-
Notifications
You must be signed in to change notification settings - Fork 0
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add miniscope interface #8
Changes from all commits
705e74e
31f8ef5
262a5ab
6196ac4
ae83880
File filter
Filter by extension
Conversations
Jump to
Diff view
Diff view
There are no files selected for viewing
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,316 @@ | ||
from roiextractors.imagingextractor import ImagingExtractor | ||
from roiextractors.multiimagingextractor import MultiImagingExtractor | ||
from roiextractors.extraction_tools import PathType, DtypeType, get_package | ||
|
||
from typing import Optional | ||
import json | ||
import datetime | ||
|
||
from pydantic import DirectoryPath | ||
from pathlib import Path | ||
import numpy as np | ||
|
||
from copy import deepcopy | ||
from pathlib import Path | ||
from typing import Literal, Optional | ||
|
||
import numpy as np | ||
from pydantic import DirectoryPath, validate_call | ||
from pynwb import NWBFile | ||
|
||
from neuroconv.datainterfaces.ophys.baseimagingextractorinterface import BaseImagingExtractorInterface | ||
from neuroconv.utils import DeepDict, dict_deep_update | ||
|
||
|
||
class MiniscopeImagingExtractor(MultiImagingExtractor): | ||
|
||
def __init__(self, folder_path: DirectoryPath): | ||
|
||
self.miniscope_videos_folder_path = Path(folder_path) | ||
assert self.miniscope_videos_folder_path.exists(), f"Miniscope videos folder not found in {Path(folder_path)}" | ||
|
||
self._miniscope_avi_file_paths = [p for p in self.miniscope_videos_folder_path.iterdir() if p.suffix == ".avi"] | ||
assert len(self._miniscope_avi_file_paths) > 0, f"No .avi files found in {self.miniscope_videos_folder_path}" | ||
import natsort | ||
self._miniscope_avi_file_paths = natsort.natsorted(self._miniscope_avi_file_paths) | ||
|
||
imaging_extractors = [] | ||
for file_path in self._miniscope_avi_file_paths: | ||
extractor = _MiniscopeSingleVideoExtractor(file_path=file_path) | ||
imaging_extractors.append(extractor) | ||
|
||
super().__init__(imaging_extractors=imaging_extractors) | ||
|
||
self._sampling_frequency = self._imaging_extractors[0].get_sampling_frequency() | ||
self._image_size = self._imaging_extractors[0].get_image_size() | ||
self._dtype = self._imaging_extractors[0].get_dtype() | ||
|
||
def get_num_frames(self) -> int: | ||
return self._num_frames | ||
|
||
def get_num_channels(self) -> int: | ||
return 1 | ||
|
||
def get_image_size(self) -> tuple[int, int]: | ||
return self._image_size | ||
|
||
def get_sampling_frequency(self): | ||
return self._sampling_frequency | ||
|
||
def get_dtype(self) -> DtypeType: | ||
return self._dtype | ||
|
||
def get_channel_names(self) -> list[str]: | ||
return ["OpticalChannel"] | ||
|
||
|
||
class _MiniscopeSingleVideoExtractor(ImagingExtractor): | ||
"""An auxiliar extractor to get data from a single Miniscope video (.avi) file. | ||
|
||
This format consists of a single video (.avi) | ||
Multiple _MiniscopeSingleVideoExtractor are combined into the MiniscopeImagingExtractor for public access. | ||
""" | ||
|
||
extractor_name = "_MiniscopeSingleVideo" | ||
|
||
def __init__(self, file_path: PathType): | ||
"""Create a _MiniscopeSingleVideoExtractor instance from a file path. | ||
|
||
Parameters | ||
---------- | ||
file_path: PathType | ||
The file path to the Miniscope video (.avi) file. | ||
""" | ||
from neuroconv.datainterfaces.behavior.video.video_utils import VideoCaptureContext | ||
|
||
self._video_capture = VideoCaptureContext | ||
self._cv2 = get_package(package_name="cv2", installation_instructions="pip install opencv-python-headless") | ||
