The mesmerize-napari from nel-lab

Have ROIs and Traces accessing same cmap

analogous ROIs and traces should have same colour

TODO: Offline Analysis

Nice to have's:

Show reconstructed movie (denoised, deconvolved etc.), try to compute it on the fly @kushalkolar
View raw and mcorr movie side by side
View windowed mean mcorr video, button for this in the viz tab in the main GUI
- np.mean(movie[i - 2: i + 2], axis=0)
Visualization tab in the main GUI
Draw a square ROI and display traces within this square
In addition to current CNMF GUI, a separate window that shows stitched spatial components, one component per frame, as user moves sliders show the temporal component.
Viz tab, option to view ROIs as a mask or outline. Outline doesn't work when ROIs look really bad.
Dropdown menu to select which movie to use as input file path for param selection @kushalkolar

Eval GUI is broken

Eval GUI appears to be broken? Tested with the Sue demo movie.

@ArjunPutcha have you tried the eval gui recently?

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
File ~/Insync/kushalkolar@gmail.com/drive/repos/mesmerize-napari/mesmerize_napari/evaluate_components.py:49, in EvalComponentsWidgets.update_components(self=)
     46 cnmf_obj: CNMF = self.cnmf_viewer.cnmf_obj
     47 cnmf_obj.params.quality.update(params)
---> 49 cnmf_obj.estimates.filter_components(
        cnmf_obj = 
        cnmf_obj.estimates = 
        self.cnmf_viewer = 
        self = 
        self.cnmf_viewer.batch_item.cnmf = 
        self.cnmf_viewer.batch_item = algo                                                             cnmf
name                                                        test-cnmf
input_movie_path    a89c5872-4e50-44e4-8735-6dce4c6f2411-mcorr_els...
params              {'cnmf_kwargs': {'p': 2, 'nb': 1, 'rf': None, ...
outputs             {'mean-projection-path': 5f7a4906-9228-4868-8a...
uuid                             5f7a4906-9228-4868-8a82-c449071eb15d
Name: 1, dtype: object
        cnmf_obj.params = CNMFParams:

data:

{'caiman_version': '1.9.10',
 'decay_time': 1,
 'dims': b'NoneType',
 'dxy': (1, 1),
 'fnames': b'NoneType',
 'fr': 30,
 'last_commit': 'GITW-ae589c04ca7f8bead391b2dd14573c8cc9b3015e',
 'mmap_C': b'NoneType',
 'mmap_F': b'NoneType',
 'var_name_hdf5': 'mov'}

spatial_params:

{'block_size_spat': 5000,
 'dist': 3,
 'expandCore':  (5, 5) int64,
 'extract_cc': True,
 'maxthr': 0.1,
 'medw': b'NoneType',
 'method_exp': 'dilate',
 'method_ls': 'lasso_lars',
 'n_pixels_per_process': 208,
 'nb': 1,
 'normalize_yyt_one': True,
 'nrgthr': 0.9999,
 'num_blocks_per_run_spat': 20,
 'se': b'NoneType',
 'ss': b'NoneType',
 'thr_method': 'nrg',
 'update_background_components': True}

temporal_params:

{'ITER': 2,
 'bas_nonneg': False,
 'block_size_temp': 5000,
 'fudge_factor': 0.96,
 'lags': 5,
 'memory_efficient': False,
 'method_deconvolution': 'oasis',
 'nb': 1,
 'noise_method': 'mean',
 'noise_range':  (2,) float64,
 'num_blocks_per_run_temp': 20,
 'optimize_g': False,
 'p': 2,
 's_min': b'NoneType',
 'solvers':  (2,) |S4,
 'verbosity': False}

init_params:

{'K': 10,
 'SC_kernel': 'heat',
 'SC_nnn': 20,
 'SC_normalize': True,
 'SC_sigma': 1,
 'SC_thr': 0,
 'SC_use_NN': False,
 'alpha_snmf': 100,
 'center_psf': False,
 'gSig':  (2,) int64,
 'gSiz':  (2,) int64,
 'init_iter': 2,
 'kernel': b'NoneType',
 'lambda_gnmf': 1,
 'maxIter': 5,
 'max_iter_snmf': 500,
 'method_init': 'greedy_roi',
 'min_corr': 0.85,
 'min_pnr': 20,
 'nIter': 5,
 'nb': 1,
 'normalize_init': True,
 'options_local_NMF': b'NoneType',
 'perc_baseline_snmf': 20,
 'ring_size_factor': 1.5,
 'rolling_length': 100,
 'rolling_sum': True,
 'seed_method': 'auto',
 'sigma_smooth_snmf': (0.5, 0.5, 0.5),
 'ssub': 1,
 'ssub_B': 2,
 'tsub': 1}

preprocess_params:

