A high‑performance, Rust‑accelerated toolkit for multi‑modality cardiac image fusion and registration ﮩ٨ـﮩﮩ٨ـ♡ﮩ٨ـﮩﮩ٨ـ.

Overview

multimoda-rs aligns and fuses diverse cardiac imaging modalities — IVUS, OCT and CCTA — into unified high-resolution 3D models. Originally developed to quantify dynamic lumen deformation in coronary artery anomalies (CAAs), it is equally applicable to longitudinal studies (e.g., pre/post-stenting) and general coronary artery disease workflows. The Rust backend parallelizes computationally intensive registration steps for speeds well beyond pure Python.

Key Features

• Intravascular Registration: align pullback sequences (rest/stress, diastole/systole) using Hausdorff distance on vessel contours and catheter centroids; four modes: full, double-pair, single-pair, single.
• Centerline Alignment: register intravascular geometries onto a CCTA-derived centerline via three-point landmark or manual rotation.
• CCTA Fusion: automatically label CCTA geometries by vessel region and morph them to match intravascular measurements.
• Flexible Input: accepts CSV files (AIVUS format) or raw NumPy arrays.

Installation

bash pip install multimodars

Optional extras:

bash pip install "multimodars[viz]" # mesh visualisation (pyglet) pip install "multimodars[meshlab]" # MeshLab integration

For building from source or the full developer setup (tests, linting, docs), see the Installation guide.

Quick Example

Checkout our Documentation for in-depth explanations of the functionality and preloaded notebooks including 3D plots. Otherwise clone the repo to follow along the different examples:

```python import multimodars as mm import numpy as np

Align four intravascular phases (rest-dia, rest-sys, stress-dia, stress-sys)

rest, stress, dia, sys, _ = mm.fromfilefull( "examples/data/ivusrest", # inputpathab "examples/data/ivusstress", # inputpathcd writeobj=True, outputpathab="output/rest", outputpathcd="output/stress", outputpathac="output/diastole", outputpath_bd="output/systole", )

Align geometry onto a CCTA-derived centerline

clraw = np.genfromtxt("examples/data/centerlineraw.csv", delimiter=",") centerline = mm.numpytocenterline(clraw) alignedpair, clresampled = mm.alignthreepoint( centerline, rest, aorticrefpt=(12.2605, -201.3643, 1751.0554), upperrefpt=(11.7567, -202.1920, 1754.7975), lowerrefpt=(15.6605, -202.1920, 1749.9655), write=True, outputdir="output/aligned", ) ```

Pipeline

1. Intrapullback alignment — frames within each pullback are co-registered to remove cardiac-motion artefacts, yielding clean diastolic and systolic geometries:

2. Inter-pullback alignment — registered pullbacks (rest vs. stress, diastole vs. systole) are aligned against each other to reveal stress-induced and pulsatile deformation:

3. CCTA labeling — the CCTA-derived geometry is automatically segmented by vessel region (aorta, RCA, LCA, intramural) to prepare it for fusion:

4. CCTA–morphing — the labeled CCTA geometry is morphed along the centerline to match the high-resolution intravascular geometry:

5. CCTA/intravascular fusio — the morphed CCTA geometry is stitched to the intravascular geometry, replacing a section with a high resolution verison:

Documentation

Full documentation — installation, step-by-step tutorials, interactive (preloaded) Jupyter notebooks, and API reference — is available at multimoda-rs.readthedocs.io.

License

Distributed under the MIT License. See LICENSE for details.

Citation

Please kindly cite the following paper if you use this repository.

@article{stark2025multimodars, title = {multimodars: A Rust-powered toolkit for multi-modality cardiac image fusion and registration}, author = {Stark, Anselm W. and Ilic, Marc and Mokhtari, Ali and Mohammadi Kazaj, Pooya and Graeni, Christoph and Shiri, Isaac}, journal = {arXiv preprint arXiv:2510.06241}, year = {2025} }

Stark, Anselm W., Marc Ilic, Ali Mokhtari, Pooya Mohammadi Kazaj, Christoph Graeni, and Isaac Shiri. "multimodars: A Rust-powered toolkit for multi-modality cardiac image fusion and registration." arXiv preprint arXiv:2510.06241 (2025).