ROI Extractors

Python-based module for extracting from, converting between, and handling recorded and optical imaging data from several file formats. Inspired by SpikeExtractors.

Developed by CatalystNeuro.

Getting Started:

Installation:

pip install roiextractors

Usage:

Supported file types:

Imaging

1. HDF5
2. TIFF
3. STK
4. FLI
5. SBX

Segmentation

1. calciumImagingAnalysis (CNMF-E, EXTRACT)
2. SIMA
3. NWB
4. suite2p
5. Numpy (a data format for manual input of optical physiology data as various numpy datasets)

Functionality:

Interconversion amongst the various data formats as well as conversion to the NWB format and back.

Features:

1. SegmentationExtractor object:
   • seg_obj.get_channel_names() : List of optical channel names
   • seg_obj.get_num_channels() : Number of channels
   • seg_obj.get_movie_framesize(): (height, width) of raw movie
   • seg_obj.get_movie_location(): Location of storage of movie/tiff images
   • seg_obj.get_image_masks(self, roi_ids=None): Image masks as (ht, wd, num_rois) with each value as the weight given during segmentation operation.
   • seg_obj.get_pixel_masks(roi_ids=None): Get pixel masks as (total_pixels(ht*wid), no_rois)
   • seg_obj.get_traces(self, roi_ids=None, start_frame=None, end_frame=None): df/F trace as (num_rois, num_frames)
   • seg_obj.get_sampling_frequency(): Sampling frequency of movie/df/F trace.
   • seg_obj.get_roi_locations(): Centroid pixel location of the ROI (Regions Of Interest) as (x,y).
   • seg_obj.get_num_rois(): Total number of ROIs after segmentation operation.
   • seg_obj.get_roi_ids(): Any integer tags associated with an ROI, defaults to 0:num_of_rois

SegmentationExtractor object creation:

import roiextractors
import numpy as np

seg_obj_cnmfe = roiextractors.CnmfeSegmentationExtractor('cnmfe_filename.mat') # cnmfe
seg_obj_extract = roiextractors.ExtractSegmentationExtractor('extract_filename.mat') # extract
seg_obj_sima = roiextractors.SimaSegmentationExtractor('sima_filename.sima') # SIMA
seg_obj_numpy = roiextractors.NumpySegmentationExtractor(
                    filepath = 'path-to-file',
                    masks=np.random.rand(movie_size[0],movie_size[1],no_rois),
                    signal=np.random.randn(num_rois,num_frames),
                    roi_idx=np.random.randint(no_rois,size=[1,no_rois]),
                    no_of_channels=None,
                    summary_image=None,
                    channel_names=['Blue']) # Numpy object
seg_obj_nwb = roiextractors.NwbSegmentationExtractor(
                    filepath_of_nwb, optical_channel_name=None, # optical channel to extract and store info from
                    imaging_plane_name=None, image_series_name=None, # imaging plane to extract and store data from
                    processing_module_name=None,
                    neuron_roi_response_series_name=None, # roi_response_series name to extract and store data from
                    background_roi_response_series_name=None) # nwb object

Data format conversion: SegmentationExtractor to NWB:

roiextractors.NwbSegmentationExtractor.write_segmentation(seg_obj, saveloc,
                    propertydict=[{'name': 'ROI feature 1,
                                   'description': 'additional attribute of each ROI',
                                   'data': np.random.rand(1,no_rois),
                                   'id': seg_obj.get_roi_ids()},
                                  {'name': 'ROI feature 2,
                                   'description': 'additional attribute of each ROI',
                                   'data': np.random.rand(1,no_rois),
                                   'id': seg_obj.get_roi_ids()}],
                    nwbfile_kwargs={'session_description': 'nwbfiledesc',
                                    'experimenter': 'experimenter name',
                                    'lab': 'test lab',
                                    'session_id': 'test sess id'},
                    emission_lambda=400.0, excitation_lambda=500.0)

Example Datasets:

• Example datasets for each of the file formats can be downloaded here.

Class descriptions:

• SegmentationExtractor: An abstract class that contains all the meta-data and output data from the ROI segmentation operation when applied to the pre-processed data. It also contains methods to read from and write to various data formats ouput from the processing pipelines like SIMA, CaImAn, Suite2p, CNNM-E.

• NumpySegmentationExtractor: Contains all data coming from a file format for which there is currently no support. To construct this, all data must be entered manually as arguments.

• CnmfeSegmentationExtractor: This class inherits from the SegmentationExtractor class, having all its funtionality specifically applied to the dataset output from the 'CNMF-E' ROI segmentation method.

• ExtractSegmentationExtractor: This class inherits from the SegmentationExtractor class, having all its funtionality specifically applied to the dataset output from the 'EXTRACT' ROI segmentation method.

• SimaSegmentationExtractor: This class inherits from the SegmentationExtractor class, having all its funtionality specifically applied to the dataset output from the 'SIMA' ROI segmentation method.

• NwbSegmentationExtractor: Extracts data from the NWB data format. Also implements a static method to write any format specific object to NWB.

• Suite2PSegmentationExtractor: Extracts data from suite2p format.

Troubleshooting

Installing SIMA with python>=3.7:

Will need a manual installation for package dependency SIMA since it does not currently support python 3.7:

1. Download SIMA wheels distribution here.
2. pip install <download-path-to-wheels.whl>
3. pip install roiextractors

Funded by

• Stanford University as part of the Ripple U19 project (U19NS104590).
• LBNL as part of the NWB U24 (U24NS120057).