CTF Experimental Data

Multidistance CFT phase retrieval

This examples covers the usage of pyPhase for phase retrieval using CTF on a real dataset.

The images are is in tiff format and are not preprocessed.

Install pyPhase. Then:

from pyphase import dataset
from pyphase import phaseretrieval
from pyphase import utilities 

import matplotlib.pyplot as plt # for visualization purposes
from skimage import io

Dataset setup

Define the projects name

name = 'safold'

and set the path to the data

path = '/data/staff/tomograms/safold'

Make sure that the data folders and files have the right format and structure.

You can have a look at the data

fig, ax = plt.subplots(nrows = 2, ncols = 3, figsize = (30,20))

for i,row in enumerate(ax):
    for j,col in enumerate(row):
        file = '/data/staff/tomograms/users/diogo/Data/safold/safold_{}_0040.tif'.format((i+1)*(j+1))

        im = io.imread(file)
        col.imshow(im, cmap='gray')

Projections

pyPhase will look the parameter in the projects information files.

By default, it will include all the projections' distances found, which will be used for the phase retrieval. If you wish to use only some of the distances you may edit the file and change it according to your specifications.

Create the DataSet object

ds = dataset.TiffDataSet(path,name)

To have pyPhase perform flat-field correction and do shifts correction set

Preprocessing

ds.preprocess=1
ds.correct_shifts=1

and set

ds.correct_motion = 0

Use the utilities module to choose a registration algorithm

RA = utilities.ElastixCorrelationRegistrationAlgorithm()

Observe that using utilitie requires having elaxtix installed.

To get the shifts run

ds.Align(RA, interval = 8)

where interval is the number of projections in between the images to align. This create a file shifts.picle, in containing the shifts. These then fit into a polynomial using

ds.FitShifts()

Note:

Magnification correction is not yet implemented.

Phase retrieval

At this stage we can create our CTF retriever object with

alpha = 1e-8
retriever = phaseretrieval.CTF(ds, alpha)

Refere to the API for other phase retrieval algorithms available in pyPhase.

Choose the range of projections to which to apply the phase retrieval

start = 26
end = 27

Retrieve the phase with

retriever.ReconstructProjections(ds, start, end)

For each projection, a file named safold_[version_prefix]_00[n].tif with the retrieved phase and and file named safold[version_prefix]_att__00[n], with the retrieved attenuation, is created in /data/staff/tomograms/safold_.

Visualizing the results

Look at the results by plotting the retrieved phase and retrieved attenuation for one projection

plt.figure(figsize = [8,8])
im = io.imread(path + 'safold_/safold_PP_0026.tif')
plt.imshow(imm, cmap = 'gray')

Projections

plt.figure(figsize = [8,8])
im = io.imread(path + 'safold_/safold_PP_att_0026.tif')
plt.imshow(imm, cmap = 'gray')

Projections