self.file_path = file_path | ||
super().__init__() | ||
|
||
cap = self._cv2.VideoCapture(str(self.file_path)) | ||
|
||
self._num_frames = int(cap.get(self._cv2.CAP_PROP_FRAME_COUNT)) | ||
|
||
# Get the frames per second (fps) | ||
self._sampling_frequency = cap.get(self._cv2.CAP_PROP_FPS) | ||
self.frame_width = int(cap.get(self._cv2.CAP_PROP_FRAME_WIDTH)) | ||
self.frame_height = int(cap.get(self._cv2.CAP_PROP_FRAME_HEIGHT)) | ||
|
||
_, frame = cap.read() | ||
self._dtype = frame.dtype | ||
|
||
# Release the video capture object | ||
cap.release() | ||
|
||
def get_num_frames(self) -> int: | ||
return self._num_frames | ||
|
||
def get_num_channels(self) -> int: | ||
return 1 | ||
|
||
def get_image_size(self) -> tuple[int, int]: | ||
return (self.frame_height, self.frame_width) | ||
|
||
def get_sampling_frequency(self): | ||
return self._sampling_frequency | ||
|
||
def get_dtype(self) -> DtypeType: | ||
return self._dtype | ||
|
||
def get_channel_names(self) -> list[str]: | ||
return ["OpticalChannel"] | ||
|
||
def get_video( | ||
self, start_frame: Optional[int] = None, end_frame: Optional[int] = None, channel: int = 0 | ||
) -> np.ndarray: | ||
"""Get the video frames. | ||
|
||
Parameters | ||
---------- | ||
start_frame: int, optional | ||
Start frame index (inclusive). | ||
end_frame: int, optional | ||
End frame index (exclusive). | ||
channel: int, optional | ||
Channel index. | ||
|
||
Returns | ||
------- | ||
video: numpy.ndarray | ||
The video frames. | ||
|
||
Notes | ||
----- | ||
The grayscale conversion is based on minian | ||
https://github.com/denisecailab/minian/blob/f64c456ca027200e19cf40a80f0596106918fd09/minian/utilities.py#LL272C12-L272C12 | ||
""" | ||
if channel != 0: | ||
raise NotImplementedError( | ||
f"The {self.extractor_name}Extractor does not currently support multiple color channels." | ||
) | ||
|
||
end_frame = end_frame or self.get_num_frames() | ||
start_frame = start_frame or 0 | ||
|
||
video = np.empty(shape=(end_frame - start_frame, *self.get_image_size()), dtype=self.get_dtype()) | ||
with self._video_capture(file_path=str(self.file_path)) as video_obj: | ||
# Set the starting frame position | ||
video_obj.current_frame = start_frame | ||
for frame_number in range(end_frame - start_frame): | ||
frame = next(video_obj) | ||
video[frame_number] = self._cv2.cvtColor(frame, self._cv2.COLOR_RGB2GRAY) | ||
|
||
return video | ||
|
||
|
||
class MiniscopeImagingInterface(BaseImagingExtractorInterface): | ||
"""Data Interface for MiniscopeImagingExtractor.""" | ||
|
||
Extractor = MiniscopeImagingExtractor | ||
display_name = "Miniscope Imaging" | ||
associated_suffixes = (".avi", ".csv", ".json") | ||
info = "Interface for Miniscope imaging data." | ||
|
||
@classmethod | ||
def get_source_schema(cls) -> dict: | ||
source_schema = super().get_source_schema() | ||
source_schema["properties"]["folder_path"][ | ||
"description" | ||
] = "The folder where the Miniscope videos are contained" | ||
|
||
return source_schema | ||
|
||
@validate_call | ||
def __init__(self, folder_path: DirectoryPath): | ||
""" | ||
Initialize reading the Miniscope imaging data. | ||
|
||
Parameters | ||
---------- | ||
folder_path : DirectoryPath | ||
The folder where the Miniscope videos are contained. The video files are expected to be in folder_path | ||
|
||
""" | ||
from ndx_miniscope.utils import get_recording_start_times, read_miniscope_config | ||
|
||
super().__init__(folder_path=folder_path) | ||
|
||
self.miniscope_folder = Path(folder_path) | ||
# This contains the general metadata and might contain behavioral videos | ||
self.session_folder = self.miniscope_folder.parent | ||
