:
m_rig = cm.load(mc_list[-2].fname_tot_rig)
m_rig.resize(1, 1, .2).play(gain=1, offset=offset_mov, fr=30)
pl.imshow(mc.total_template_rig, cmap='gray')
#%%
cm.movie(np.array([mc_.total_template_rig for mc_ in mc_list])).play(
fr=8, gain=1, offset=offset_mov)
# %% visualize templates
cm.movie(np.array(mc.templates_rig)).play(
fr=10, gain=1, magnification=1, offset=offset_mov)
# %% plot rigid shifts
pl.close()
pl.plot(np.vstack([mc.shifts_rig for mc in mc_list]))
pl.legend(['x shifts', 'y shifts'])
pl.xlabel('frames')
pl.ylabel('pixels')
# TODO: show screenshot 9
# %% restart cluster to clean up memory
# TODO: todocument
# THIS SOMETIMES GETS STUCK IN SCRIPT MODE
c, dview, n_processes = cm.cluster.setup_cluster(
backend='local', n_processes=None, single_thread=False)
cv2.waitKey(300)
count_neuro += 1
print(count_neuro)
break
# crop_img = cv2.resize(crop_img,dsize=None,fx=2,fy=2)
# newshape = np.array(crop_img.shape)//2
# crop_img = crop_img[newshape[0]-half_crop:newshape[0]+half_crop,newshape[0]-half_crop:newshape[0]+half_crop]
# NOTE: its img[y: y + h, x: x + w] and *not* img[x: x + w, y: y + h]
id_file += 1
all_masks_gt = np.vstack(all_masks_gt)
#%%
cm.movie(np.squeeze(all_masks_gt[labels_gt == 0])).play(
gain=3., magnification=10)
#%%
np.savez('ground_truth_components_minions.npz',
all_masks_gt=all_masks_gt, labels_gt=labels_gt, traces_gt=traces_gt)
#%%
import itertools
def grouper(n, iterable, fillvalue=None):
"grouper(3, 'ABCDEFG', 'x') --> ABC DEF Gxx"
args = [iter(iterable)] * n
return itertools.zip_longest(*args, fillvalue=fillvalue)
#%% curate once more. Remove wrong negatives
negatives = np.where(labels_gt == 1)[0]
def save_memmap(filenames, base_name='Yr', resize_fact=(1, 1, 1), remove_init=0, idx_xy=None,
order='F', xy_shifts=None, is_3D=False, add_to_movie=0, border_to_0=0, dview = None,
n_chunks=100):
""" Efficiently write data from a list of tif files into a memory mappable file
Parameters:
----------
filenames: list
list of tif files or list of numpy arrays
base_name: str
the base used to build the file name. IT MUST NOT CONTAIN "_"
resize_fact: tuple
x,y, and z downsampling factors (0.5 means downsampled by a factor 2)
remove_init: int
number of frames to remove at the begining of each tif file
(used for resonant scanning images if laser in rutned on trial by trial)
idx_xy: tuple size 2 [or 3 for 3D data]
for selecting slices of the original FOV, for instance
idx_xy = (slice(150,350,None), slice(150,350,None))
order: string
whether to save the file in 'C' or 'F' order
xy_shifts: list
x and y shifts computed by a motion correction algorithm to be applied before memory mapping
is_3D: boolean
whether it is 3D data
add_to_movie: floating-point
value to add to each image point, typically to keep negative values out.
border_to_0: (undocumented)
dview: (undocumented)
n_chunks: (undocumented)
Returns:
-------
fname_new: the name of the mapped file, the format is such that
the name will contain the frame dimensions and the number of frames
"""
if type(filenames) is not list:
raise Exception('input should be a list of filenames')
if len(filenames) > 1:
is_inconsistent_order = False
for file__ in filenames:
if ('order_' + order not in file__) or ('.mmap' not in file__):
is_inconsistent_order = True
if is_inconsistent_order: # Here we make a bunch of memmap files in the right order. Same parameters
fname_new = cm.save_memmap_each(filenames,
base_name = base_name,
order = order,
border_to_0 = border_to_0,
dview = dview,
resize_fact = resize_fact,
remove_init = remove_init,
idx_xy = idx_xy,
xy_shifts = xy_shifts,
add_to_movie = add_to_movie)
else:
fname_new = filenames
# The goal is to make a single large memmap file, which we do here
if order == 'F':
raise exception('You cannot merge files in F order, they must be in C order')
fname_new = cm.save_memmap_join(fname_new, base_name=base_name, dview=dview, n_chunks=n_chunks, add_to_mov = add_to_movie)
else:
# TODO: can be done online
Ttot = 0
for idx, f in enumerate(filenames):
if isinstance(f, str): # Might not always be filenames.
print(f)
if is_3D:
Yr = f if not(isinstance(f, basestring)) else tifffile.imread(f)
if idx_xy is None:
Yr = Yr[remove_init:]
elif len(idx_xy) == 2:
Yr = Yr[remove_init:, idx_xy[0], idx_xy[1]]
else:
Yr = Yr[remove_init:, idx_xy[0], idx_xy[1], idx_xy[2]]
else:
Yr = cm.load(f, fr=1, in_memory=True) if (isinstance(f, basestring) or isinstance(f, list)) else cm.movie(f) # TODO: Rewrite more legibly
if xy_shifts is not None:
Yr = Yr.apply_shifts(xy_shifts, interpolation='cubic', remove_blanks=False)
if idx_xy is None:
if remove_init > 0:
Yr = Yr[remove_init:]
elif len(idx_xy) == 2:
Yr = Yr[remove_init:, idx_xy[0], idx_xy[1]]
else:
raise Exception('You need to set is_3D=True for 3D data)')
Yr = np.array(Yr)[remove_init:, idx_xy[0], idx_xy[1], idx_xy[2]]
if border_to_0 > 0:
min_mov = Yr.calc_min()
Yr[:, :border_to_0, :] = min_mov
Yr[:, :, :border_to_0] = min_mov
Yr[:, :, -border_to_0:] = min_mov
Yr[:, -border_to_0:, :] = min_mov
fx, fy, fz = resize_fact
if fx != 1 or fy != 1 or fz != 1:
if 'movie' not in str(type(Yr)):
Yr = cm.movie(Yr, fr=1)
Yr = Yr.resize(fx=fx, fy=fy, fz=fz)
T, dims = Yr.shape[0], Yr.shape[1:]
Yr = np.transpose(Yr, list(range(1, len(dims) + 1)) + [0])
Yr = np.reshape(Yr, (np.prod(dims), T), order='F')
Yr = np.ascontiguousarray(Yr, dtype=np.float32) + 0.0001 + add_to_movie
if idx == 0:
fname_tot = base_name + '_d1_' + str(dims[0]) + '_d2_' + str(dims[1]) + '_d3_' + str(
