Convert Micromanager multidimensional tiff data into zarr format

This notebook explains how to use Pycromanger to readout the multi-dimensional data saved by Micro-manager and convert it into zarr format.

This is useful when the multi-dimensional data is large (more than hundreds of GBs). Currently there is no python reader that can directly readout the large multi-dimensional data saved by Micromanger. Therefore, Pycromanger can be used as a reader to bridge this gap.

To run this notebook, first open Micromanger, then load the multi-dimensional data into Micromanger using virtual stack. Pycromanger will then be able to read out that data via a Java-Python bridge.

After the data is converted to zarr, one can easier load the data into Python and perform downsteam processing.

import numpy as np
import zarr

from pycromanager import Bridge

Construct java objects

bridge = Bridge()
mm = bridge.get_studio()
ds = mm.displays().get_active_data_viewer().get_data_provider()

Set parameters

nb_tp = 500  # number of time points
nb_ch = 2  # number of channels
nz = 600  # number of z slices
ny = 2048  # number of y pixels in each image
nx = 2048  # number of x pixels in each image
save_to = r"D:\zarr_data\data.zarr"  # save to this directory

Create Zarr folder

if not os.path.isdir(save_to):
    # create new zarr folder
    root = zarr.open(save_to, mode="w")
    ch0 = root.zeros(
        "ch0", shape=(nb_tp, nz, ny, nx), chunks=(1, 64, 512, 512), dtype="i2"
    )
    ch1 = root.zeros(
        "ch1", shape=(nb_ch, nz, ny, nx), chunks=(1, 64, 512, 512), dtype="i2"
    )
    print(root.tree())

# open the zarr folder to save data into it
root = zarr.open(save_to, mode="rw")
ch0 = root["ch0"]
ch1 = root["ch1"]

Readout data from MM and save to zarr

data_stack = np.zeros((nz, ny, nx))  # initialize nparray to store a z-stack
for t in range(nb_tp):
    for ch in range(nb_ch):
        for z in range(nb_slices):
            coords_string = "t={0},p=0,c={1},z={2}".format(t, ch, z)
            coords = mm.data().create_coords(coords_string)
            mimg = ds.get_image(coords)
            data_stack[z, :, :] = np.reshape(
                mimg.get_raw_pixels(), newshape=[mimg.get_height(), mimg.get_width()]
            )
        if ch == 0:
            ch0[t, :, :, :] = data_stack
        else:
            ch1[t, :, :, :] = data_stack