Linking RayStation AI auto-contouring with Eclipse TPS - a scripted workflow for clinical integration

Sep 29, 2026

Efficient, accurate delineation of target volumes and organs-at-risk (OARs) is essential in radiotherapy planning, yet deep learning auto-contouring integration across different vendor platforms remains difficult. The study presents an automated, cross-platform workflow combining RayStation AI-based auto-contouring with the Eclipse TPS. Custom C# and Python scripts automate DICOM-based data exchange, triggering segmentation in RayStation 2024B and returning contours to Eclipse for review with minimal user interaction.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages9

Reading Time23 min

AccessFree to download

Abstract

Introduction

Clinical need and limitations of manual and atlas-based contouring

Motivation for integrating deep learning AI segmentation

Materials and methods

Results

Conclusion

© 2025 Greater Poland Cancer Centre. Published by Via Medica. All rights reserved. e-ISSN 2083–4640 ISSN 1507–1367 ISSN: 1507–1367 RESEARCH PAPER Reports of Practical Oncology and Radiotherapy 2025, Volume 30, Number 6, pages: 749–757 DOI: 10.5603/rpor.109182 Submitted: 12.09.2025 Accepted: 20.10.2025 Linking RayStation AI auto-contouring with Eclipse TPS: a scripted workflow for clinical integration Adam Ryczkowski1, Agata Jodda1, Tomasz Piotrowski1–3 1Department of Medical Physics, Greater Poland Cancer Centre, Poznan, Poland 2Department of Electroradiology, Poznan University of Medical Sciences, Poznan, Poland 3Department of Biomedical Physics, Adam Mickiewicz University in Poznan, Poznan, Poland ABSTRACT Background: Efficient and accurate delineation of target volumes and organs-at-risk (OARs) is critical in radiotherapy planning. Deep learning-based auto-contouring improves consistency and reduces manual effort, but integrating AI tools into established workflows is challenging when platforms differ. This work presents an automated integration of RayStation’s AI-based auto-contouring with the Eclipse treatment planning system (TPS) . Materials and methods: Custom scripts were developed in Eclipse (C\#) and RayStation (Python), supported by DICOM networking, to enable automated data exchange. The Eclipse script exports computed tomography (CT) images and a list of requested structures to RayStation. Upon receipt, RayStation 2024B automatically triggers a Python script that performs deep learning segmentation using built-in models. The resulting contours are then exported back to Eclipse for review, requiring no user interaction beyond initial input. Results: The integration was validated on 35 clinical cases. The complete process, from CT export to contour availability in Eclipse, averaged 1.1 minutes per case. The solution significantly reduces contouring time and improves workflow consistency. Conclusion: This automated cross-platform integration enables deep learning segmentation from RayStation to be accessed directly within Eclipse, offering a streamlined and scalable contouring solution. It enhances planning efficiency, reduces inter-observer variability, and allows clinicians to benefit from state-of-the-art AI tools within their existing TPS. The framework is modular and adaptable to evolving clinical protocols, serving as a practical example of multi-vendor system interoperability in modern radiotherapy planning. Keywords: deep learning auto-contouring; radiotherapy treatment planning; RayStation — Eclipse integration; artificial intelligence in radiation oncology; DICOM automation Rep Pract Oncol Radiother 2025;30(6):749–757 Introduction Accurate delineation of target volumes and organs-at-risk (OARs) is a critical yet time-consuming step in radiotherapy treatment planning. Manual contouring by clinicians can delay planning and introduce inter-observer variability [1–3] . Traditional auto-contouring approaches, such as atlas-based segmentation, have been used to alleviate this burden by propagating contours from reference atlases [4] . While atlas-based methods can reduce initial contouring time, they often still Address for correspondence: Tomasz Piotrowski, Department of Medical Physics, Greater Poland Cancer Centre, Poznan, Poland; e-mail: [tomasz.piotrowski@me.com](tomasz.piotrowski@me.com) This article is available in open access under Creative Common Attribution-Non-Commercial-No Derivatives 4.0 International (CC BY-NC-ND 4.0) license, allowing to download articles and share them with others as long as they credit the authors and the publisher, but without permission to change them in any way or use them commercially [https://journals.viamedica.pl/rpor](https://journals.viamedica.pl/rpor) 749 require significant manual correction to achieve clinically acceptable accuracy [5] . In recent years, deep learning (DL) algorithms have emerged asa promising soluti

FAQs

Why is target and OAR delineation critical in radiotherapy planning?

Accurate delineation is essential because it directly affects radiotherapy treatment planning quality, while manual contouring is time-consuming and can introduce inter-observer variability.

What is the proposed workflow for integrating RayStation AI auto-contouring with Eclipse TPS?

Eclipse exports CT images and requested structures via custom scripts, RayStation 2024B triggers a Python deep learning segmentation using built-in models, and the resulting contours are exported back to Eclipse for review.

How was the integration validated and what performance was reported?

Validation used 35 clinical cases. The end-to-end process from CT export to contour availability in Eclipse averaged 1.1 minutes per case, reducing contouring time and improving workflow consistency.

Sep 29, 2026

Uploaded by RuangKosong