Evaluation of Deep Learning to Augment Image-Guided Radiotherapy for Head and Neck and Prostate Cancers | Head and Neck Cancer | JAMA Network Open | JAMA Network
[Skip to Navigation]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address 18.207.129.82. Please contact the publisher to request reinstatement.
1.
Pan  HY, Haffty  BG, Falit  BP,  et al.  Supply and demand for radiation oncology in the United States: updated projections for 2015 to 2025.   Int J Radiat Oncol Biol Phys. 2016;96(3):493-500. doi:10.1016/j.ijrobp.2016.02.064PubMedGoogle ScholarCrossref
2.
Sklan  A, Collingridge  D.  Treating head and neck cancer: for better or for worse?   Lancet Oncol. 2017;18(5):570-571. doi:10.1016/S1470-2045(17)30269-3PubMedGoogle ScholarCrossref
3.
Barnett  GC, West  CM, Dunning  AM,  et al.  Normal tissue reactions to radiotherapy: towards tailoring treatment dose by genotype.   Nat Rev Cancer. 2009;9(2):134-142. doi:10.1038/nrc2587PubMedGoogle ScholarCrossref
4.
Vorwerk  H, Zink  K, Schiller  R,  et al.  Protection of quality and innovation in radiation oncology: the prospective multicenter trial the German Society of Radiation Oncology (DEGRO-QUIRO study).   Strahlenther Onkol. 2014;190(5):433-443. doi:10.1007/s00066-014-0634-0PubMedGoogle ScholarCrossref
5.
Cazzaniga  LF, Marinoni  MA, Bossi  A,  et al.  Interphysician variability in defining the planning target volume in the irradiation of prostate and seminal vesicles.   Radiother Oncol. 1998;47(3):293-296. doi:10.1016/S0167-8140(98)00028-0PubMedGoogle ScholarCrossref
6.
Cooper  JS, Mukherji  SK, Toledano  AY,  et al.  An evaluation of the variability of tumor-shape definition derived by experienced observers from CT images of supraglottic carcinomas (ACRIN protocol 6658).   Int J Radiat Oncol Biol Phys. 2007;67(4):972-975. doi:10.1016/j.ijrobp.2006.10.029PubMedGoogle ScholarCrossref
7.
Dubois  DF, Prestidge  BR, Hotchkiss  LA, Prete  JJ, Bice  WS  Jr.  Intraobserver and interobserver variability of MR imaging- and CT-derived prostate volumes after transperineal interstitial permanent prostate brachytherapy.   Radiology. 1998;207(3):785-789. doi:10.1148/radiology.207.3.9609905PubMedGoogle ScholarCrossref
8.
Fiorino  C, Reni  M, Bolognesi  A, Cattaneo  GM, Calandrino  R.  Intra- and inter-observer variability in contouring prostate and seminal vesicles: implications for conformal treatment planning.   Radiother Oncol. 1998;47(3):285-292. doi:10.1016/S0167-8140(98)00021-8PubMedGoogle ScholarCrossref
9.
Fotina  I, Lütgendorf-Caucig  C, Stock  M, Pötter  R, Georg  D.  Critical discussion of evaluation parameters for inter-observer variability in target definition for radiation therapy.   Strahlenther Onkol. 2012;188(2):160-167. doi:10.1007/s00066-011-0027-6PubMedGoogle ScholarCrossref
10.
Gardner  SJ, Wen  N, Kim  J,  et al.  Contouring variability of human- and deformable-generated contours in radiotherapy for prostate cancer.   Phys Med Biol. 2015;60(11):4429-4447. doi:10.1088/0031-9155/60/11/4429PubMedGoogle ScholarCrossref
11.
Valicenti  RK, Sweet  JW, Hauck  WW,  et al.  Variation of clinical target volume definition in three-dimensional conformal radiation therapy for prostate cancer.   Int J Radiat Oncol Biol Phys. 1999;44(4):931-935. doi:10.1016/s0360-3016(99)00090-5PubMedGoogle ScholarCrossref
12.
Ohri  N, Shen  X, Dicker  AP, Doyle  LA, Harrison  AS, Showalter  TN.  Radiotherapy protocol deviations and clinical outcomes: a meta-analysis of cooperative group clinical trials.   J Natl Cancer Inst. 2013;105(6):387-393. doi:10.1093/jnci/djt001PubMedGoogle ScholarCrossref
13.
Balagopal  A, Kazemifar  S, Nguyen  D,  et al.  Fully automated organ segmentation in male pelvic CT images.   Phys Med Biol. 2018;63(24):245015. doi:10.1088/1361-6560/aaf11cPubMedGoogle Scholar
14.
Lou  B, Doken  S, Zhuang  T,  et al.  An image-based deep learning framework for individualising radiotherapy dose.   Lancet Digit Health. 2019;1(3):e136-e147. doi:10.1016/S2589-7500(19)30058-5PubMedGoogle ScholarCrossref
15.
Nikolov  S, Blackwell  S, Zverovitch  A,  et al. Deep learning to achieve clinically applicable segmentation of head and neck anatomy for radiotherapy. arXiv. Preprint published online September 12, 2018. Accessed October 29, 2020. https://arxiv.org/pdf/ 1809.04430.pdf
