Evaluation of nodal outlining based on ESTRO guidelines in breast radiotherapy planning
PO-0934
Abstract
Evaluation of nodal outlining based on ESTRO guidelines in breast radiotherapy planning
Authors: Welgemoed|, Camarie(1)*[c.welgemoed@nhs.net];Spezi|, Emiliano(2);Chowdhury|, Amani(1);Riddle|, Pippa(1);Amy|, Clifford(1);McLauchlan|, Ruth(1);Dorothy|, Gujral(1);Aboagye|, Eric(3);
(1)Imperial College Healthcare NHS Trust, Radiotherapy, London, United Kingdom;(2)Cardiff University, School of Engineering, Cardiff, United Kingdom;(3)Imperial College London, Surgery and Cancer, London, United Kingdom;
Show Affiliations
Hide Affiliations
Purpose or Objective
The nodal field-based planning technique in breast radiotherapy was superseded in 2016 by a volume-based technique. This technique requires outlining of nodal volumes, allowing for improved dose conformance to treatment volumes and minimising radiation to surrounding normal tissues. However, volume outlining can betedious and time-consuming, requires expert skills and confer large inter- and intra-user variability1, 2, 3.. Yang described how the use of guidelines improved outlining accuracy, nonetheless inter- and intra-variability still exist due to differences in training, image quality and clinical experience levels3.
The purpose of this work was to evaluate local inter- and intra-user variability in manual outlining for nodal target volume delineation. Accuracy outcomes were expected to demonstrate the quality of local outlining practice;
potentially inform future training needs, and contribute to quality improvement of radiotherapy planning and treatment delivery.
Material and Methods
4 Clinicians were exposed to equal image quality conditions and retrospectively outlined nodal volumes, as well as the brachial plexus on 12 anonymised CT data sets. The clinicians comprised of 2 consultant breast oncologists (DD and EE) and 2 specialist registrars (AA and BB). The European Society for Radiotherapy and Oncology guidelines4 were followed when outlining the level 4 to 1, inter-pectoral and internal mammary nodes. Their outlines were compared to 12 “gold standard outlines” (clinician approved outlines), and the Jaccard coefficient (JCI) calculated to measure conformance to the “gold standard” outlines. A Jaccard coefficient value of 1.0 resembles 100% conformance to the ‘Gold Standard’ outline and 0.8 would be seen as reasonably acceptable5,6.
Results
Manual outlining differences for level 4 and 1 are demonstrated in figure 1. Results show a larger intra-observer range for level 4 than level 1 nodes. Furthermore, the variation of outlined volumes between clinicians is not statistically significant (JCI range 0-56–0.61 and 0.65-0.75 for level 4 and 1 nodes respectively). However, this importantly also demonstrates a universal failure to meet the ‘gold standard’, which is more pronounced with level 4 nodal volumes.
Analysis from the x, y and z coordinates indicates risk of under-dosing the nodal volumes and exposing critical organs. The JCI for nodal 01 outlined by clinician DD was 0.23 (Fig.1). 3D analysis for the same volume demonstrated under-outlining of 1.8cm inferiorly and 1.4cm laterally and at depth. Consequently, the volume is 8.6 cubic centimetre smaller than the “gold standard” which may be of clinical significance (Fig.2).
Conclusion
‘From this study it can be reasoned that in order for this volume-based technique to deliver its predicted clinical improvement, further adjustments are required through amendment of current guidelines and employment of automated and computerised systems’ (or artificial intelligence systems).