1 if len(dims) == 2 else dims[2]) + '_order_' + str(order) # TODO: Rewrite more legibly
if isinstance(f, str):
fname_tot = os.path.join(os.path.split(f)[0], fname_tot)
if len(filenames) > 1:
big_mov = np.memmap(fname_tot, mode='w+', dtype=np.float32,
shape=prepare_shape((np.prod(dims), T)), order=order)
big_mov[:, Ttot:Ttot + T] = Yr
del big_mov
else:
print('SAVING WITH numpy.tofile()')
Yr.tofile(fname_tot)
else:
big_mov = np.memmap(fname_tot, dtype=np.float32, mode='r+',
shape=prepare_shape((np.prod(dims), Ttot + T)), order=order)
big_mov[:, Ttot:Ttot + T] = Yr
del big_mov
sys.stdout.flush()
Ttot = Ttot + T
fname_new = fname_tot + '_frames_' + str(Ttot) + '_.mmap'
try:
# need to explicitly remove destination on windows
os.unlink(fname_new)
except OSError:
pass
os.rename(fname_tot, fname_new)
return fname_new
def save_memmap(filenames, base_name='Yr', resize_fact=(1, 1, 1), remove_init=0, idx_xy=None, order='F',xy_shifts=None,is_3D=False,add_to_movie=0,border_to_0=0):
""" Saves efficiently a list of tif files into a memory mappable file
Parameters
----------
filenames: list
list of tif files
base_name: str
the base used to build the file name. IT MUST NOT CONTAIN "_"
resize_fact: tuple
x,y, and z downampling factors (0.5 means downsampled by a factor 2)
remove_init: int
number of frames to remove at the begining of each tif file (used for resonant scanning images if laser in rutned on trial by trial)
idx_xy: tuple size 2 [or 3 for 3D data]
for selecting slices of the original FOV, for instance idx_xy=(slice(150,350,None),slice(150,350,None))
order: string
whether to save the file in 'C' or 'F' order
xy_shifts: list
x and y shifts computed by a motion correction algorithm to be applied before memory mapping
is_3D: boolean
whether it is 3D data
Return
-------
fname_new: the name of the mapped file, the format is such that the name will contain the frame dimensions and the number of f
"""
#TODO: can be done online
Ttot = 0
for idx, f in enumerate(filenames):
print(f)
if is_3D:
import tifffile
print("Using tifffile library instead of skimage because of 3D")
if idx_xy is None:
Yr = tifffile.imread(f)[remove_init:]
elif len(idx_xy) == 2:
Yr = tifffile.imread(f)[remove_init:, idx_xy[0], idx_xy[1]]
else:
Yr = tifffile.imread(f)[remove_init:, idx_xy[0], idx_xy[1], idx_xy[2]]
# elif :
#
# if xy_shifts is not None:
# raise Exception('Calblitz not installed, you cannot motion correct')
#
# if idx_xy is None:
# Yr = imread(f)[remove_init:]
# elif len(idx_xy) == 2:
# Yr = imread(f)[remove_init:, idx_xy[0], idx_xy[1]]
# else:
# raise Exception('You need to set is_3D=True for 3D data)')
else:
Yr=cm.load(f,fr=1)
if xy_shifts is not None:
Yr=Yr.apply_shifts(xy_shifts,interpolation='cubic',remove_blanks=False)
if idx_xy is None:
Yr = np.array(Yr)[remove_init:]
elif len(idx_xy) == 2:
Yr = np.array(Yr)[remove_init:, idx_xy[0], idx_xy[1]]
else:
raise Exception('You need to set is_3D=True for 3D data)')
Yr = np.array(Yr)[remove_init:, idx_xy[0], idx_xy[1], idx_xy[2]]
if border_to_0>0:
min_mov=np.min(Yr)
Yr[:,:border_to_0,:]=min_mov
Yr[:,:,:border_to_0]=min_mov
Yr[:,:,-border_to_0:]=min_mov
Yr[:,-border_to_0:,:]=min_mov
fx, fy, fz = resize_fact
if fx != 1 or fy != 1 or fz != 1:
Yr = cm.movie(Yr, fr=1)
Yr = Yr.resize(fx=fx, fy=fy, fz=fz)
T, dims = Yr.shape[0], Yr.shape[1:]
Yr = np.transpose(Yr, list(range(1, len(dims) + 1)) + [0])
Yr = np.reshape(Yr, (np.prod(dims), T), order='F')
if idx == 0:
fname_tot = base_name + '_d1_' + str(dims[0]) + '_d2_' + str(dims[1]) + '_d3_' + str(
1 if len(dims) == 2 else dims[2]) + '_order_' + str(order)
fname_tot = os.path.join(os.path.split(f)[0],fname_tot)
big_mov = np.memmap(fname_tot, mode='w+', dtype=np.float32,
shape=(np.prod(dims), T), order=order)
else:
big_mov = np.memmap(fname_tot, dtype=np.float32, mode='r+',
shape=(np.prod(dims), Ttot + T), order=order)
# np.save(fname[:-3]+'npy',np.asarray(Yr))
big_mov[:, Ttot:Ttot + T] = np.asarray(Yr, dtype=np.float32) + 1e-10 + add_to_movie
big_mov.flush()
del big_mov
Ttot = Ttot + T
fname_new = fname_tot + '_frames_' + str(Ttot) + '_.mmap'
os.rename(fname_tot, fname_new)
return fname_new
with open(json_path, "w") as json_file:
json_file.write(simplejson.dumps(simplejson.loads(model_json), indent=4))
print('Saved trained model at %s ' % json_path)
if not os.path.isdir(save_dir):
os.makedirs(save_dir)
model_path = os.path.join(save_dir, model_name + '.h5')
model.save(model_path)
print('Saved trained model at %s ' % model_path)
#%% visualize_results
predictions = model.predict(all_masks_gt, batch_size=32, verbose=1)
cm.movie(np.squeeze(all_masks_gt[np.where(predictions[:, 0] >= 0.5)[0]])).play(
gain=3., magnification=5, fr=10)
#%%
cm.movie(np.squeeze(all_masks_gt[np.where(predictions[:, 1] >= 0.5)[0]])).play(
gain=3., magnification=5, fr=10)
#%%
cm.movie(np.squeeze(all_masks_gt[np.where(predictions[:, 2] >= 0.5)[0]])).play(
gain=3., magnification=5, fr=10)
#%% retrieve and test
json_file = open(json_path, 'r')
loaded_model_json = json_file.read()
json_file.close()
loaded_model = model_from_json(loaded_model_json)
loaded_model.load_weights(model_path)
opt = keras.optimizers.rmsprop(lr=0.0001, decay=1e-6)
loaded_model.compile(loss=keras.losses.categorical_crossentropy,
max_shifts, dview=dview, splits=splits, num_splits_to_process=num_splits_to_process,
num_iter=num_iter, template=total_template_rig, shifts_opencv=shifts_opencv, save_movie_rigid=save_movie_rigid, add_to_movie=add_to_movie)
