jandoG / napari_interactive_image_cropping

Interactively crop fluorescence microscopy 2D/3D time-lapse images.

README

This repository contains jupyter notebooks which interactively crop fluorescence microscopy 2D time-lapse images with 2 channels.

Analysis goal:

• Crop interactively different regions in the 2D files, normalize and save them
• Expected input images: 2D, 2 channels, time-lapse, time image files saved in folder (need to concatenate)
• Expected output: cropped image 2D , single channel time-lapse, normalized to combat variabilities in intensities. Additionally, drift corrected.

Workflow using Napari:

• Read images from folder - concatenate and display (XYCT).
• Display a rectangular Region of Interest (ROI).
• Ask user to adjust ROI (Region of Interest) to crop.
• Crop image to the ROI coordinates.
• Ask user to specify channel to normalize.
• Normalize the histogram of the cropped single channel time sequence image - options between percentile/ clahe methods.
• Drift correct the image using image registration.
• Option to normalize other channel.
• Option to apply same transformation as previous channel (drift correction).
• Save images: merge channels, save images with proper calibration and imagej compatible. Ask user to specify calibration and frame interval at the start and apply this to images before saving.

Python libraries used

• tifffile: reading and writing images
• napari: display and cropping
• scikit image: histogram normalization
• pystackreg: image registration to correct for drifting nuclei.

Installation

• Use the environment file unser set_up_instructions to install napari and other required libraries.
• Keep the script imgprocessing_functions.py next to the jupyter notebook while running.

This project was original done in collaboration with Ksenia Kuznetsova, PhD at Vastenhouw lab (Oct 2021, MPI-CBG, Dresden). The code is published with permission.