Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Randomized Controlled Trial
. 2022 Jan;11(1):166-175.
doi: 10.1002/cam4.4441. Epub 2021 Nov 23.

A blind randomized validated convolutional neural network for auto-segmentation of clinical target volume in rectal cancer patients receiving neoadjuvant radiotherapy

Yijun Wu  1 , Kai Kang  1 , Chang Han  1 , Shaobin Wang  2 , Qi Chen  2 , Yu Chen  2 , Fuquan Zhang  1 , Zhikai Liu  1
Affiliations
Randomized Controlled Trial

A blind randomized validated convolutional neural network for auto-segmentation of clinical target volume in rectal cancer patients receiving neoadjuvant radiotherapy

Yijun Wu et al. Cancer Med. 2022 Jan.

Abstract

Background: Delineation of clinical target volume (CTV) for radiotherapy is a time-consuming and labor-intensive work. This study aims to propose a novel convolutional neural network (CNN)-based model for fast auto-segmentation of CTV. To evaluate its performance and clinical utility, a blind randomized validation method was used.

Methods: Our proposed model was based on the generally accepted U-Net architecture using computed tomography slices with CTV contours delineated by experienced radiation clinicians from 135 rectal patients receiving neoadjuvant radiotherapy. The Dice similarity coefficient (DSC) and 95th percentile Hausdorff distance (95HD) were used to measure segmentation performance. The validated dataset of additional 20 patients for clinical evaluation by 10 experienced oncology clinicians from 7 centers was randomly and blindly divided into two groups for clinicians' scoring and Turing test, respectively. Second evaluation was performed with different randomization after 2 weeks.

Results: The mean DSC and 95HD values of the proposed model were 0.90 ± 0.02 and 8.11 ± 1.93 mm for CTV of rectal cancer patients, respectively. The average time for automatic segmentation in the validation groups was 15 s per patient. By clinicians' scoring, the AI model performed better than manually delineating, though the differences were not significant (Week 0: 2.59 vs. 2.52, p = 0.086; Week 2: 2.55 vs. 2.47, p = 0.115). Additionally, the mean positive rates in the Turing test were 40.5% in Week 0 and 45.2% in Week 2, which demonstrated the great intelligence of our model.

Conclusions: Our proposed model can be used clinically for assisting contouring of CTVs in rectal cancer patients receiving neoadjuvant radiotherapy, which improves the efficiency and consistency of radiation clinicians' work.

Keywords: clinical evaluation; convolutional neural network; deep learning; neoadjuvant radiotherapy; rectal cancer.

PubMed Disclaimer

Conflict of interest statement

Shaobin Wang, Qi Chen, and Yu Chen were employed by the company MedMind Technology Co., Ltd. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Figures

FIGURE 1
FIGURE 1
Development of our U‐Net architecture. En is encoder block. Down and Conv are down sample and convolution layers. WEN is connection weight responding to a specific connection
FIGURE 2
FIGURE 2
Flow chart for clinicians' scoring and Turing test. AI, artificial intelligence; GT, ground truth
FIGURE 3
FIGURE 3
Sample contours for clinicians' scoring in patients with rectal cancer receiving neoadjuvant radiotherapy. Rejected (0 point): (A, B); Major revision (1 point): (C, D); Minor revision (2 points): (E–H); and Totally accepted (3 points): (I–L)
FIGURE 4
FIGURE 4
Sample slices for Turing test in patients with rectal cancer receiving neoadjuvant radiotherapy. Red: artificial intelligence (AI) contour; Green: ground truth (GT) contour. (A–D): AI performs better than GT; (E–H): AT performs worse than GT
FIGURE 5
FIGURE 5
(A, B) Mean score of 10 clinicians for each contour. (C, D) Mean score of 10 clinicians for each slice in Turing test. AI, artificial intelligence; GT, ground truth
See this image and copyright information in PMC

References

    1. Sung H, Ferlay J, Siegel R, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209–249. - PubMed
    1. Folkesson J, Birgisson H, Pahlman L, Cedermark B, Glimelius B, Gunnarsson U. Swedish Rectal Cancer Trial: long lasting benefits from radiotherapy on survival and local recurrence rate. J Clin Oncol. 2005;23(24):5644‐5650. - PubMed
    1. Myerson RJ, Garofalo MC, El Naqa I, et al. Elective clinical target volumes for conformal therapy in anorectal cancer: a radiation therapy oncology group consensus panel contouring atlas. Int J Radiat Oncol Biol Phys. 2009;74(3):824‐830. - PMC - PubMed
    1. Ng M, Leong T, Chander S, et al. Australasian Gastrointestinal Trials Group (AGITG) contouring atlas and planning guidelines for intensity‐modulated radiotherapy in anal cancer. Int J Radiat Oncol Biol Phys. 2012;83(5):1455‐1462. - PubMed
    1. Roels S, Duthoy W, Haustermans K, et al. Definition and delineation of the clinical target volume for rectal cancer. Int J Radiat Oncol Biol Phys. 2006;65(4):1129‐1142. - PubMed

Publication types

MeSH terms

LinkOut - more resources