total_shifts.append(shifts_rig)
templates_all.append(total_template_rig)
t2 = time.time() - t1
print(t2)
pl.imshow(total_template_rig, cmap='gray',
vmax=np.percentile(total_template_rig, 95))
pl.pause(1)
#%%
pl.close()
pl.plot(np.concatenate(total_shifts, 0))
#%% visualize all templates
cm.movie(np.array(templates_all)).play(fr=2, gain=5)
#%% PIECEWISE RIGID MOTION CORRECTION
total_shifts_els = []
templates_all_els = []
new_templ = total_template_rig.copy()
strides = params_movie['strides']
overlaps = params_movie['overlaps']
shifts_opencv = True
save_movie = True
splits = params_movie['splits_els']
num_splits_to_process_list = params_movie['num_splits_to_process_els']
upsample_factor_grid = params_movie['upsample_factor_grid']
max_deviation_rigid = params_movie['upsample_factor_grid']
for file_to_process in all_files:
spatial_filter_, time_trace_, of_or = cm.behavior.behavior.extract_motor_components_OF(m, n_components, mask=mask,
resize_fact=resize_fact, only_magnitude=only_magnitude, verbose=True, method_factorization='dict_learn', max_iter_DL=max_iter_DL)
#%%
mags, dircts, dircts_thresh, spatial_masks_thrs = cm.behavior.behavior.extract_magnitude_and_angle_from_OF(
spatial_filter_, time_trace_, of_or, num_std_mag_for_angle=num_std_mag_for_angle, sav_filter_size=3, only_magnitude=only_magnitude)
#%%
idd = 0
axlin = pl.subplot(n_components, 2, 2)
for mag, dirct, spatial_filter in zip(mags, dircts_thresh, spatial_filter_):
pl.subplot(n_components, 2, 1 + idd * 2)
min_x, min_y = np.min(np.where(mask), 1)
# max_x,max_y = np.max(np.where(mask),1)
spfl = spatial_filter
spfl = cm.movie(spfl[None, :, :]).resize(
1 / resize_fact, 1 / resize_fact, 1).squeeze()
max_x, max_y = np.add((min_x, min_y), np.shape(spfl))
mask[min_x:max_x, min_y:max_y] = spfl
mask[mask < np.nanpercentile(spfl, 70)] = np.nan
# spfl = ld['spatial_filter'][idd]
pl.imshow(m[0], cmap='gray')
pl.imshow(mask, alpha=.5)
pl.axis('off')
axelin = pl.subplot(n_components, 2, 2 + idd * 2, sharex=axlin)
pl.plot(mag / 10, 'k')
dirct[mag < 0.5 * np.std(mag)] = np.nan
pl.plot(dirct, 'r-', linewidth=2)
# gt_pix_2 = scipy.signal.savgol_filter(np.diff(np.array(pts[8]).T,axis=0)[2:],3,1)
def get_behavior_traces(fname,t0,t1,freq,ISI,draw_rois=False,plot_traces=False,mov_filt_1d=True,window_hp=201,window_lp=3,interpolate=True,EXPECTED_ISI=.25):
"""
From hdf5 movies extract eyelid closure and wheel movement
Parameters
----------
fname: str
file name of the hdf5 file
t0,t1: float.
Times of beginning and end of trials (in general 0 and 8 for our dataset) to build the absolute time vector
freq: float
frequency used to build the final time vector
ISI: float
inter stimulu interval
draw_rois: bool
whether to manually draw the eyelid contour
plot_traces: bool
whether to plot the traces during extraction
mov_filt_1d: bool
whether to filter the movie after extracting the average or ROIs. The alternative is a 3D filter that can be very computationally expensive
window_lp, window_hp: ints
number of frames to be used to median filter the data. It is needed because of the light IR artifact coming out of the eye
Returns
-------
res: dict
dictionary with fields
'eyelid': eyelid trace
'wheel': wheel trace
'time': absolute tim vector
'trials': corresponding indexes of the trials
'trial_info': for each trial it returns start trial, end trial, time CS, time US, trial type (CS:0 US:1 CS+US:2)
'idx_CS_US': idx trial CS US
'idx_US': idx trial US
'idx_CS': idx trial CS
"""
CS_ALONE=0
US_ALONE=1
CS_US=2
meta_inf = fname[:-7]+'data.h5'
time_abs=np.linspace(t0,t1,freq*(t1-t0))
T=len(time_abs)
t_us=0
t_cs=0
n_samples_ISI=np.int(ISI*freq)
t_uss=[]
ISIs=[]
eye_traces=[]
wheel_traces=[]
trial_info=[]
tims=[]
with h5py.File(fname) as f:
with h5py.File(meta_inf) as dt:
rois=np.asarray(dt['roi'],np.float32)
trials = list(f.keys())
trials.sort(key=lambda x: np.int(x.replace('trial_','')))
trials_idx=[np.int(x.replace('trial_',''))-1 for x in trials]
trials_idx_=[]
for tr,idx_tr in zip(trials[:],trials_idx[:]):
if plot_traces:
pl.cla()
print(tr)
trial=f[tr]
mov=np.asarray(trial['mov'])
if draw_rois:
pl.imshow(np.mean(mov,0))
pl.xlabel('Draw eye')
pts=pl.ginput(-1)
pts = np.asarray(pts, dtype=np.int32)
data = np.zeros(np.shape(mov)[1:], dtype=np.int32)
# if CV_VERSION == 2:
#lt = cv2.CV_AA
# elif CV_VERSION == 3:
lt = cv2.LINE_AA
cv2.fillConvexPoly(data, pts, (1,1,1), lineType=lt)
rois[0]=data
pl.close()
pl.imshow(np.mean(mov,0))
pl.xlabel('Draw wheel')
pts=pl.ginput(-1)
pts = np.asarray(pts, dtype=np.int32)
data = np.zeros(np.shape(mov)[1:], dtype=np.int32)
# if CV_VERSION == 2:
#lt = cv2.CV_AA
# elif CV_VERSION == 3:
lt = cv2.LINE_AA
cv2.fillConvexPoly(data, pts, (1,1,1), lineType=lt)
rois[1]=data
pl.close()
# eye_trace=np.mean(mov*rois[0],axis=(1,2))
# mov_trace=np.mean((np.diff(np.asarray(mov,dtype=np.float32),axis=0)**2)*rois[1],axis=(1,2))
mov=np.transpose(mov,[0,2,1])
mov=mov[:,:,::-1]
if mov.shape[0]>0:
ts=np.array(trial['ts'])
if np.size(ts)>0:
assert np.std(np.diff(ts))<0.005, 'Time stamps of behaviour are unreliable'
if interpolate:
new_ts=np.linspace(0,ts[-1,0]-ts[0,0],np.shape(mov)[0])
if dt['trials'][idx_tr,-1] == US_ALONE:
t_us=np.maximum(t_us,dt['trials'][idx_tr,3]-dt['trials'][idx_tr,0])
mmm=mov[:n_samples_ISI].copy()
mov=mov[:-n_samples_ISI]
mov=np.concatenate([mmm,mov])
elif dt['trials'][idx_tr,-1] == CS_US:
t_cs=np.maximum(t_cs,dt['trials'][idx_tr,2]-dt['trials'][idx_tr,0])
