Installation

This section describes how to install the pipeline and its required dependencies. It covers setting up a Python environment, installing core packages, and configuring optional GPU support for accelerated segmentation. Following these steps ensures the pipeline runs correctly on Linux, Windows, or Mac systems.

Prerequisites

Cellpose segmentation can run on CPU without a GPU, but processing may be slower.

  • Python 3.10+ (recommended)

  • Linux/Windows NVIDIA GPU with CUDA 11+ for accelerated segmentation (optional)

  • Apple Silicon Mac (M1-M3) 16 GB RAM for processing large images (recommended)

Recommended: create a virtual environment to avoid dependency conflicts

python -m venv cellseg_env
source cellseg_env/bin/activate  # Linux/macOS
cellseg_env\Scripts\activate     # Windows

Install Core Dependencies

Install the required Python packages:

pip install cellpose spatialdata geopandas anndata scikit-image numpy pandas matplotlib

Optional GPU Support

If you have a compatible NVIDIA GPU (Linux/Windows) or an Apple Silicon Mac (M1–M3), you can install the GPU-enabled version of Cellpose.

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
pip install cellpose[all]

# Show all TensorFlow logs
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '0'
import tensorflow as tf

# Confirming system information, TensorFlow version & GPU access
print('Python path:', sys.executable)
print('TensorFlow version:', tf.__version__)
print('GPU available:', tf.config.list_physical_devices('GPU'))