{'check_nan': True,
 'compute_g': False,
 'include_noise': False,
 'lags': 5,
 'max_num_samples_fft': 3072,
 'n_pixels_per_process': 208,
 'noise_method': 'mean',
 'noise_range':  (2,) float64,
 'p': 2,
 'pixels': b'NoneType',
 'sn': b'NoneType'}

patch_params:

{'border_pix': 0,
 'del_duplicates': False,
 'in_memory': True,
 'low_rank_background': True,
 'memory_fact': 1,
 'n_processes': 23,
 'nb_patch': 1,
 'only_init': False,
 'p_patch': 0,
 'p_ssub': 2,
 'p_tsub': 2,
 'remove_very_bad_comps': False,
 'rf': b'NoneType',
 'skip_refinement': False,
 'stride': 30}

online:

{'N_samples_exceptionality': 30,
 'W_update_factor': 1,
 'batch_update_suff_stat': False,
 'dist_shape_update': False,
 'ds_factor': 1,
 'epochs': 1,
 'expected_comps': 500,
 'full_XXt': False,
 'init_batch': 2000,
 'init_method': 'bare',
 'iters_shape': 5,
 'max_comp_update_shape': inf,
 'max_num_added': 5,
 'max_shifts_online': 10,
 'min_SNR': 2.5,
 'min_num_trial': 5,
 'minibatch_shape': 100,
 'minibatch_suff_stat': 5,
 'motion_correct': True,
 'movie_name_online': 'online_movie.mp4',
 'n_refit': 0,
 'normalize': False,
 'num_times_comp_updated': inf,
 'opencv_codec': 'H264',
 'path_to_model': '/home/kushal/caiman_data/model/cnn_model_online.h5',
 'ring_CNN': False,
 'rval_thr': 0.8,
 'save_online_movie': False,
 'show_movie': False,
 'simultaneously': False,
 'sniper_mode': False,
 'stop_detection': False,
 'test_both': False,
 'thresh_CNN_noisy': 0.5,
 'thresh_fitness_delta': -50,
 'thresh_fitness_raw': -152.44944831836074,
 'thresh_overlap': 0.5,
 'update_freq': 200,
 'update_num_comps': True,
 'use_corr_img': False,
 'use_dense': True,
 'use_peak_max': True}

quality:

{'SNR_lowest': 0.0,
 'cnn_lowest': 0.0,
 'gSig_range': b'NoneType',
 'max_ecc': 3,
 'min_SNR': 0.5,
 'min_cnn_thr': 0.5,
 'rval_lowest': 0.2000000000000001,
 'rval_thr': 0.0,
 'use_cnn': True,
 'use_ecc': False}

merging:

{'do_merge': True,
 'max_merge_area': b'NoneType',
 'merge_parallel': False,
 'merge_thr': 0.8}

motion:

{'border_nan': 'copy',
 'gSig_filt': b'NoneType',
 'indices':  (2,) |S32,
 'is3D': False,
 'max_deviation_rigid': 3,
 'max_shifts': (6, 6),
 'min_mov': b'NoneType',
 'niter_rig': 1,
 'nonneg_movie': True,
 'num_frames_split': 80,
 'num_splits_to_process_els': b'NoneType',
 'num_splits_to_process_rig': b'NoneType',
 'overlaps': (32, 32),
 'pw_rigid': False,
 'shifts_opencv': True,
 'splits_els': 14,
 'splits_rig': 14,
 'strides': (96, 96),
 'upsample_factor_grid': 4,
 'use_cuda': False}

ring_CNN:

{'loss_fn': 'pct',
 'lr': 0.001,
 'lr_scheduler': b'NoneType',
 'max_epochs': 100,
 'n_channels': 2,
 'path_to_model': b'NoneType',
 'patience': 3,
 'pct': 0.01,
 'remove_activity': False,
 'reuse_model': False,
 'use_add': False,
 'use_bias': False,
 'width': 5}
     50     imgs=self.cnmf_viewer.batch_item.cnmf.get_input_memmap(),
     51     params=cnmf_obj.params,
     52 )
     54 self.cnmf_viewer.update_visible_components()