t_us=np.maximum(t_us,dt['trials'][idx_tr,3]-dt['trials'][idx_tr,0])
t_uss.append(t_us)
ISI=t_us-t_cs
ISIs.append(ISI)
n_samples_ISI=np.int(ISI*freq)
else:
t_cs=np.maximum(t_cs,dt['trials'][idx_tr,2]-dt['trials'][idx_tr,0])
new_ts=new_ts
tims.append(new_ts)
else:
start,end,t_CS,t_US= dt['trials'][idx_tr,:-1]-dt['trials'][idx_tr,0]
f_rate=np.median(np.diff(ts[:,0]))
ISI=t_US-t_CS
idx_US=np.int(old_div(t_US,f_rate))
idx_CS=np.int(old_div(t_CS,f_rate))
fr_before_US=np.int(old_div((t_US - start -.1),f_rate))
fr_after_US=np.int(old_div((end -.1 - t_US),f_rate))
idx_abs=np.arange(-fr_before_US,fr_after_US)
time_abs=idx_abs*f_rate
assert np.abs(ISI-EXPECTED_ISI)<.01, str(np.abs(ISI-EXPECTED_ISI)) + ':the distance form CS and US is different from what expected'
# trig_US=
# new_ts=
mov_e=cb.movie(mov*rois[0][::-1].T,fr=old_div(1,np.mean(np.diff(new_ts))))
mov_w=cb.movie(mov*rois[1][::-1].T,fr=old_div(1,np.mean(np.diff(new_ts))))
x_max_w,y_max_w=np.max(np.nonzero(np.max(mov_w,0)),1)
x_min_w,y_min_w=np.min(np.nonzero(np.max(mov_w,0)),1)
x_max_e,y_max_e=np.max(np.nonzero(np.max(mov_e,0)),1)
x_min_e,y_min_e=np.min(np.nonzero(np.max(mov_e,0)),1)
mov_e=mov_e[:,x_min_e:x_max_e,y_min_e:y_max_e]
mov_w=mov_w[:,x_min_w:x_max_w,y_min_w:y_max_w]
# mpart=mov[:20].copy()
# md=cse.utilities.mode_robust(mpart.flatten())
# N=np.sum(mpart<=md)
# mpart[mpart>md]=md
# mpart[mpart==0]=md
# mpart=mpart-md
# std=np.sqrt(np.sum(mpart**2)/N)
# thr=md+10*std
#
# thr=np.minimum(255,thr)
# return mov
if mov_filt_1d:
mov_e=np.mean(mov_e, axis=(1,2))
window_hp_=window_hp
window_lp_=window_lp
if plot_traces:
pl.plot(old_div((mov_e-np.mean(mov_e)),(np.max(mov_e)-np.min(mov_e))))
else:
window_hp_=(window_hp,1,1)
window_lp_=(window_lp,1,1)
bl=signal.medfilt(mov_e,window_hp_)
mov_e=signal.medfilt(mov_e-bl,window_lp_)
if mov_filt_1d:
eye_=np.atleast_2d(mov_e)
else:
eye_=np.atleast_2d(np.mean(mov_e, axis=(1,2)))
wheel_=np.concatenate([np.atleast_1d(0),np.nanmean(np.diff(mov_w,axis=0)**2,axis=(1,2))])
if np.abs(new_ts[-1] - time_abs[-1])>1:
raise Exception('Time duration is significantly larger or smaller than reference time')
wheel_=np.squeeze(wheel_)
eye_=np.squeeze(eye_)
f1=scipy.interpolate.interp1d(new_ts , eye_,bounds_error=False,kind='linear')
eye_=np.array(f1(time_abs))
f1=scipy.interpolate.interp1d(new_ts , wheel_,bounds_error=False,kind='linear')
wheel_=np.array(f1(time_abs))
if plot_traces:
pl.plot( old_div((eye_), (np.nanmax(eye_)-np.nanmin(eye_))),'r')
pl.plot( old_div((wheel_ -np.nanmin(wheel_)), np.nanmax(wheel_)),'k')
pl.pause(.01)
trials_idx_.append(idx_tr)
eye_traces.append(eye_)
wheel_traces.append(wheel_)
trial_info.append(dt['trials'][idx_tr,:])
res=dict()
res['eyelid'] = eye_traces
res['wheel'] = wheel_traces
res['time'] = time_abs - np.median(t_uss)
res['trials'] = trials_idx_
res['trial_info'] = trial_info
res['idx_CS_US'] = np.where(list(map(int,np.array(trial_info)[:,-1]==CS_US)))[0]
res['idx_US'] = np.where(list(map(int,np.array(trial_info)[:,-1]==US_ALONE)))[0]
res['idx_CS'] = np.where(list(map(int,np.array(trial_info)[:,-1]==CS_ALONE)))[0]
return res
for i, j in zip(x, y):
cv2.rectangle(data, (i * 8, j * 8),
(i * 8 + 64, j * 8 + 64), (0, 0, 255), 5)
return data
annotated = locate(data)
plt.title("Augmented")
plt.imshow(annotated)
plt.show()
#%% visualize_results
num_sampl = 30000
predictions = model.predict(
all_masks_gt[:num_sampl, :, :, None], batch_size=32, verbose=1)
cm.movie(np.squeeze(all_masks_gt[np.where(predictions[:num_sampl, 0] >= 0.95)[
0]])).play(gain=3., magnification=5, fr=10)
#%%
cm.movie(np.squeeze(all_masks_gt[np.where(predictions[:num_sampl, 1] >= 0.95)[
0]])).play(gain=3., magnification=5, fr=10)
#%%
pl.imshow(montage2d(all_masks_gt[np.where((labels_gt[:num_sampl] == 0) & (
predictions[:num_sampl, 1] > 0.95))[0]].squeeze()))
#%%
pl.imshow(montage2d(all_masks_gt[np.where((labels_gt[:num_sampl] == 1) & (
predictions[:num_sampl, 0] > 0.95))[0]].squeeze()))
#%%
pl.imshow(montage2d(all_masks_gt[np.where(
(predictions[:num_sampl, 0] > 0.95))[0]].squeeze()))
#%% retrieve and test
json_file = open(json_path, 'r')
loaded_model_json = json_file.read()
m_res = glob.glob('Sue*.tif')
final_size = (512 - 24, 512 - 24)
winsize = 100
resize_fact_flow = .2
for mv in m_res:
tmpl, correlations, flows_orig, norms, smoothness = compute_metrics_motion_correction(
mv,
final_size[0],
final_size[1],
winsize=winsize,
play_flow=False,
resize_fact_flow=resize_fact_flow)
#%% run comparisons SUITE2P
for mvs in glob.glob('Sue*2000*16*.mat'):
print(mvs)
cm.movie(scipy.io.loadmat(mvs)['data'].transpose(
[2, 0, 1])).save(mvs[:-3] + '.hdf5')
#%%
m_fluos = glob.glob('M_FLUO*.hdf5')
final_size = (64 - 20, 128 - 20)
winsize = 32
resize_fact_flow = 1
for mv in m_fluos:
tmpl, correlations, flows_orig, norms, smoothness = compute_metrics_motion_correction(
mv,
final_size[0],
final_size[1],
winsize=winsize,
play_flow=False,
resize_fact_flow=resize_fact_flow)
#% run comparisons resonant
m_res = glob.glob('Sue_2000*16*.hdf5')
1, 1, downsample_factor), add_to_movie=add_to_movie)
name_new.sort()
fname_new = cm.save_memmap_join(name_new, base_name='Yr', dview=dview)
#%% LOAD MEMORY MAPPABLE FILE
Yr, dims, T = cm.load_memmap(fname_new)
d1, d2 = dims
images = np.reshape(Yr.T, [T] + list(dims), order='F')
Y = np.reshape(Yr, dims + (T,), order='F')
#%% play movie, press q to quit
play_movie = False
if play_movie:
cm.movie(images).play(fr=50, magnification=3, gain=2.)
