Python scientific stack

The feature extraction pipeline leverages industry-standard python scientific libraries for efficient image processing and computation.

Core libraries

Scikit-image

Purpose: image processing and analysis Website: https://scikit-image.org/

Features used:

  • Image segmentation algorithms

  • Morphological operations

  • 3D image analysis

  • Regional property measurement

Scipy

Purpose: scientific computing Website: https://scipy.org/

Features used:

  • Advanced signal processing

  • Statistical distributions

  • Interpolation and filtering

  • Multi-dimensional image processing

Mahotas

Purpose: image analysis and processing Website: https://mahotas.readthedocs.io/en/latest

Features used:

  • Gray-level co-occurrence matrix (GLCM)

  • Texture feature extraction

  • Morphological operations

Numpy

Purpose: numerical computing Website: https://numpy.org/

Features used:

  • Multi-dimensional array operations

  • Vectorized computations

  • Statistical calculations

GPU-accelerated libraries

Cupy

Purpose: GPU-accelerated arrays (drop-in numpy replacement) Website: https://docs.cupy.dev/en/stable

Features:

  • GPU memory management

  • Parallel numerical operations

  • 10-100X speedup over CPU numpy

Cucim

Purpose: GPU-accelerated image processing Website: https://docs.rapids.ai/api/cucim/stable/

Features:

  • GPU-accelerated scikit-image operations

  • Fast morphological operations

  • Efficient image filtering