Beginner's Guide to Labeling Cells with napari and Python in a Jupyter Notebook Environment

Labeling Cells with Napari and Python: A Step-by-Step Guide for BioImage Analysis In this tutorial, we will explore how to label cells using Napari, an interactive multi-dimensional image viewer for Python that is particularly well-suited for microscopy data. This guide is designed to be practical and beginner-friendly, catering to biologists, data scientists, and image analysts. By the end of this tutorial, you will be able to load microscopy images into Python, navigate Napari’s labeling tools, and save your labeled images in a structured format for further analysis or machine/deep learning applications. We will conduct all steps within a Jupyter Notebook environment, seamlessly combining Python scripting with visual exploration and annotation using Napari. To get the most out of this tutorial, it is assumed that you have a basic understanding of microscopy, Python syntax, and how to work with Jupyter Notebooks. Familiarity with image segmentation concepts is also beneficial. Step 1: Create the Folder Structure To efficiently manage and label your microscopy images, especially .lif files from a Leica microscope, it's crucial to set up a well-organized folder system. Here, we will use a four-folder structure to keep our project tidy: Raw .lif Files: Store the original .lif microscopy files here. These should be exactly as exported from your Leica microscope to ensure the integrity of the raw data. Processed Images: This folder will contain the images after they have been converted from .lif format to a more manageable format, such as TIFF or PNG. Labeled Images: Save the annotated images generated during the labeling process in this folder. Each labeled image should correspond to a processed image. Annotations: Store the metadata and label information for each image in this folder. This data will be useful for downstream analysis and training machine learning models. Create these folders in a location that is easy to access and manage, such as within your project directory. Step 2: Install Required Libraries Before we begin, ensure you have the necessary libraries installed. You can install them using pip in your Jupyter Notebook: python !pip install napari lif-reader tifffile These libraries include: - Napari: An interactive multi-dimensional image viewer. - Lif Reader: A library to read Leica .lif files. - Tifffile: A library to handle TIFF image files. Step 3: Load Microscopy Data Next, we need to load the .lif files from the "Raw .lif Files" folder. We will use the lif_reader library to do this: ```python import os from lif_reader import LifFile Define the path to the raw .lif files raw_folder = 'path/to/your/Raw .lif Files' List all .lif files in the folder lif_files = [f for f in os.listdir(raw_folder) if f.endswith('.lif')] Load the first .lif file lif_file_path = os.path.join(raw_folder, lif_files[0]) lif_file = LifFile(lif_file_path) Access the first image channel image = lif_file.get_image(0).get_pixel_data() Display the image to verify it's loaded correctly import matplotlib.pyplot as plt plt.imshow(image) plt.show() ``` This script reads the first .lif file in the specified folder and extracts the pixel data for the first image channel. It then displays the image to confirm that the data has been loaded correctly. Step 4: Convert and Save Processed Images Once the images are loaded, we will convert them to a more convenient format, such as TIFF, and save them in the "Processed Images" folder: ```python import tifffile Define the path to the processed images folder processed_folder = 'path/to/your/Processed Images' Create the processed images folder if it doesn't exist os.makedirs(processed_folder, exist_ok=True) Save the processed image tifffile.imwrite(os.path.join(processed_folder, 'image_0.tiff'), image) ``` This code snippet creates the "Processed Images" folder if it does not already exist and saves the processed image in TIFF format. Step 5: Label Images Using Napari Now, we will use Napari to label the cells in the processed images. Start by launching Napari and loading the image: ```python import napari Launch Napari viewer viewer = napari.Viewer() Add the image to the viewer viewer.add_image(image) Add a labels layer to the viewer labels_layer = viewer.add_labels(np.zeros_like(image, dtype=np.uint8)) ``` In the Napari viewer, you can manually draw labels on the image using the brush tool. Each cell can be assigned a unique label, typically represented by different colors. Step 6: Save Labeled Images and Annotations After labeling the cells, save both the labeled images and the annotations: ```python import numpy as np Define the path to the labeled images folder labeled_images_folder = 'path/to/your/Labeled Images' os.makedirs(labeled_images_folder, exist_ok=True) Define the path to the annotations folder annotations_folder = 'path/to/your/Annotations' os.makedirs(annotations_folder, exist_ok=True) Save the labeled image tifffile.imwrite(os.path.join(labeled_images_folder, 'labeled_image_0.tiff'), labels_layer.data) Save the label metadata (annotations) np.save(os.path.join(annotations_folder, 'annotation_0.npy'), labels_layer.data) ``` This script saves the labeled image in TIFF format and the label metadata in NumPy array format, ensuring that all labeled data is stored in a structured manner. Step 7: Repeat for Additional Images If you have multiple images to label, repeat steps 3 through 6 for each one. Automate this process by iterating over the list of .lif files: ```python for index, file in enumerate(lif_files): lif_file_path = os.path.join(raw_folder, file) lif_file = LifFile(lif_file_path) image = lif_file.get_image(0).get_pixel_data() # Save the processed image tifffile.imwrite(os.path.join(processed_folder, f'image_{index}.tiff'), image) # Launch Napari viewer viewer = napari.Viewer() viewer.add_image(image) labels_layer = viewer.add_labels(np.zeros_like(image, dtype=np.uint8)) # Wait for user to label the image input("Press Enter to proceed once labeling is complete...") # Save the labeled image and annotations tifffile.imwrite(os.path.join(labeled_images_folder, f'labeled_image_{index}.tiff'), labels_layer.data) np.save(os.path.join(annotations_folder, f'annotation_{index}.npy'), labels_layer.data) ``` This loop automates the process of converting and labeling multiple images. After labeling each image, the user must press Enter to proceed to the next one. Conclusion By following these steps, you can effectively label cells in microscopy images using Napari and Python. This structured approach ensures that your raw data, processed images, labeled images, and annotations are organized and ready for further analysis or machine learning tasks. Napari’s interactive capabilities make manual labeling intuitive and efficient, even for complex multi-dimensional images. With this foundation, you can advance your bioimage analysis projects and explore more sophisticated techniques in image processing and AI.