#%% movie cannot be negative!
if np.min(images) < 0:
raise Exception('Movie too negative, add_to_movie should be larger')
#%% correlation image. From here infer neuron size and density
Cn = cm.movie(images)[:3000].local_correlations(swap_dim=False)
#%% run seeded CNMF
cnm = cnmf.CNMF(n_processes, method_init='greedy_roi', k=Ain.shape[1], gSig=gSig, merge_thresh=merge_thresh,
p=p, dview=dview, Ain=Ain, method_deconvolution='oasis', rolling_sum=False, rf=None)
cnm = cnm.fit(images)
# TODO : needinfo
pl.imshow(new_templ, cmap='gray')
pl.pause(.1)
mc_list.append(mc)
# we are going to keep this part because it helps the user understand what we need.
# needhelp why it is not the same as in the notebooks ?
# TODO: show screenshot 2,3
# %%
# load motion corrected movie
m_rig = cm.load(mc.fname_tot_rig)
pl.imshow(mc.total_template_rig, cmap='gray')
# %% visualize templates
cm.movie(np.array(mc.templates_rig)).play(
fr=5, gain=3, magnification=2, offset=offset_mov)
# %% plot rigid shifts
pl.close()
pl.plot(mc.shifts_rig)
pl.legend(['x shifts', 'y shifts'])
pl.xlabel('frames')
pl.ylabel('pixels')
# %% inspect movie
downsample_ratio = params_display['downsample_ratio']
# TODO: todocument
offset_mov = np.min(m_orig[:100])
m_rig.resize(1, 1, downsample_ratio).play(
gain=3, offset=offset_mov * .5, fr=100, magnification=1, bord_px=bord_px_rig)
# %% restart cluster to clean up memory
# TODO: todocument
pl.xlabel('rigid')
pl.xlim([0, 1])
pl.ylim([0, 1])
#%
pl.subplot(2, 2, 3)
pl.title('rigid mean')
pl.imshow(np.mean(mr, 0), cmap='gray', vmax=300)
pl.axis('off')
pl.subplot(2, 2, 4)
pl.imshow(np.mean(mc, 0), cmap='gray', vmax=300)
pl.title('pw-rigid mean')
pl.axis('off')
#%%
mc = cm.movie(mc)
mc[np.isnan(mc)] = 0
mc.resize(1, 1, .25).play(gain=10., fr=50)
#%%
ccimage = m.local_correlations(eight_neighbours=True, swap_dim=False)
ccimage_rig = mr.local_correlations(eight_neighbours=True, swap_dim=False)
ccimage_els = mc.local_correlations(eight_neighbours=True, swap_dim=False)
#%%
pl.subplot(3, 1, 1)
pl.imshow(ccimage)
pl.subplot(3, 1, 2)
pl.imshow(ccimage_rig)
pl.subplot(3, 1, 3)
pl.imshow(ccimage_els)
#%%
pl.subplot(2, 1, 1)
max_size_neuro = max_radius**2*np.pi
with np.load(gt_file, encoding = 'latin1') as ld:
print(ld.keys())
d1_or = int(ld['d1'])
d2_or = int(ld['d2'])
dims_or = (d1_or,d2_or)
A_gt = ld['A_gt'][()].toarray()
C_gt = ld['C_gt']
Cn_orig = ld['Cn']
#locals().update(ld)
#A_gt = scipy.sparse.coo_matrix(A_gt[()])
#dims = (d1,d2)
if ds_factor > 1:
A_gt = cm.movie(np.reshape(A_gt,dims_or+(-1,),order='F')).transpose(2,0,1).resize(1./ds_factor,1./ds_factor)
if plot_on:
pl.figure(); pl.imshow(A_gt.sum(0))
A_gt2 = np.array(np.reshape(A_gt,(A_gt.shape[0],-1),order='F')).T
Cn_orig = cv2.resize(Cn_orig,None,fx=1./ds_factor,fy=1./ds_factor)
else:
A_gt2 = A_gt.copy()
A_gt_thr = cm.source_extraction.cnmf.spatial.threshold_components(A_gt2, dims, medw=None, thr_method='max', maxthr=global_params['max_thr'], extract_cc=True,
se=None, ss=None, dview=None)
A_gt_thr_bin = A_gt_thr > 0
size_neurons_gt = A_gt_thr_bin.sum(0)
idx_size_neurons_gt = np.where((size_neurons_gt>min_size_neuro) & (size_neurons_gt < max_size_neuro) )[0]
print(A_gt_thr.shape)
#%% filter for size found neurons
strides=strides,
overlaps=overlaps,
splits_els=splits_els,
num_splits_to_process_els=num_splits_to_process_els,
upsample_factor_grid=upsample_factor_grid,
max_deviation_rigid=max_deviation_rigid,
shifts_opencv=True,
nonneg_movie=True)
#%%
mc.motion_correct_rigid(save_movie=True)
#%% load motion corrected movie
m_rig = cm.load(mc.fname_tot_rig)
pl.imshow(mc.total_template_rig, cmap='gray')
#%% visualize templates
cm.movie(np.array(mc.templates_rig)).play(fr=20,
gain=1,
magnification=1,
offset=offset_mov)
#%% plot rigid shifts
pl.close()
pl.plot(mc.shifts_rig)
pl.legend(['x shifts', 'y shifts'])
pl.xlabel('frames')
pl.ylabel('pixels')
#%% inspect movie
bord_px_rig = np.ceil(np.max(mc.shifts_rig)).astype(np.int)
#%%
downsample_ratio = .2
m_rig.resize(1, 1, downsample_ratio).play(gain=2,
offset=offset_mov * .25,
fr=30,
magnification=1,
mags = ld['mags']
dircts_thresh = ld['dircts']
spatial_filter_ = ld['spatial_filter_']
mask_orig = ld['mask']
idd = 0
axlin = pl.subplot(n_components, 2, 2)
for mag, dirct, spatial_filter in zip(mags, dircts_thresh, spatial_filter_):
pl.subplot(n_components, 2, 1 + idd * 2)
m = cm.load(os.path.join(os.path.split(e)[0], 'trial1.mat'))
mask = mask_orig.astype(np.float32).copy()
min_x, min_y = np.min(np.where(mask), 1)
# max_x,max_y = np.max(np.where(mask),1)
spfl = spatial_filter
spfl = cm.movie(spfl[None, :, :]).resize(
1 / resize_fact, 1 / resize_fact, 1).squeeze()
max_x, max_y = np.add((min_x, min_y), np.shape(spfl))
mask[min_x:max_x, min_y:max_y] = spfl
mask[mask < np.nanpercentile(spfl, 70)] = np.nan
# spfl = ld['spatial_filter'][idd]
pl.imshow(m[0], cmap='gray')
pl.imshow(mask, alpha=.5)
pl.axis('off')
axelin = pl.subplot(n_components, 2, 2 + idd * 2, sharex=axlin)
pl.plot(mag, 'k')