|
||
self._miniscope_config = read_miniscope_config(folder_path=self.miniscope_folder) | ||
|
||
# use the frame rate of the json configuration to set the metadata | ||
frame_rate_string = self._miniscope_config["frameRate"] | ||
# frame_rate_string look like "30.0FPS", extract the float part | ||
self._metadata_frame_rate = float(frame_rate_string.split("FPS")[0]) | ||
|
||
|
||
self.photon_series_type = "OnePhotonSeries" | ||
|
||
def _get_session_start_time(self): | ||
|
||
general_metadata_json = self.session_folder/ "metaData.json" | ||
assert general_metadata_json.exists(), f"General metadata json not found in {self.session_folder}" | ||
|
||
## Read metadata | ||
with open(general_metadata_json) as f: | ||
general_metadata = json.load(f) | ||
|
||
session_start_time = datetime.datetime( | ||
year=general_metadata["year"], | ||
month=general_metadata["month"], | ||
day=general_metadata["day"], | ||
hour=general_metadata["hour"], | ||
minute=general_metadata["minute"], | ||
second=general_metadata["second"], | ||
microsecond=general_metadata["msec"] * 1000, # Convert milliseconds to microseconds | ||
) | ||
|
||
return session_start_time | ||
|
||
def get_metadata(self) -> DeepDict: | ||
from neuroconv.tools.roiextractors import get_nwb_imaging_metadata | ||
|
||
metadata = super().get_metadata() | ||
default_metadata = get_nwb_imaging_metadata(self.imaging_extractor, photon_series_type=self.photon_series_type) | ||
metadata = dict_deep_update(metadata, default_metadata) | ||
metadata["Ophys"].pop("TwoPhotonSeries", None) | ||
|
||
session_start_time = self._get_session_start_time() | ||
metadata["NWBFile"].update(session_start_time=session_start_time) | ||
|
||
device_metadata = metadata["Ophys"]["Device"][0] | ||
miniscope_config = deepcopy(self._miniscope_config) | ||
device_name = miniscope_config.pop("name") | ||
device_metadata.update(name=device_name, **miniscope_config) | ||
# Add link to Device for ImagingPlane | ||
imaging_plane_metadata = metadata["Ophys"]["ImagingPlane"][0] | ||
imaging_plane_metadata.update( | ||
device=device_name, | ||
imaging_rate=self._metadata_frame_rate, | ||
) | ||
one_photon_series_metadata = metadata["Ophys"]["OnePhotonSeries"][0] | ||
one_photon_series_metadata.update(unit="px") | ||
|
||
return metadata | ||
|
||
def get_metadata_schema(self) -> dict: | ||
metadata_schema = super().get_metadata_schema() | ||
metadata_schema["properties"]["Ophys"]["definitions"]["Device"]["additionalProperties"] = True | ||
return metadata_schema | ||
|
||
def get_original_timestamps(self) -> np.ndarray: | ||
|
||
timestamps_file_path = self.miniscope_folder / "timeStamps.csv" | ||
assert timestamps_file_path.exists(), f"Miniscope timestamps file not found in {self.miniscope_folder}" | ||
|
||
import pandas as pd | ||
|
||
timetsamps_df = pd.read_csv(timestamps_file_path) | ||
timestamps_milliseconds = timetsamps_df["Time Stamp (ms)"].values.astype(float) | ||
timestamps_seconds = timestamps_milliseconds / 1000.0 | ||
|
||
# Shift when the first timestamp is negative | ||
# TODO: Figure why, I copied from miniscope | ||
if timestamps_seconds[0] < 0.0: | ||
timestamps_seconds += abs(timestamps_seconds[0]) | ||
|
||
return np.asarray(timestamps_seconds) | ||
|
||
def add_to_nwbfile( | ||
self, | ||
nwbfile: NWBFile, | ||
metadata: Optional[dict] = None, | ||
photon_series_type: Literal["TwoPhotonSeries", "OnePhotonSeries"] = "OnePhotonSeries", | ||
stub_test: bool = False, | ||
stub_frames: int = 100, | ||
): | ||
from ndx_miniscope.utils import add_miniscope_device | ||
|
||
