PrairieView Reader Added

CHANGELOG.md

@@ -2,6 +2,9 @@
 
 Observes [Semantic Versioning](https://semver.org/spec/v2.0.0.html) standard and [Keep a Changelog](https://keepachangelog.com/en/1.0.0/) convention.
 
+## [0.3.0] - 2022-10-7
++ Add - Function `prairieviewreader` to parse metadata from Bruker PrarieView acquisition system
+
 ## 0.2.1 - 2022-07-13
 + Add - Adopt `black` formatting
 + Add - Code of Conduct
@@ -19,4 +22,6 @@ Observes [Semantic Versioning](https://semver.org/spec/v2.0.0.html) standard and
 + Change - Rename the package `element-data-loader` to `element-interface`.
 
 ## 0.1.0a0 - 2021-06-21
-+ Add - Readers for: `ScanImage`, `Suite2p`, `CaImAn`.
++ Add - Readers for: `ScanImage`, `Suite2p`, `CaImAn`.
+
+[0.3.0]: https://github.com/datajoint/element-interface/releases/tag/0.3.0

element_interface/prairieviewreader.py

@@ -0,0 +1,144 @@
+import pathlib
+import xml.etree.ElementTree as ET
+from datetime import datetime
+import numpy as np
+
+
+def get_pv_metadata(pvtiffile):
+    """Extract metadata for calcium imaging scans generated by Bruker Systems PrairieView acquisition software.
+
+    The PrairieView software generates one .ome.tif imaging file per frame acquired. The metadata for all frames is contained one .xml file. This function locates the .xml file and generates a dictionary necessary to populate the DataJoint ScanInfo and Field tables.
+
+    PrairieView works with resonance scanners with a single field.
+
+    PrairieView does not support bidirectional x and y scanning.
+
+    ROI information is not contained in the .xml file.
+
+    All images generated using PrairieView have square dimensions (e.g. 512x512).
+
+
+    Args:
+        pvtiffile: An absolute path to the .ome.tif image file.
+
+    Raises:
+        FileNotFoundError: No .xml file containing information about the acquired scan was found at path in parent directory at `pvtiffile`.
+
+    Returns:
+        metainfo: A dict mapping keys to corresponding metadata values fetched from the .xml file.
+    """
+
+    # May return multiple xml files. Only need one that contains scan metadata.
+    xml_files = pathlib.Path(pvtiffile).parent.glob("*.xml")
+
+    for xml_file in xml_files:
+        tree = ET.parse(xml_file)
+        root = tree.getroot()
+        if root.find(".//Sequence"):
+            break
+    else:
+        raise FileNotFoundError(
+            f"No PrarieView metadata XML file found at {pvtiffile.parent}"
+        )
+
+    bidirectional_scan = False  # Does not support bidirectional
+
+    n_fields = 1  # Always contains 1 field
+
+    # Get all channels and find unique values
+    channel_list = [
+        int(channel.attrib.get("channel"))
+        for channel in root.iterfind(".//Sequence/Frame/File/[@channel]")
+    ]
+    n_channels = len(set(channel_list))
+
+    # One "Frame" per depth. Gets number of frames in first sequence
+    planes = [
+        int(plane.attrib.get("index"))
+        for plane in root.findall(".//Sequence/[@cycle='1']/Frame")
+    ]
+    n_depths = len(set(planes))
+
+    n_frames = len(root.findall(".//Sequence/Frame"))
+
+    roi = 1
+    # x and y coordinate values for the center of the field
+    x_field = float(
+        root.find(
+            ".//PVStateValue/[@key='currentScanCenter']/IndexedValue/[@index='XAxis']"
+        ).attrib.get("value")
+    )
+    y_field = float(
+        root.find(
+            ".//PVStateValue/[@key='currentScanCenter']/IndexedValue/[@index='YAxis']"
+        ).attrib.get("value")
+    )
+
+    framerate = 1 / float(
+        root.findall('.//PVStateValue/[@key="framePeriod"]')[0].attrib.get("value"))  # rate = 1/framePeriod
+
+    usec_per_line = float(
+        root.findall(".//PVStateValue/[@key='scanLinePeriod']")[0].attrib.get("value")) * 1e6  # Convert from seconds to microseconds
+
+    scan_datetime = datetime.strptime(
+        root.attrib.get("date"), "%m/%d/%Y %I:%M:%S %p")
+
+    total_duration = float(
+        root.findall(".//Sequence/Frame")[-1].attrib.get("relativeTime")
+    )
+
+    bidirection_z = bool(
+        root.find(".//Sequence").attrib.get("bidirectionalZ"))
+
+    px_height = int(
+        root.findall(
+            ".//PVStateValue/[@key='pixelsPerLine']")[0].attrib.get("value")
+    )
+    # All PrairieView-acquired images have square dimensions (512 x 512; 1024 x 1024)
+    px_width = px_height
+
+    um_per_pixel = float(
+        root.find(
+            ".//PVStateValue/[@key='micronsPerPixel']/IndexedValue/[@index='XAxis']"
+        ).attrib.get("value")
+    )
+
+    um_height = um_width = float(px_height) * um_per_pixel
+
+    z_min = float(root.findall(
+        ".//Sequence/[@cycle='1']/Frame/PVStateShard/PVStateValue/[@key='positionCurrent']/SubindexedValues/SubindexedValue/[@subindex='0']"
+    )[0].attrib.get("value"))
+    z_max = float(root.findall(
+        ".//Sequence/[@cycle='1']/Frame/PVStateShard/PVStateValue/[@key='positionCurrent']/SubindexedValues/SubindexedValue/[@subindex='0']"
+    )[-1].attrib.get("value"))
+    z_step = float(root.find(
+        ".//PVStateShard/PVStateValue/[@key='micronsPerPixel']/IndexedValue/[@index='ZAxis']"
+    ).attrib.get("value"))
+    z_fields = np.arange(z_min, z_max + 1, z_step)
+    assert z_fields.size == n_depths
+
+    metainfo = dict(
+        num_fields=n_fields,
+        num_channels=n_channels,
+        num_planes=n_depths,
+        num_frames=n_frames,
+        num_rois=roi,
+        x_pos=None,
+        y_pos=None,
+        z_pos=None,
+        frame_rate=framerate,
+        bidirectional=bidirectional_scan,
+        bidirectional_z=bidirection_z,
+        scan_datetime=scan_datetime,
+        usecs_per_line=usec_per_line,
+        scan_duration=total_duration,
+        height_in_pixels=px_height,
+        width_in_pixels=px_width,
+        height_in_um=um_height,
+        width_in_um=um_width,
+        fieldX=x_field,
+        fieldY=y_field,
+        fieldZ=z_fields,
+    )
+
+    return metainfo

element_interface/version.py

@@ -1,2 +1,2 @@
 """Package metadata"""
-__version__ = "0.2.1"
+__version__ = "0.3.0"