dirct[mag < 0.5 * np.std(mag)] = np.nan
pl.plot(dirct, 'r-', linewidth=2)
gt_pix_2 = scipy.signal.savgol_filter(
def run_equalizer(selected_rows, states_df, parameters, session_wise=False):
'''
This function is meant to help with differences in contrast in different trials and session, to equalize general
brightness or reduce photobleaching. It corrects the video and saves them in the corrected version. It can be run
with the already aligned videos or trial by trial. for trial by trial, a template is required.
params: selected_rows: pd.DataFrame -> A dataframe containing the analysis states you want to have equalized
params: states_df: pd.DataFrame -> A dataframe containing all the analysis data base
params: parameters: dict -> contains parameters concerning equalization
returns : None
'''
step_index = 4
# Sort the dataframe correctly
df = selected_rows.sort_values(by=paths.multi_index_structure)
# Determine the output path
output_tif_file_path = os.environ[
'DATA_DIR'] + f'data/interim/equalizer/main/'
mouse, session, init_trial, *r = df.iloc[0].name
#histogram_name = f'mouse_{mouse}_session_{session}_init_trial_{init_trial}'
#output_steps_file_path = f'data/interim/equalizer/meta/figures/histograms/'+histogram_name
try:
df.reset_index()[['session', 'trial', 'is_rest'
]].set_index(['session', 'trial', 'is_rest'],
verify_integrity=True)
except ValueError:
logging.error(
'You passed multiple of the same trial in the dataframe df')
return df
#creates an output dictionary for the data base
output = {
'main': {},
'meta': {
'analysis': {
'analyst': os.environ['ANALYST'],
'date': datetime.datetime.today().strftime("%m-%d-%Y"),
'time': datetime.datetime.today().strftime("%H:%M:%S")
},
'duration': {}
}
}
if session_wise:
row_local = df.iloc[0]
input_tif_file_list = eval(row_local['alignment_output'])['main']
movie_original = cm.load(
input_tif_file_list) # load video as 3d array already concatenated
if parameters['make_template_from_trial'] == 0:
movie_equalized = do_equalization(movie_original)
else:
movie_equalized = np.empty_like(movie_original)
source = movie_original[0:100, :, :]
# equalize all the videos loads in m_list_reshape with the histogram of source
for j in range(int(movie_original.shape[0] / 100)):
want_to_equalize = movie_original[j * 100:(j + 1) * 100, :, :]
movie_equalized[j * 100:(j + 1) *
100, :, :] = do_equalization_from_template(
reference=want_to_equalize, source=source)
#Save the movie
index = row_local.name
new_index = db.replace_at_index1(index, 4 + 4,
2) ## version 2 is for session wise
row_local.name = new_index
equalized_path = movie_equalized.save(
output_tif_file_path + db.create_file_name(4, row_local.name) +
'.mmap',
order='C')
output['main'] = equalized_path
#auxiliar = eval(row_local.loc['alignment_output'])
#auxiliar.update({'equalizing_output' : output})
# row_local.loc['alignment_output'] = str(auxiliar)
row_local.loc['equalization_output'] = output
states_df = db.append_to_or_merge_with_states_df(states_df, row_local)
else:
# Get necessary parameters and create a list with the paths to the relevant files
decoding_output_list = []
input_tif_file_list = []
trial_index_list = []
for idx, row in df.iterrows():
decoding_output = eval(row.loc['decoding_output'])
decoding_output_list.append(decoding_output)
input_tif_file_list.append(decoding_output['main'])
trial_index_list.append(db.get_trial_name(idx[2], idx[3]))
# this was something for ploting while testing, can be removed
#colors = []
#for i in range(len(df)):
# colors.append('#%06X' % randint(0, 0xFFFFFF))
#load the videos as np.array to be able to manipulate them
m_list = []
legend = []
shape_list = []
h_step = parameters['histogram_step']
for i in range(len(input_tif_file_list)):
im = io.imread(input_tif_file_list[i]) #load video as 3d array
m_list.append(im) # and adds all the videos to a list
shape_list.append(
im.shape
) # list of sizes to cut the videos in time for making all of them having the same length
#legend.append('trial = ' + f'{df.iloc[i].name[2]}')
min_shape = min(shape_list)
new_shape = (100 * int(min_shape[0] / 100), min_shape[1], min_shape[2]
) # new videos shape
m_list_reshape = []
m_list_equalized = []
source = m_list[0][0:100, :, :]
#equalize all the videos loaded in m_list_reshape with the histogram of source
for i in range(len(input_tif_file_list)):
video = m_list[i]
if parameters['make_template_from_trial'] == 0:
equalized_video = do_equalization(video)
else:
m_list_reshape.append(video[:new_shape[0], :, :])
equalized_video = np.empty_like(video[:new_shape[0], :, :])
for j in range(int(min_shape[0] / 100)):
want_to_equalize = m_list_reshape[i][j * 100:(j + 1) *
100, :, :]
equalized_video[j * 100:(j + 1) *
100, :, :] = do_equalization_from_template(
reference=want_to_equalize,
source=source)
m_list_equalized.append(equalized_video)
#convert the 3d np.array to a caiman movie and save it as a tif file, so it can be read by motion correction script.
