torch-refine-tilt-axis-angle 0.0.2

Tilt-axis angle optimization for tilt series using common lines in PyTorch

torch-refine-tilt-axis-angle

Tilt-axis angle optimization for tilt series using common lines.

Overview

torch-refine-tilt-axis-angle provides an implementation for finding the in plane rotation for a tilt-series that aligns the tilt axis with the y-axis. This is done by extracting lines from the 2D Fourier transform of the tilt-series images and minimizing the mean squared error between them, i.e. the common lines. It makes use of the pytorch LBFGS optimizer to find the minimum.

Installation

pip install torch-refine-tilt-axis-angle

Usage

import torch
from torch_refine_tilt_axis_angle import refine_tilt_axis_angle
from torch_grid_utils import circle

# Load or create your tilt series
# tilt_series shape: (batch, height, width) - batch is number of tilt images
# Example: tilt_series with shape (61, 512, 512) - 61 tilt images of 512x512 pixels
tilt_series = torch.randn(61, 512, 512)

# Specify an initial guess for the tilt axis angle (the default is 0.0)
# This can be the value from an MDOC file.
initial_tilt_axis_angle = 50.0

# Define a circular mask that removes some noise and help the optimization target.
alignment_mask = circle(236, (512, 512), smoothing_radius=20)

# Run tilt axis angle refinement.
new_tilt_axis_angle = refine_tilt_axis_angle(
    tilt_series=tilt_series,
    alignment_mask=alignment_mask,
    tilt_axis_angle=initial_tilt_axis_angle,
    # grid_points=1,  # optionally increase the spline grid points (default 1)
    # return_single_angle=False,  # optionally write out an angle for each image 
)

Use uv to run an example with simulated data and visualize the results.

uv run examples/simulate_tilt_axis_refinement.py

License

This package is distributed under the BSD 3-Clause License.