from neuroconv.tools.roiextractors import add_photon_series_to_nwbfile | ||
|
||
miniscope_timestamps = self.get_original_timestamps() | ||
imaging_extractor = self.imaging_extractor | ||
|
||
if stub_test: | ||
stub_frames = min([stub_frames, self.imaging_extractor.get_num_frames()]) | ||
imaging_extractor = self.imaging_extractor.frame_slice(start_frame=0, end_frame=stub_frames) | ||
miniscope_timestamps = miniscope_timestamps[:stub_frames] | ||
|
||
imaging_extractor.set_times(times=miniscope_timestamps) | ||
|
||
device_metadata = metadata["Ophys"]["Device"][0] | ||
# Cast to string because miniscope extension requires so | ||
device_metadata["gain"] = str(device_metadata["gain"]) | ||
device_metadata.pop("ewl") | ||
add_miniscope_device(nwbfile=nwbfile, device_metadata=device_metadata) | ||
|
||
add_photon_series_to_nwbfile( | ||
imaging=imaging_extractor, | ||
nwbfile=nwbfile, | ||
metadata=metadata, | ||
photon_series_type=photon_series_type, | ||
) |
Original file line number | Diff line number | Diff line change |
---|---|---|
|
@@ -776,6 +776,8 @@ This is the metadata file in the same directory that the minian videos: | |
} | ||
``` | ||
|
||
Something concerning is that the 30FPS does not correspond to the videos. The metadata says 30 FPS but the videos are 60 FPS. | ||
|
||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. I am not sure that the 60FPS reported in the file property is actually correct. There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. I think I checked against OpenCV, why are you suspicious? There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. (looking at subject EEG3-4) Because 60fps is a bit unusual for onephoton imaging (higher than the usual 20-30 Hz) + the timestamps are 33 ms apart (even if they are not always regular), that is 30Hz + if you account for 30fps, the time span of the calcium imaging recording is the same of the behavioral recording which would make more sense instead of recording with miniscope just half of the time (as it would be accounting for 60fps) There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Thanks for writing this. Let me take a close look. |
||
And this is the one in the parent folder. Both are called `metaData.json` | ||
|
||
```json | ||
|
@@ -800,6 +802,38 @@ And this is the one in the parent folder. Both are called `metaData.json` | |
} | ||
``` | ||
|
||
One big question is why the json in examples that we have in gin is different, the most important issue is that the session start time has its own key on the gin data but not here: | ||
|
||
```json | ||
{ | ||
"animalName": "", | ||
"baseDirectory": "C:/mData/2021_10_07/C6-J588_Disc5/15_03_28", | ||
"cameras": [ | ||
"BehavCam 2" | ||
], | ||
"experimentName": "", | ||
"miniscopes": [ | ||
"Miniscope" | ||
], | ||
"nameExpMouse": "C6-J588_Disc5", | ||
"recordingStartTime": { | ||
"day": 7, | ||
"hour": 15, | ||
"minute": 3, | ||
"month": 10, | ||
"msec": 635, | ||
"msecSinceEpoch": 1633644208635, | ||
"second": 28, | ||
"year": 2021 | ||
}, | ||
"researcherName": "" | ||
} | ||
``` | ||
|
||
Find out why is this different. | ||
|
||
Also, | ||
|
||
## Cross Registration | ||
|
||
|
||
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I guess the timestamps refer to the session start time. If this is correct, when shifting all timestamps, we should also shift back the session start time.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Let's discuss this briefly in the meeting. I copied this from the current miniscope code on the corresponding extension. I would rather fix the timing issues in one go when we are at the synch stage.