for i in range(len(input_tif_file_list)):
# Save the movie
row_local = df.iloc[i]
movie_original = cm.movie(m_list_reshape[i])
movie_equalized = cm.movie(m_list_equalized[i])
# Write necessary variables to the trial index and row_local
index = row_local.name
new_index = db.replace_at_index1(index, 4 + 0,
1) ## version 1 is for trial wise
row_local.name = new_index
output['main'] = output_tif_file_path + db.create_file_name(
4, row_local.name) + '.tif'
#auxiliar = eval(row_local.loc['decoding_output'])
#auxiliar.update({'equalizing_output' : output})
#row_local.loc['decoding_output'] = str(auxiliar)
row_local.loc['equalization_output'] = output
movie_equalized.save(output_tif_file_path +
db.create_file_name(4, row_local.name) +
'.tif')
states_df = db.append_to_or_merge_with_states_df(
states_df, row_local)
db.save_analysis_states_database(states_df,
paths.analysis_states_database_path,
paths.backup_path)
return
# pl.xlim([0, dim_r[1]])
# pl.ylim([0, dim_r[0]])
pl.pause(.01)
#%% final movie shifts size
np.multiply(shape_grid, 2)
movie_shifts_x = np.zeros(shape_grid + (T,))
movie_shifts_y = np.zeros(shape_grid + (T,))
for idx, r in enumerate(res_p):
x1, x2 = np.unravel_index(idx, shape_grid)
movie_shifts_x[x1, x2, :] = np.array(r[0][-1])[:, 0]
#%%
for idx, r in enumerate(res_p):
mm = cm.movie(r[-1])
mm.play(fr=100, magnification=7, gain=2.)
#%%
#%%
imgtot = np.zeros(np.prod(dim_r))
for idx, r in enumerate(res_p):
# pl.subplot(2,5,idx+1)
img = r[0][2]
# lq,hq = np.percentile(img,[10,90])
# pl.imshow(img,cmap='gray',vmin=lq,vmax=hq)
imgtot[r[1]] = np.maximum(img.T.flatten(), imgtot[r[1]])
m = cm.load('test_mmap_d1_49_d2_114_d3_1_order_C_frames_6764_.mmap')
lq, hq = np.percentile(imgtot, [1, 97])
fn_parts = fn.split('_')
d1 = int(fn_parts[fn_parts.index('d1') + 1]) # column, x
d2 = int(fn_parts[fn_parts.index('d2') + 1]) # row, y
d3 = int(fn_parts[fn_parts.index('d3') + 1]) # channel
d4 = int(fn_parts[fn_parts.index('frames') + 1]) # frame, T
order = fn_parts[fn_parts.index('order') + 1]
mov = np.memmap(filename=os.path.join(data_folder, fn),
shape=(d1, d2, d4),
order=order,
dtype=np.float32,
mode='r')
mov = mov.transpose((2, 1, 0))
print('movie shape: {}'.format(mov.shape))
f = plt.figure(figsize=(8, 5))
ax = f.add_subplot(111)
fig = ax.imshow(np.mean(mov, axis=0),
vmin=300,
vmax=1500,
cmap='inferno',
interpolation='nearest')
f.colorbar(fig)
plt.show()
input("Press enter to continue ...")
print('playing {} ...'.format(fn))
cm.movie(mov).play(fr=30, magnification=1, gain=2.)
m_orig = np.array(images)
patch_size = 50
half_crop = np.minimum(gSig[0] * 4 + 1, patch_size) // 2
idx_included = np.arange(A_gt.shape[-1])
# idx_included = idx_components_cnn[np.arange(50)] # neurons that are displayed
all_pos_crops = []
all_neg_crops = []
all_dubious_crops = []
base_name = fname_new.split('/')[5]
dims = (d1, d2)
idx_included = np.array([x for x in idx_included if x is not None])
base_name = base_name + '_' + \
str(idx_included[0]) + '_' + str(idx_included[-1])
print(base_name)
idx_excluded = np.setdiff1d(np.arange(A_gt.shape[-1]), idx_included)
m_res = m_orig - cm.movie(np.reshape(A_gt.tocsc()[:, idx_excluded].dot(
C_gt[idx_excluded]) + b_gt.dot(f_gt), dims + (-1,), order='F').transpose([2, 0, 1]))
mean_proj = np.mean(m_res, 0)
std_mov = mean_proj.std()
min_mov = np.median(mean_proj)
#%%
# will use either the overfeat like (full FOV, train_net_cifar_edge_cutter_FOV.py) network or the single patch one (train_net_cifar_edge_cutter.py)
full_fov = False
count_start = 30
dims = np.array(dims)
#bin_ = 10
cms_total = [np.array(scipy.ndimage.center_of_mass(np.reshape(a.toarray(
), dims, order='F'))).astype(np.int) for a in A_gt.tocsc()[:, idx_included].T]
cms_total = np.maximum(cms_total, half_crop)
cms_total = np.array([np.minimum(cms, dims - half_crop)
for cms in cms_total]).astype(np.int)
for count in range(count_start, T):
all_results = dict()
ALL_CCs = []
for params_movie in np.array(params_movies)[ID]:
# params_movie['gnb'] = 3
params_display = {'downsample_ratio': .2, 'thr_plot': 0.8}
fname_new = os.path.join(base_folder, params_movie['fname'])
print(fname_new)
# %% LOAD MEMMAP FILE
Yr, dims, T = cm.load_memmap(fname_new)
d1, d2 = dims
images = np.reshape(Yr.T, [T] + list(dims), order='F')
# TODO: needinfo
Y = np.reshape(Yr, dims + (T, ), order='F')
m_images = cm.movie(images)
if plot_on:
if m_images.shape[0] < 5000:
Cn = m_images.local_correlations(
swap_dim=params_movie['swap_dim'], frames_per_chunk=1500)
Cn[np.isnan(Cn)] = 0
else:
Cn = np.array(
cm.load(('/'.join(params_movie['gtname'].split('/')[:-2] +
['projections', 'correlation_image.tif'])
))).squeeze()
check_nan = False
# %% start cluster
# TODO: show screenshot 10
try:
resize_fact_flow = 1
for mv in m_fluos:
tmpl, correlations, flows_orig, norms, smoothness = compute_metrics_motion_correction(
mv, final_size[0], final_size[1], winsize=winsize, play_flow=False, resize_fact_flow=resize_fact_flow)
#% run comparisons resonant
m_res = glob.glob('Sue*.tif')
final_size = (512 - 24, 512 - 24)
winsize = 100
resize_fact_flow = .2
for mv in m_res:
tmpl, correlations, flows_orig, norms, smoothness = compute_metrics_motion_correction(
mv, final_size[0], final_size[1], winsize=winsize, play_flow=False, resize_fact_flow=resize_fact_flow)
#%% run comparisons SUITE2P
for mvs in glob.glob('Sue*2000*16*.mat'):
print(mvs)
cm.movie(scipy.io.loadmat(mvs)['data'].transpose(
[2, 0, 1])).save(mvs[:-3] + '.hdf5')
#%%
m_fluos = glob.glob('M_FLUO*.hdf5')
final_size = (64 - 20, 128 - 20)
winsize = 32
resize_fact_flow = 1
for mv in m_fluos:
tmpl, correlations, flows_orig, norms, smoothness = compute_metrics_motion_correction(
mv, final_size[0], final_size[1], winsize=winsize, play_flow=False, resize_fact_flow=resize_fact_flow)
#% run comparisons resonant
m_res = glob.glob('Sue_2000*16*.hdf5')
final_size = (512 - 24, 512 - 24)
winsize = 100
resize_fact_flow = .2
for mv in m_res:
tmpl, correlations, flows_orig, norms, smoothness = compute_metrics_motion_correction(
cm.utils.visualization.view_patches_bar(Yr,
cnm.A[:, idx_components],
cnm.C[idx_components],
cnm.b,
cnm.f,
dims[0],
dims[1],
YrA=cnm.YrA[idx_components],
img=cn_filter)
#%%
cm.stop_server(dview=dview)
#%% denoised movie
cm.movie(
np.reshape(cnm.A.tocsc()[:, idx_components].dot(cnm.C[idx_components]) +
cnm.b.dot(cnm.f),
dims + (-1, ),
order='F').transpose(2, 0, 1)).play(magnification=3, gain=1.)
