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

2025-11-11

AI Summary

To address the challenge of jointly optimizing physical models and deep learning in X-ray CT reconstruction, this paper proposes an end-to-end differentiable CT reconstruction framework. Methodologically, it introduces high-accuracy, fully differentiable forward/back-projection operators implemented via native PyTorch CUDA kernels for efficient GPU acceleration. The framework supports diverse acquisition geometries—including parallel-, fan-, and cone-beam configurations—as well as arbitrary scanning trajectories, and incorporates an integrated artifact simulation module. Crucially, it unifies the physical imaging model and neural networks into a single trainable pipeline, leveraging automatic differentiation and GPU parallelism. Experimental results demonstrate substantial improvements in both reconstruction accuracy and training efficiency, while preserving flexibility and extensibility. The open-source implementation provides a foundational tool for interpretable and verifiable AI research in medical imaging.

Technology Category

• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Mixed Discrete/Continuous Search
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

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

Problem

• Developing differentiable CT operators for neural network integration
• Extending compatibility to PyTorch with CUDA kernel support
• Creating end-to-end trainable CT reconstruction pipelines

Innovation

• Extends compatibility to PyTorch framework
• Adds native CUDA kernel support for geometries
• Provides tools for simulating imaging artifacts