Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

An update to PYRO-NN: A Python Library for Differentiable CT Operators

Linda-Sophie Schneider, Yipeng Sun, Chengze Ye, Markus Michen, Andreas Maier

TL;DR

The paper updates PYRO-NN, a Python library for differentiable CT reconstruction, by extending compatibility to PyTorch and adding native CUDA kernels for projection and backprojection across parallel, fan, and cone-beam geometries. It provides tools for realistic artifact simulation, arbitrary trajectory modeling, and end-to-end trainable pipelines via a high-level Python API, enabling seamless integration of physics-based models with data-driven learning. By embedding differentiable CT operators and enabling gradient flow without large explicit system matrices, PYRO-NN supports end-to-end optimization of reconstruction pipelines. The work demonstrates practical usage through an FBP-style pipeline and introduces enhancements for complex trajectories and artifacts, positioning PYRO-NN as a versatile framework bridging classical CT methods and deep learning in modern imaging contexts.

Abstract

Deep learning has brought significant advancements to X-ray Computed Tomography (CT) reconstruction, offering solutions to challenges arising from modern imaging technologies. These developments benefit from methods that combine classical reconstruction techniques with data-driven approaches. Differentiable operators play a key role in this integration by enabling end-to-end optimization and the incorporation of physical modeling within neural networks. In this work, we present an updated version of PYRO-NN, a Python-based library for differentiable CT reconstruction. The updated framework extends compatibility to PyTorch and introduces native CUDA kernel support for efficient projection and back-projection operations across parallel, fan, and cone-beam geometries. Additionally, it includes tools for simulating imaging artifacts, modeling arbitrary acquisition trajectories, and creating flexible, end-to-end trainable pipelines through a high-level Python API. Code is available at: https://github.com/csyben/PYRO-NN

An update to PYRO-NN: A Python Library for Differentiable CT Operators

TL;DR

The paper updates PYRO-NN, a Python library for differentiable CT reconstruction, by extending compatibility to PyTorch and adding native CUDA kernels for projection and backprojection across parallel, fan, and cone-beam geometries. It provides tools for realistic artifact simulation, arbitrary trajectory modeling, and end-to-end trainable pipelines via a high-level Python API, enabling seamless integration of physics-based models with data-driven learning. By embedding differentiable CT operators and enabling gradient flow without large explicit system matrices, PYRO-NN supports end-to-end optimization of reconstruction pipelines. The work demonstrates practical usage through an FBP-style pipeline and introduces enhancements for complex trajectories and artifacts, positioning PYRO-NN as a versatile framework bridging classical CT methods and deep learning in modern imaging contexts.

Abstract

Deep learning has brought significant advancements to X-ray Computed Tomography (CT) reconstruction, offering solutions to challenges arising from modern imaging technologies. These developments benefit from methods that combine classical reconstruction techniques with data-driven approaches. Differentiable operators play a key role in this integration by enabling end-to-end optimization and the incorporation of physical modeling within neural networks. In this work, we present an updated version of PYRO-NN, a Python-based library for differentiable CT reconstruction. The updated framework extends compatibility to PyTorch and introduces native CUDA kernel support for efficient projection and back-projection operations across parallel, fan, and cone-beam geometries. Additionally, it includes tools for simulating imaging artifacts, modeling arbitrary acquisition trajectories, and creating flexible, end-to-end trainable pipelines through a high-level Python API. Code is available at: https://github.com/csyben/PYRO-NN

Paper Structure

This paper contains 10 sections, 1 equation, 1 figure.

Table of Contents

  1. Introduction
  2. The PYRO-NN Framework
  3. Modular Layer Design
  4. CT Reconstruction in Neural Networks
  5. Code Example: Filtered Backprojection in PYRO-NN
  6. Framework Updates and Enhancements
  7. PyTorch Integration
  8. Advanced Simulation Tool
  9. Flexible Trajectory Modeling
  10. Conclusion and Future Directions

Figures (1)

  • Figure 1: Visualization of simulated artifacts with Shepp Logan Phantom.