Standardising Breast Radiotherapy Structure Naming Conventions: A Machine Learning Approach

Abstract

In progressing the use of big data in health systems, standardised nomenclature is required to enable data pooling and analyses. In many radiotherapy planning systems and their data archives, target volumes (TV) and organ-at-risk (OAR) structure nomenclature has not been standardised. Machine learning (ML) has been utilised to standardise volumes nomenclature in retrospective datasets. However, only subsets of the structures have been targeted. Within this paper, we proposed a new approach for standardising all the structures nomenclature by using multi-modal artificial neural networks. A cohort consisting of 1613 breast cancer patients treated with radiotherapy was identified from Liverpool & Macarthur Cancer Therapy Centres, NSW, Australia. Four types of volume characteristics were generated to represent each target and OAR volume: textual features, geometric features, dosimetry features, and imaging data. Five datasets were created from the original cohort, the first four represented different subsets of volumes and the last one represented the whole list of volumes. For each dataset, 15 sets of combinations of features were generated to investigate the effect of using different characteristics on the standardisation performance. The best model reported 99.416% classification accuracy over the hold-out sample when used to standardise all the nomenclatures in a breast cancer radiotherapy plan into 21 classes. Our results showed that ML based automation methods can be used for standardising naming conventions in a radiotherapy plan taking into consideration the inclusion of multiple modalities to better represent each volume.

End-to-end framework for automated collection of large multicentre radiotherapy datasets demonstrated in a Danish Breast Cancer Group cohort

“…The semi-automatic method used in CORDIAL-RT for categorising structure names required a fair amount of domain knowledge and time. A recent study [8] demonstrated an ML based method for standardising structure nomenclature on 1613 breast cancer patients with promising results. This could make the process faster and less subjective, however the method was only demonstrated in a single centre and was language dependant.…”

“…To address the task of curation, standardisation, and analysis of DICOM files, a vendor-agnostic tool is needed. Tools with standardisation capabilities exist, but these are either single purpose like nomenclature standardisation [8] , focused on dose analysis such as the DVH Analytics package [9] or not open source like the DcmCollab system [10] . While not made for explorative data curation, a system like DcmCollab which focuses on storage, security and GDPR compliance, could however be used as a storage solution after the dataset has been curated.…”

Multimodal Deep Learning Methods on Image and Textual Data to Predict Radiotherapy Structure Names

“…ChemProps [20] was introduced in 2021 for composite polymer name standardization. Furthermore, organ at risk delineation and standardization in radiotherapy were studied in 2021 [21] and 2023 [22,23]. However, for radiotherapy breast structure standardization [23], accuracy was mostly used to evaluate the performances which is biased towards the majority class.…”

Perceptions on and roadblocks to implementation of standardized nomenclature in radiation oncology: A survey from TG‐263U1

“…11 Several groups have published automated solutions for identifying, naming, or relabeling structures in accordance with the TG-263 standard to improve compliance. [12][13][14][15] Another group performed a phased rollout with TG-263 compliant templates that resulted in an increase from 68.2% to 97.8% of compliant structures over a span of 26 months. 11 Although there are several solutions to the roadblocks associated with TG-263 compliance, many challenges remain that may be unique, or shared, by individual clinics and would be valuable to understand prior to the release of the TG-263 update (TG-263U1).…”

“…Several groups have published automated solutions for identifying, naming, or relabeling structures in accordance with the TG‐263 standard to improve compliance 12–15 . Another group performed a phased rollout with TG‐263 compliant templates that resulted in an increase from 68.2% to 97.8% of compliant structures over a span of 26 months 11 .…”