Added a utility class to read lif files.

.gitignore

@@ -0,0 +1,2 @@
+.DS_Store
+.idea/*

bioloaders/loaders.py

@@ -0,0 +1,67 @@
+"""
+A utility class to load various types of images and convert them to numpy.arrays
+
+Author: Sarah Kefayati
+Email: [email protected]
+"""
+
+__author__ = "Sarah Kefayati"
+__email__ = "[email protected]"
+
+import os
+import shutil
+
+import bioformats as bf
+import javabridge as jv
+import numpy as np
+import pandas as pd
+
+
+class Loaders:
+
+    @staticmethod
+    def lif_reader(lif_file_path, pickled_data_path):
+        """
+        This function reads lif file, separates channels, series, and time points and
+        converts the images into 3D numpy.arrays. It then stores the images into pickle
+        files and saves their path in a pandas.dataframe.
+
+        :param lif_file_path: Input lif file.
+        :param pickled_data_path: Path of the folder to store pickled images.
+        :return: A dataframe containing the metadata for pickled images.
+        """
+        jv.start_vm(class_path=bf.JARS, max_heap_size='8G')
+        md = bf.get_omexml_metadata(lif_file_path)
+        o = bf.OMEXML(md)
+        n_channels = o.image().Pixels.channel_count
+        n_series = o.get_image_count()
+        time_points = o.image().Pixels.SizeT
+        size_x = o.image().Pixels.SizeX
+        size_y = o.image().Pixels.SizeY
+        size_z = o.image().Pixels.SizeZ
+        rdr = bf.ImageReader(lif_file_path, perform_init=True)
+
+        if os.path.exists(pickled_data_path):
+            shutil.rmtree(pickled_data_path)
+        os.makedirs(pickled_data_path)
+
+        records = []
+
+        for c in range(0, n_channels):
+            for n in range(n_series):
+                for t in range(time_points):
+                    img_out = np.empty(shape=(size_x, size_y, size_z))
+
+                    full_path_out = os.path.join(pickled_data_path, 'image_c{}_n{}_t{}'.format(c, n, t))
+
+                    if not os.path.exists(full_path_out):
+                        for z in range(size_z):
+                            img_out[:, :, z] = rdr.read(c=c, z=z, t=t, series=n, rescale=False)
+                        np.save(full_path_out, img_out, allow_pickle=True, fix_imports=True)
+
+                    records.append((full_path_out, c, n, t))
+
+        df = pd.DataFrame.from_records(records, columns=['image_path', 'channel', 'series', 'time'])
+
+        jv.kill_vm()
+        return df

requirements.txt

@@ -1,3 +1,5 @@
+javabridge
+python-bioformats
 numpy
 scipy
 matplotlib