File ~/Insync/kushalkolar@gmail.com/drive/repos/CaImAn/caiman/source_extraction/cnmf/estimates.py:1142, in Estimates.filter_components(self=, imgs= (2000, 60, 80) float32, params=CNMFParams:

data:

{'caiman_version': '1.9.10',...use_add': False,
 'use_bias': False,
 'width': 5}, new_dict={}, dview=None, select_mode='All')
   1139     self.evaluate_components(imgs, params, dview=dview)
   1140 else:
   1141     self.idx_components, self.idx_components_bad, self.cnn_preds = \
-> 1142     select_components_from_metrics(self.A, dims, params.get('init', 'gSig'),
        self.idx_components_bad =  (1,) int64
        self = 
        self.idx_components, self.idx_components_bad, self.cnn_preds = ( (9,) int64,  (1,) int64,  (10,) float32)
        self.idx_components =  (9,) int64
        self.cnn_preds =  (10,) float32
        dims = (60, 80)
        params = CNMFParams:

data:

{'caiman_version': '1.9.10',
 'decay_time': 1,
 'dims': b'NoneType',
 'dxy': (1, 1),
 'fnames': b'NoneType',
 'fr': 30,
 'last_commit': 'GITW-ae589c04ca7f8bead391b2dd14573c8cc9b3015e',
 'mmap_C': b'NoneType',
 'mmap_F': b'NoneType',
 'var_name_hdf5': 'mov'}

spatial_params:

{'block_size_spat': 5000,
 'dist': 3,
 'expandCore':  (5, 5) int64,
 'extract_cc': True,
 'maxthr': 0.1,
 'medw': b'NoneType',
 'method_exp': 'dilate',
 'method_ls': 'lasso_lars',
 'n_pixels_per_process': 208,
 'nb': 1,
 'normalize_yyt_one': True,
 'nrgthr': 0.9999,
 'num_blocks_per_run_spat': 20,
 'se': b'NoneType',
 'ss': b'NoneType',
 'thr_method': 'nrg',
 'update_background_components': True}

temporal_params:

{'ITER': 2,
 'bas_nonneg': False,
 'block_size_temp': 5000,
 'fudge_factor': 0.96,
 'lags': 5,
 'memory_efficient': False,
 'method_deconvolution': 'oasis',
 'nb': 1,
 'noise_method': 'mean',
 'noise_range':  (2,) float64,
 'num_blocks_per_run_temp': 20,
 'optimize_g': False,
 'p': 2,
 's_min': b'NoneType',
 'solvers':  (2,) |S4,
 'verbosity': False}

init_params:

{'K': 10,
 'SC_kernel': 'heat',
 'SC_nnn': 20,
 'SC_normalize': True,
 'SC_sigma': 1,
 'SC_thr': 0,
 'SC_use_NN': False,
 'alpha_snmf': 100,
 'center_psf': False,
 'gSig':  (2,) int64,
 'gSiz':  (2,) int64,
 'init_iter': 2,
 'kernel': b'NoneType',
 'lambda_gnmf': 1,
 'maxIter': 5,
 'max_iter_snmf': 500,
 'method_init': 'greedy_roi',
 'min_corr': 0.85,
 'min_pnr': 20,
 'nIter': 5,
 'nb': 1,
 'normalize_init': True,
 'options_local_NMF': b'NoneType',
 'perc_baseline_snmf': 20,
 'ring_size_factor': 1.5,
 'rolling_length': 100,
 'rolling_sum': True,
 'seed_method': 'auto',
 'sigma_smooth_snmf': (0.5, 0.5, 0.5),
 'ssub': 1,
 'ssub_B': 2,
 'tsub': 1}

preprocess_params:

{'check_nan': True,
 'compute_g': False,
 'include_noise': False,
 'lags': 5,
 'max_num_samples_fft': 3072,
 'n_pixels_per_process': 208,
 'noise_method': 'mean',
 'noise_range':  (2,) float64,
 'p': 2,
 'pixels': b'NoneType',
 'sn': b'NoneType'}

patch_params:

{'border_pix': 0,
 'del_duplicates': False,
 'in_memory': True,
 'low_rank_background': True,
 'memory_fact': 1,
 'n_processes': 23,
 'nb_patch': 1,
 'only_init': False,
 'p_patch': 0,
 'p_ssub': 2,
 'p_tsub': 2,
 'remove_very_bad_comps': False,
 'rf': b'NoneType',
 'skip_refinement': False,
 'stride': 30}

online:

