pyomarker 0.0.1.dev14

A library of tools and functionality for frequentist and Bayesian imaging biomarker analysis.

pyomarker

Bayesian and classical methods for quantitative imaging biomarker reliability and uncertainty.

Quantitative imaging biomarkers are increasingly used in clinical research, yet their repeatability and uncertainty are often poorly characterised. pyomarker provides well-defined frequentist and Bayesian tools for analysing test–retest data, estimating measurement reliability, and quantifying uncertainty in imaging biomarker studies. It leverages Hamiltonian Monte Carlo sampling of parameter posterior distributions through the Stan probabilistic programming framework.

---

⚠️ Development status
pyomarker is under active development. APIs may change. If you would like more information about upcoming releases, please contact me.

---

Key features

• Classical repeatability metrics (Bland–Altman, ICC, CoV)
• Bayesian models for uncertainty-aware biomarker analysis
• Designed for quantitative imaging workflows

Installation

pip install pyomarker

📘 Documentation: https://icr-computational-imaging.github.io/pyomarker/

Example

import numpy as np
from pyomarker.models.test_retest.real.bland_altman import BlandAltman

x1 = np.array([1.2, 1.4, 1.1, 1.3])
x2 = np.array([1.3, 1.5, 1.0, 1.2])

ba = BlandAltman(ci=0.90).fit(x1, x2)
print(ba.metrics())

Check our examples page for more.