#%% only neurons
cm.movie(
np.reshape(cnm.A.tocsc()[:, idx_components].dot(cnm.C[idx_components]),
dims + (-1, ),
order='F').transpose(2, 0, 1)).play(magnification=3, gain=10.)
#%% only the background
cm.movie(
np.reshape(cnm.b.dot(cnm.f), dims + (-1, ),
order='F').transpose(2, 0, 1)).play(magnification=3, gain=1.)
#%% residuals
cm.movie(
np.array(Y) -
np.reshape(cnm.A.tocsc()[:, :].dot(cnm.C[:]) + cnm.b.dot(cnm.f),
#%% VIEW TRACES (accepted and rejected)
view_patches_bar(Yr, cnm2.A.tocsc()[:, idx_components], cnm2.C[idx_components],
cnm2.b, cnm2.f, dims[0], dims[1], YrA=cnm2.YrA[idx_components],
img=Cn)
view_patches_bar(Yr, cnm2.A.tocsc()[:, idx_components_bad], cnm2.C[idx_components_bad],
cnm2.b, cnm2.f, dims[0], dims[1], YrA=cnm2.YrA[idx_components_bad],
img=Cn)
#%% Extract DF/F values
F_dff = detrend_df_f(cnm2.A, cnm2.b, cnm2.C, cnm2.f, YrA=cnm2.YrA,
quantileMin=8, frames_window=250)
#%% Show final traces
cnm2.view_patches(Yr, dims=dims, img=Cn)
#%% STOP CLUSTER and clean up log files
cm.stop_server(dview=dview)
log_files = glob.glob('*_LOG_*')
for log_file in log_files:
os.remove(log_file)
#%% reconstruct denoised movie
denoised = cm.movie(cnm2.A.dot(cnm2.C) +
cnm2.b.dot(cnm2.f)).reshape(dims + (-1,), order='F').transpose([2, 0, 1])
#%% play along side original data
cm.concatenate([m_els.resize(1, 1, downsample_ratio),
denoised.resize(1, 1, downsample_ratio)],
axis=2).play(fr=60, gain=15, magnification=2, offset=0) # press q to exit
bord_px_rig = np.ceil(np.max(mc.shifts_rig)).astype(np.int)
# TODO : needinfo
pl.imshow(new_templ, cmap='gray')
pl.pause(.1)
mc_list.append(mc)
# we are going to keep this part because it helps the user understand what we need.
# needhelp why it is not the same as in the notebooks ?
# TODO: show screenshot 2,3
# %%
# load motion corrected movie
m_rig = cm.load(mc.fname_tot_rig)
pl.imshow(mc.total_template_rig, cmap='gray')
# %% visualize templates
cm.movie(np.array(mc.templates_rig)).play(
fr=10, gain=5, magnification=2, offset=offset_mov)
# %% plot rigid shifts
pl.close()
pl.plot(mc.shifts_rig)
pl.legend(['x shifts', 'y shifts'])
pl.xlabel('frames')
pl.ylabel('pixels')
# %% inspect movie
downsample_ratio = params_display['downsample_ratio']
# TODO: todocument
offset_mov = -np.min(m_orig[:100])
m_rig.resize(1, 1, downsample_ratio).play(
gain=10, offset=offset_mov * .25, fr=30, magnification=2, bord_px=bord_px_rig)
# %%
# a computing intensive but parralellized part
t1 = time.time()
#mc = cm.load('M_FLUO_4_d1_64_d2_128_d3_1_order_F_frames_4620_.mmap')
#mc = cm.load('M_FLUO_t_d1_64_d2_128_d3_1_order_F_frames_6764_.mmap')
#%%
mc.resize(1,1,.2).play(gain=30,fr = 30, offset = 300,magnification=1.)
#%%
m.resize(1,1,.2).play(gain=10,fr = 30, offset = 0,magnification=1.)
#%%
cm.concatenate([mr.resize(1,1,.5),mc.resize(1,1,.5)],axis=1).play(gain=10,fr = 100, offset = 300,magnification=1.)
#%%
import h5py
with h5py.File('sueann_pw_rigid_movie.mat') as f:
mef = np.array(f['M2'])
mef = cm.movie(mef.transpose([0,2,1]))
#%%
cm.concatenate([mef.resize(1,1,.15),mc.resize(1,1,.15)],axis=1).play(gain=30,fr = 40, offset = 300,magnification=1.)
#%%
(mef-mc).resize(1,1,.1).play(gain=50,fr = 20, offset = 0,magnification=1.)
#%%
(mc-mef).resize(1,1,.1).play(gain=50,fr = 20, offset = 0,magnification=1.)
#%%
T,d1,d2 = np.shape(m)
shape_mov = (d1*d2,m.shape[0])
Y = np.memmap('M_FLUO_4_d1_64_d2_128_d3_1_order_F_frames_4620_.mmap',mode = 'r',dtype=np.float32, shape=shape_mov, order='F')
mc = cm.movie(np.reshape(Y,(d2,d1,T),order = 'F').transpose([2,1,0]))
mc.resize(1,1,.25).play(gain=10.,fr=50)
#%%