{'N_samples_exceptionality': 30,
 'W_update_factor': 1,
 'batch_update_suff_stat': False,
 'dist_shape_update': False,
 'ds_factor': 1,
 'epochs': 1,
 'expected_comps': 500,
 'full_XXt': False,
 'init_batch': 2000,
 'init_method': 'bare',
 'iters_shape': 5,
 'max_comp_update_shape': inf,
 'max_num_added': 5,
 'max_shifts_online': 10,
 'min_SNR': 2.5,
 'min_num_trial': 5,
 'minibatch_shape': 100,
 'minibatch_suff_stat': 5,
 'motion_correct': True,
 'movie_name_online': 'online_movie.mp4',
 'n_refit': 0,
 'normalize': False,
 'num_times_comp_updated': inf,
 'opencv_codec': 'H264',
 'path_to_model': '/home/kushal/caiman_data/model/cnn_model_online.h5',
 'ring_CNN': False,
 'rval_thr': 0.8,
 'save_online_movie': False,
 'show_movie': False,
 'simultaneously': False,
 'sniper_mode': False,
 'stop_detection': False,
 'test_both': False,
 'thresh_CNN_noisy': 0.5,
 'thresh_fitness_delta': -50,
 'thresh_fitness_raw': -152.44944831836074,
 'thresh_overlap': 0.5,
 'update_freq': 200,
 'update_num_comps': True,
 'use_corr_img': False,
 'use_dense': True,
 'use_peak_max': True}

quality:

{'SNR_lowest': 0.0,
 'cnn_lowest': 0.0,
 'gSig_range': b'NoneType',
 'max_ecc': 3,
 'min_SNR': 0.5,
 'min_cnn_thr': 0.5,
 'rval_lowest': 0.2000000000000001,
 'rval_thr': 0.0,
 'use_cnn': True,
 'use_ecc': False}

merging:

{'do_merge': True,
 'max_merge_area': b'NoneType',
 'merge_parallel': False,
 'merge_thr': 0.8}

motion:

{'border_nan': 'copy',
 'gSig_filt': b'NoneType',
 'indices':  (2,) |S32,
 'is3D': False,
 'max_deviation_rigid': 3,
 'max_shifts': (6, 6),
 'min_mov': b'NoneType',
 'niter_rig': 1,
 'nonneg_movie': True,
 'num_frames_split': 80,
 'num_splits_to_process_els': b'NoneType',
 'num_splits_to_process_rig': b'NoneType',
 'overlaps': (32, 32),
 'pw_rigid': False,
 'shifts_opencv': True,
 'splits_els': 14,
 'splits_rig': 14,
 'strides': (96, 96),
 'upsample_factor_grid': 4,
 'use_cuda': False}

ring_CNN:

{'loss_fn': 'pct',
 'lr': 0.001,
 'lr_scheduler': b'NoneType',
 'max_epochs': 100,
 'n_channels': 2,
 'path_to_model': b'NoneType',
 'patience': 3,
 'pct': 0.01,
 'remove_activity': False,
 'reuse_model': False,
 'use_add': False,
 'use_bias': False,
 'width': 5}
        self.A = '
with 1979 stored elements in Compressed Sparse Column format>
        self.r_values =  (10,) float64
        self.SNR_comp =  (10,) float64
        opts = {'SNR_lowest': 0.0, 'cnn_lowest': 0.0, 'gSig_range': b'NoneType', 'min_SNR': 0.5, 'min_cnn_thr': 0.5, 'rval_lowest': 0.2000000000000001, 'rval_thr': 0.0, 'use_cnn': True, 'use_ecc': False, 'max_ecc': 3}
        opts['rval_thr'] = 0.0
        opts['rval_lowest'] = 0.2000000000000001
        opts['min_SNR'] = 0.5
        opts['SNR_lowest'] = 0.0
        opts['min_cnn_thr'] = 0.5
        opts['cnn_lowest'] = 0.0
        opts['use_cnn'] = True
        opts['gSig_range'] = b'NoneType'
   1143                                    self.r_values, self.SNR_comp,
   1144                                    predictions=self.cnn_preds,
   1145                                    r_values_min=opts['rval_thr'],
   1146                                    r_values_lowest=opts['rval_lowest'],
   1147                                    min_SNR=opts['min_SNR'],
   1148                                    min_SNR_reject=opts['SNR_lowest'],
   1149                                    thresh_cnn_min=opts['min_cnn_thr'],
   1150                                    thresh_cnn_lowest=opts['cnn_lowest'],
   1151                                    use_cnn=opts['use_cnn'],
   1152                                    gSig_range=opts['gSig_range'])
   1153     if opts['use_ecc']:
   1154         idx_ecc = np.where(self.ecc  opts['max_ecc'])[0]

