Elsevier

Radiotherapy and Oncology

Volume 153, December 2020, Pages 139-145
Radiotherapy and Oncology

Original Article
Clinical evaluation of atlas- and deep learning-based automatic segmentation of multiple organs and clinical target volumes for breast cancer

https://doi.org/10.1016/j.radonc.2020.09.045Get rights and content

Highlights

  • The DLBAS model performed better for the majority of CTVs and OARs in comparison to ABAS solutions.

  • ABAS performed relatively well for structures with distinct boundaries.

  • DLBAS showed more robustness and less dependency on input data.

  • DLBAS has great potential to help optimise the radiation therapy planning workflow.

Abstract

Manual segmentation is the gold standard method for radiation therapy planning; however, it is time-consuming and prone to inter- and intra-observer variation, giving rise to interests in auto-segmentation methods. We evaluated the feasibility of deep learning-based auto-segmentation (DLBAS) in comparison to commercially available atlas-based segmentation solutions (ABAS) for breast cancer radiation therapy.

This study used contrast-enhanced planning computed tomography scans from 62 patients with breast cancer who underwent breast-conservation surgery. Contours of target volumes (CTVs), organs, and heart substructures were generated using two commercial ABAS solutions and DLBAS using fully convolutional DenseNet. The accuracy of the segmentation was assessed using 14 test patients using the Dice Similarity Coefficient and Hausdorff Distance referencing the expert contours. A sensitivity analysis was performed using non-contrast planning CT from 14 additional patients.

Compared to ABAS, the proposed DLBAS model yielded more consistent results and the highest average Dice Similarity Coefficient values and lowest Hausdorff Distances, especially for CTVs and the substructures of the heart. ABAS showed limited performance in soft-tissue-based regions, such as the esophagus, cardiac arteries, and smaller CTVs. The results of sensitivity analysis between contrast and non-contrast CT test sets showed little difference in the performance of DLBAS and conversely, a large discrepancy for ABAS.

The proposed DLBAS algorithm was more consistent and robust in its performance than ABAS across the majority of structures when examining both CTVs and normal organs. DLBAS has great potential to aid a key process in the radiation therapy workflow, helping optimise and reduce the clinical workload.

Section snippets

Data

Sixty-two breast cancer patients, with a mix of left- and right-sided breast cancer, who received RT after breast-conserving surgery between May 1st, 2016 and May 1st, 2019 and underwent contrast-enhanced planning computed tomography (CT) were selected for this study after Institutional Review Board approval. The CT scans were acquired on a Siemens Sensation Open (Siemens, Forchheim, Germany) and a Toshiba Aquilion (Toshiba Medical Systems, Japan), using the following common CT image

Results

A typical example of an axial CT scan segmentation produced from DLBAS and ABAS can be seen in Supplementary Video 1. Fig. 2A shows an example of CTV auto-segmentation from ABAS and DLBAS. Among 14 CTV structures, DLBAS produced the highest average DSCs in 11 of them. The statistical test reflects that these differences were significant for left and right AXL3 and IMN (Supplementary Table 1). However, the HD95 comparisons of CTVs reveal that DLBAS produced smaller surface discrepancies compared

Discussion

To the best of our knowledge, this is the first study that compares the performance of ABAS and DLBAS methods for breast cancer RT planning that includes node regions and heart structures. In this study, we demonstrated the efficacy of our DLBAS by measuring the performance across a range of structures (CTVs, OARs, and heart substructures) in a head-to-head comparison study with commercial ABAS solutions. Our results indicate that, while ABAS offered an acceptable performance level, DLBAS

Conflict of Interest

J.S.K is cofounder of Oncosoft and serves as an advisor to Rayence. This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. 2020R1A4A1016619).

References (45)

  • T.J. Whelan

    Regional nodal irradiation in early-stage breast cancer

    N Engl J Med

    (2015)
  • L.B.J. Thorsen

    DBCG-IMN: A population-based cohort study on the effect of internal mammary node irradiation in early node-positive breast cancer

    J Clin Oncol

    (2016)
  • J.P. Gross

    Radiation therapy field design and lymphedema risk after regional nodal irradiation for breast cancer

    Int J Radiat Oncol Biol Phys

    (2018)
  • H.K. Byun et al.

    Risk of lymphedema following contemporary treatment for breast cancer: an analysis of 7,426 consecutive patients from a multidisciplinary perspective

    Int J Radiat Oncol

    (2019)
  • B.M. Lee et al.

    Hypofractionated radiotherapy dose scheme and application of new techniques are associated to a lower incidence of radiation pneumonitis in breast cancer patients

    Front Oncol

    (2020)
  • B.M. Lee et al.

    Risk of radiation pneumonitis following individualized modern radiation therapy with IMRT, a breath-holding technique and prone positioning for breast cancer

    Int J Radiat Oncol

    (2018)
  • S.H. Choi et al.

    The effect of regional node irradiation on the thyroid gland in the breast cancer patients: the clinical significance of optimization of radiation target volume

    Int J Radiat Oncol

    (2019)
  • V.A.B. van den Bogaard

    Development and evaluation of an auto-segmentation tool for the left anterior descending coronary artery of breast cancer patients based on anatomical landmarks

    Radiother Oncol

    (2019)
  • White J. et al., Breast cancer atlas for radiation therapy planning: consensus definitions....
  • X.A. Li

    Variability of target and normal structure delineation for breast cancer radiotherapy: an RTOG multi-institutional and multiobserver study

    Int J Radiat Oncol Biol Phys

    (2009)
  • H. Lee

    Clinical evaluation of commercial atlas-based auto-segmentation in the head and neck region

    Front Oncol

    (2019)
  • M. La Macchia

    Systematic evaluation of three different commercial software solutions for automatic segmentation for adaptive therapy in head-and-neck, prostate and pleural cancer

    Radiat Oncol

    (2012)
  • Cited by (59)

    View all citing articles on Scopus

    We would like to express our gratitude to MIM Software and Mirada Medical for granting us access to their software and for their assistance in this study.

    View full text