16.
Wang  S, He  K, Nie  D, Zhou  S, Gao  Y, Shen  D.  CT male pelvic organ segmentation using fully convolutional networks with boundary sensitive representation.   Med Image Anal. 2019;54:168-178. doi:10.1016/j.media.2019.03.003PubMedGoogle ScholarCrossref
17.
Ronneberger  O, Fischer  P, Brox  T. U-net: convolutional networks for biomedical image segmentation. In: Navab  N, Hornegger  J, Wells  W, Frangi  A, eds.  Medical Image Computing and Computer-Assisted Intervention—MICCAI 2015. Lecture Notes in Computer Science, vol 9351. Springer; 2015. doi:10.1007/978-3-319-24574-4_28
18.
Brouwer  CL, Steenbakkers  RJ, Bourhis  J,  et al.  CT-based delineation of organs at risk in the head and neck region: DAHANCA, EORTC, GORTEC, HKNPCSG, NCIC CTG, NCRI, NRG Oncology and TROG consensus guidelines.   Radiother Oncol. 2015;117(1):83-90. doi:10.1016/j.radonc.2015.07.041PubMedGoogle ScholarCrossref
19.
TCIA CT scan dataset. Accessed October 29, 2020. https://github.com/deepmind/tcia-ct-scan-dataset
20.
Cancer Imaging Archive. Head-neck cetuximab. Published June 3, 2020. Accessed October 29, 2020. https://wiki.cancerimagingarchive.net/display/Public/Head-Neck+Cetuximab
21.
Dice  LR.  Measures of the amount of ecologic association between species.  In:  Ecology. 1945;26(3):297–302. doi:10.2307/1932409Google ScholarCrossref
22.
Lin  L, Hedayat  AS, Wu W.  Statistical Tools for Measuring Agreement. Springer Science & Business Media, 2012."https://doi.org/10.1007/978-1-4614-0562-7"
23.
Bland  JM, Altman  DG.  Statistical methods for assessing agreement between two methods of clinical measurement.   Lancet. 1986;1(8476):307-310. doi:10.1016/S0140-6736(86)90837-8PubMedGoogle ScholarCrossref
24.
Mann  HB, Whitney  DR.  On a test of whether one of two random variables is stochastically larger than the other.   Ann Math Stat. 1947;18(1):50-60. doi:10.1214/aoms/1177730491Google ScholarCrossref
25.
Zhu  W, Huang  Y, Zeng  L,  et al.  AnatomyNet: deep learning for fast and fully automated whole-volume segmentation of head and neck anatomy.   Med Phys. 2019;46(2):576-589. doi:10.1002/mp.13300PubMedGoogle ScholarCrossref
26.
Wang  Yueyue  et al  Organ at risk segmentation in head and neck CT images using a two-stage segmentation framework based on 3D U-net.   IEEE Access. 2019;7:144591–144602. doi:10.1109/ACCESS.2019.2944958Google ScholarCrossref
27.
Raudaschl  PF, Zaffino  P, Sharp  GC,  et al.  Evaluation of segmentation methods on head and neck CT: auto-segmentation challenge 2015.   Med Phys. 2017;44(5):2020-2036. doi:10.1002/mp.12197PubMedGoogle ScholarCrossref
28.
InnerEye Deep Learning. Accessed October 29, 2020. https://github.com/microsoft/InnerEye-DeepLearning/
29.
Barrett  JF, Keat  N.  Artifacts in CT: recognition and avoidance.   Radiographics. 2004;24(6):1679-1691. doi:10.1148/rg.246045065PubMedGoogle ScholarCrossref
30.
Tarroni  G, Oktay  O, Bai  W,  et al.  Learning-based quality control for cardiac MR IMAGES.   IEEE Trans Med Imaging. 2019;38(5):1127-1138. doi:10.1109/TMI.2018.2878509PubMedGoogle ScholarCrossref
31.
Sonke  JJ, Aznar  M, Rasch  C.  Adaptive radiotherapy for anatomical changes.   Semin Radiat Oncol. 2019;29(3):245-257. doi:10.1016/j.semradonc.2019.02.007PubMedGoogle ScholarCrossref
32.
Heukelom  J, Fuller  CD.  Head and neck cancer adaptive radiation therapy (ART): conceptual considerations for the informed clinician.   Semin Radiat Oncol. 2019;29(3):258-273. doi:10.1016/j.semradonc.2019.02.008Google Scholar
33.
Lagendijk  JJW, Raaymakers  BW, van Vulpen  M.  The magnetic resonance imaging-linac system.   Semin Radiat Oncol. 2014;24(3):207-209. doi:10.1016/j.semradonc.2014.02.009PubMedGoogle ScholarCrossref
34.
Valenzuela  W, Ferguson  SJ, Ignasiak  D,  et al.  FISICO: fast image segmentation correction.   PLoS One. 2016;11(5):e0156035. doi:10.1371/journal.pone.0156035PubMedGoogle Scholar
35.
Acosta  O.,, Dowling  J.,, Drean  G.,  et al Multi-atlas-based segmentation of pelvic structures from CT scans for planning in prostate cancer radiotherapy. In: El-Baz  AS, Saba  L, Suri  JS, eds.  Abdomen and Thoracic Imaging. Springer, 2014:623–656. doi:10.1007/978-1-4614-8498-1_24
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Original Investigation
    Health Informatics
    November 30, 2020