File ~/Insync/kushalkolar@gmail.com/drive/repos/CaImAn/caiman/components_evaluation.py:673, in select_components_from_metrics(A=, dims=(60, 80), gSig= (2,) int64, r_values= (10,) float64, comp_SNR= (10,) float64, r_values_min=0.0, r_values_lowest=0.2000000000000001, min_SNR=0.5, min_SNR_reject=0.0, thresh_cnn_min=0.5, thresh_cnn_lowest=0.0, use_cnn=True, gSig_range=b'NoneType', neuron_class=1, predictions= (10,) float64, **kwargs={})
    671     predictions = np.zeros(len(r_values))
    672     for size_range in gSig_range:
--> 673         predictions = np.maximum(predictions, evaluate_components_CNN(A, dims, size_range)[0][:, neuron_class])
        predictions =  (10,) float64
        size_range = 78
        np = 
        A = '
with 1979 stored elements in Compressed Sparse Column format>
        dims = (60, 80)
        np.maximum = 
        neuron_class = 1
    675 idx_components_cnn = np.where(predictions >= thresh_cnn_min)[0]
    676 bad_comps = np.where((r_values = r_values_lowest) | (comp_SNR = min_SNR_reject) |
    677                      (predictions = thresh_cnn_lowest))[0]

File ~/Insync/kushalkolar@gmail.com/drive/repos/CaImAn/caiman/components_evaluation.py:321, in evaluate_components_CNN(A=, dims=(60, 80), gSig=78, model_name='/home/kushal/caiman_data/model/cnn_model', patch_size=50, loaded_model=, isGPU=False)
    317         loaded_model = load_graph(model_file)
    319     logging.debug("Loaded model from disk")
--> 321 half_crop = np.minimum(gSig[0] * 4 + 1, patch_size), np.minimum(gSig[1] * 4 + 1, patch_size)
        np.minimum = 
        patch_size = 50
        np = 
        gSig = 78
    322 dims = np.array(dims)
    323 coms = [scipy.ndimage.center_of_mass(mm.toarray().reshape(dims, order='F')) for mm in A.tocsc().T]

TypeError: 'int' object is not subscriptable

sub-GUIs should not show re-instantiate every time a button is clicked

Basically change these:

def show_cnmf_params_gui(self):
    self.cnmf_gui = CNMFWidget(parent=self)
    self.cnmf_gui.show()

to this:

def show_cnmf_params_gui(self):
    if hasattr(self, 'cnmf_gui'):
        self.cnmf_gui.show()
        return

    self.cnmf_gui = CNMFWidget(parent=self)
    self.cnmf_gui.show()

Use opencv to draw shapes

Check contour selection and colour change works well

[BUG] Abort button needs to be functional

The Abort button needs to be made functional and the Start and Start at item buttons needs to be disabled once a batch is running. It is currently possible to flood the system with processes by clicking the Start button repeatedly.

Eval GUI should default to current eval kwargs

When launching eval gui, the values for each item (SNR, rval, etc.) should by default be whatever the predefined eval kwargs were.

[FEATURE] set framerate in CNMF & CNMFE param GUIs

add a doubleSpinBox to CNMF and CNMFE param GUIs to input fr param.

Last things for param GUIs

Input movie selection by using QComboBox to select input movie file corresponding to a layer in the viewer
framerate QDoubleSpinBox

[BUG] strange QProcess launching

Sometimes QProcess does not launch with the first click of Start but only with the second click, weird maybe a napari thing. Isolate by launching purely from backend pandas exts.

docs to do

Written tutorial
- mcorr -> cnmf(e) -> eval
- Start off with mesmerize-core in jupyter for batch creation, napari just for viz as an option. Params GUIs also possible but jupyter is recommended
Video tutorial, same stuff as written tut

nel-lab / mesmerize-napari Goto Github PK

mesmerize-napari's People

Stargazers

Watchers

mesmerize-napari's Issues

Recommend Projects

Recommend Topics

Recommend Org