komamripy

Pulseq-compatible, high-performance MRI simulation in Python.

komamripy is a thin Python interface to KomaMRI.jl, a high-performance Julia framework for MRI simulation. It exposes KomaMRI to Python through juliacall, so Python users can run fast CPU/GPU MRI simulations without writing any Julia.

Installation

Install from PyPI with uv (recommended):

uv init project-folder && cd project-folder
uv add komamripy

or with pip:

pip install komamripy

Quick start

The first import downloads Julia and precompiles KomaMRI, which can take a few minutes. Subsequent runs are fast.

import komamripy as km
import numpy as np

km.load_cuda()  # GPU backend selection, optional

sys = km.Scanner()  # scanner hardware
obj = km.brain_phantom2D()  # 2D brain phantom
seq = km.PulseDesigner.EPI_example()  # example EPI sequence

sim_params = {"return_type": "mat"}  # return the raw signal matrix
raw = km.simulate(obj, seq, sys, sim_params=sim_params)

print(np.shape(raw))

You do not need to install Julia yourself: juliacall provisions a suitable Julia automatically, and KomaMRI is installed on first import.

How it works

komamripy mirrors the KomaMRI Julia namespace: any function or type that KomaMRI exposes is available as an attribute of komamripy. Julia code therefore translates to Python almost line for line.

Simulation results are returned as Julia objects; convert array-like results to NumPy with numpy.asarray.

Status

komamripy is under active development as a Google Summer of Code project.

Project status and roadmap are tracked in JuliaHealth/komamripy#9.