    Evaluation of Deep Learning to Augment Image-Guided Radiotherapy for Head and Neck and Prostate Cancers

    Author Affiliations
    • 1Health Intelligence, Microsoft Research, Cambridge, United Kingdom
    • 2Department of Oncology, Cambridge University Hospitals NHS Foundation Trust, United Kingdom
    • 3now with Edinburgh Cancer Centre, Western General Hospital, Edinburgh, United Kingdom
    JAMA Netw Open. 2020;3(11):e2027426. doi:10.1001/jamanetworkopen.2020.27426
    Key Points

    Question  Can machine learning models achieve clinically acceptable accuracy in image segmentation tasks in radiotherapy planning and reduce overall contouring time?

    Findings  This quality improvement study was conducted on a set of 242 head and neck and 519 pelvic computed tomography scans acquired for radiotherapy planning at 8 distinct clinical sites with heterogeneous population groups and image acquisition settings. The proposed technology achieved levels of accuracy within interexpert variability; statistical agreement was observed for 13 of 15 structures while reducing the annotation time by a mean of 93% per scan.

    Meaning  The study findings highlight the opportunity for widespread adoption of autosegmentation models in radiotherapy workflows to reduce overall contouring and planning time.

    Abstract

    Importance  Personalized radiotherapy planning depends on high-quality delineation of target tumors and surrounding organs at risk (OARs). This process puts additional time burdens on oncologists and introduces variability among both experts and institutions.

    Objective  To explore clinically acceptable autocontouring solutions that can be integrated into existing workflows and used in different domains of radiotherapy.

    Design, Setting, and Participants  This quality improvement study used a multicenter imaging data set comprising 519 pelvic and 242 head and neck computed tomography (CT) scans from 8 distinct clinical sites and patients diagnosed either with prostate or head and neck cancer. The scans were acquired as part of treatment dose planning from patients who received intensity-modulated radiation therapy between October 2013 and February 2020. Fifteen different OARs were manually annotated by expert readers and radiation oncologists. The models were trained on a subset of the data set to automatically delineate OARs and evaluated on both internal and external data sets. Data analysis was conducted October 2019 to September 2020.

    Main Outcomes and Measures  The autocontouring solution was evaluated on external data sets, and its accuracy was quantified with volumetric agreement and surface distance measures. Models were benchmarked against expert annotations in an interobserver variability (IOV) study. Clinical utility was evaluated by measuring time spent on manual corrections and annotations from scratch.

    Results  A total of 519 participants’ (519 [100%] men; 390 [75%] aged 62-75 years) pelvic CT images and 242 participants’ (184 [76%] men; 194 [80%] aged 50-73 years) head and neck CT images were included. The models achieved levels of clinical accuracy within the bounds of expert IOV for 13 of 15 structures (eg, left femur, κ = 0.982; brainstem, κ = 0.806) and performed consistently well across both external and internal data sets (eg, mean [SD] Dice score for left femur, internal vs external data sets: 98.52% [0.50] vs 98.04% [1.02]; P = .04). The correction time of autogenerated contours on 10 head and neck and 10 prostate scans was measured as a mean of 4.98 (95% CI, 4.44-5.52) min/scan and 3.40 (95% CI, 1.60-5.20) min/scan, respectively, to ensure clinically accepted accuracy. Manual segmentation of the head and neck took a mean 86.75 (95% CI, 75.21-92.29) min/scan for an expert reader and 73.25 (95% CI, 68.68-77.82) min/scan for a radiation oncologist. The autogenerated contours represented a 93% reduction in time.

    Conclusions and Relevance  In this study, the models achieved levels of clinical accuracy within expert IOV while reducing manual contouring time and performing consistently well across previously unseen heterogeneous data sets. With the availability of open-source libraries and reliable performance, this creates significant opportunities for the transformation of radiation treatment planning.

    ×