What's Changed

• feat(examples): add segmentation & feature extraction examples by @alxndrkalinin in #33
• fix: address code quality issues #35-#38 by @alxndrkalinin in #39
• feat(metrics): add binomial splitting for single-image FRC/FSC by @alxndrkalinin in #40

Full Changelog: v0.7.0a1...v0.7.0a2

v0.7.0a1 — Alpha Release

Alpha release of cubic v0.7.0, adding spectral resolution metrics and band-limited image quality assessment.

Highlights

• Decorrelation Analysis (DCR) — Single-image, parameter-free resolution estimation for 2D and 3D with angular sectoring (Descloux et al. 2019)
• Histogram FSC backend — GPU-accelerated 3D FSC with proper XY/Z anisotropy correction (Koho et al. 2019)
• Band-limited metrics — Butterworth low-pass filtering, band-limited PCC/SSIM, spectrally-weighted PCC
• Package reorganization — cubic.metrics.frc → cubic.metrics.spectral (backward-compatible via deprecation shim, removal in 0.8.0)
• GPU-optional notebooks — All example notebooks detect GPU availability and fall back to CPU

Breaking changes

• pad_image_to_cube() default mode: "reflect" → "constant"
• fsc_resolution() default bin_delta: 10 → 1
• fsc_resolution() default spacing: 1.0 → None

Full changelog: v0.6.0...v0.7.0a1

Initial PyPI release of cubic, a Python library for CUDA-accelerated processing and analysis of multidimensional (2D/3D+) bioimages.

Highlights:

• Device-agnostic API wrapping scipy/scikit-image and CuPy/cuCIM
• GPU-accelerated Fourier Ring/Shell Correlation (FRC/FSC) for image resolution
• Richardson-Lucy deconvolution, average precision for segmentation, and more
• Install with pip install cubic

Citation:

Kalinin, A.A., Carpenter, A.E., Singh, S., & O'Meara, M.J. (2025). cubic: CUDA-accelerated 3D BioImage Computing. IEEE International Conference on Computer Vision Workshops (ICCVW).

Initial release accompanying our 2025 ICCVW publication