1.
National Comprehensive Cancer Network Inc NCCN Clinical Practice Guidelines in
Oncology. Head and Neck Cancers version 2. 2020. Available from URL:
https://www.nccn.org/professionals/physician_gls/pdf/head-and-neck.pdf
2.
Machiels JP, René Leemans C, Golusinski W, Grau C, Licitra L, Gregoire V; EHNS
Executive
Board.
Electronic
address:
secretariat@ehns.org;
ESMO
Guidelines
Committee. Electronic address: clinicalguidelines@esmo.org; ESTRO Executive Board.
Electronic address: info@estro.org. Squamous cell carcinoma of the oral cavity, larynx,
oropharynx and hypopharynx: EHNS-ESMO-ESTRO Clinical Practice Guidelines for
diagnosis, treatment and follow-up. Ann Oncol. 2020; 31: 1462–1475.
3.
Imai T, Satoh I, Matsumoto K, et al. Retrospective observational study of occult cervical
lymph-node metastasis in T1N0 tongue cancer. Jpn J Clin Oncol. 2017; 47: 130–136.
4.
Shibuya H, Hoshina M, Takeda M, Matsumoto S, Suzuki S, Okada N. Brachytherapy for
stage I & II oral tongue cancer: an analysis of past cases focusing on control and
complications. Int J Radiat Oncol Biol Phys. 1993; 26: 51–58.
5.
Fujita M, Hirokawa Y, Kashiwado K, et al. Interstitial brachytherapy for stage I and II
squamous cell carcinoma of the oral tongue: factors influencing local control and soft
tissue complications. Int J Radiat Oncol Biol Phys. 1999; 44: 767–775.
6.
Fasunla AJ, Greene BH, Timmesfeld N, Wiegand S, Werner JA, Sesterhenn AM. A
meta-analysis of the randomized controlled trials on elective neck dissection versus
therapeutic neck dissection in oral cavity cancers with clinically node-negative neck.
Oral Oncol. 2011; 47: 320–324.
7.
D'Cruz AK, Vaish R, Kapre N, et al. Elective versus Therapeutic Neck Dissection in
Node-Negative Oral Cancer. N Engl J Med. 2015; 373: 521–529.
8.
Lambin P, Rios-Velazquez E, Leijenaar R, et al. Radiomics: extracting more information
from medical images using advanced feature analysis. Eur J Cancer. 2012; 48: 441–446.
9.
Thawani R, McLane M, Beig N, et al. Radiomics and radiogenomics in lung cancer: A
review for the clinician. Lung Cancer. 2018; 115: 34–41.
10. Valdora F, Houssami N, Rossi F, Calabrese M, Tagliafico AS. Rapid review: radiomics
and breast cancer. Breast Cancer Res Treat. 2018; 169: 217–229.
11. Peng Z, Wang Y, Wang Y, et al. Application of radiomics and machine learning in head
and neck cancers. Int J Biol Sci. 2021; 17: 475–486.
12. Giraud P, Giraud P, Gasnier A, et al. Radiomics and Machine Learning for Radiotherapy
in Head and Neck Cancers. Front Oncol. 2019; 9: 174.
13. Elhalawani H, Lin TA, Volpe S, et al. Machine Learning Applications in Head and Neck
Radiation Oncology: Lessons From Open-Source Radiomics Challenges. Front Oncol.
2018; 8: 294.
14. Forghani R, Chatterjee A, Reinhold C, et al. Head and neck squamous cell carcinoma:
prediction of cervical lymph node metastasis by dual-energy CT texture analysis with
machine learning. Eur Radiol. 2019; 29: 6172–6181.
15. Yuan Y, Ren J, Tao X. Machine learning-based MRI texture analysis to predict occult
lymph node metastasis in early-stage oral tongue squamous cell carcinoma. Eur Radiol.
2021; 31: 6429–6437.
16. Zhong YW, Jiang Y, Dong S, et al. Tumor radiomics signature for artificial neural
network-assisted detection of neck metastasis in patient with tongue cancer. J
Neuroradiol. 2021 Aug 4. doi: 10.1016/j.neurad.2021.07.006.
17. Grégoire V, Ang K, Budach W, et al. Delineation of the neck node levels for head and
neck tumors: a 2013 update. DAHANCA, EORTC, HKNPCSG, NCIC CTG, NCRI,
RTOG, TROG consensus guidelines. Radiother Oncol. 2014; 110: 172–181.
18. Haga A, Takahashi W, Aoki S, et al. Standardization of imaging features for radiomics
analysis. J Med Invest. 2019; 66: 35–37.
19. van Griethuysen JJM, Fedorov A, Parmar C, et al. Computational Radiomics system to
decode the radiographic phenotype. Cancer Res. 2017; 77: e104.
20. Tibshirani R. The lasso method for variable selection in the Cox model. Stat Med. 1997;
16: 385–395.
21. Zhang JX, Song W, Chen ZH, et al. Prognostic and predictive value of a microRNA
signature in stage II colon cancer: a microRNA expression analysis. Lancet Oncol. 2013;
14: 1295 –1306.
22. Zhang Z. Introduction to machine learning: k-nearest neighbors. Ann Transl Med. 2016;
4: 218.
23. Cortes C, Vapnik V. Support-vector networks. Mach Learn. 1995; 20: 273–297.
24. Bel L, Allard D, Laurent JM, Cheddadi R, Bar-Hen A. CART algorithm for spatial data:
application to environmental and ecological data. Comput. Stat. Data Anal. 2009; 53:
3082–3093.
25. Breiman L. Random Forests. Mach Learn. 2001; 45: 5–32.
26. Hastie T, Rosset S, Zhu J, Zou H. Multi-class adaboost. Stat. Its Interface. 2009; 2: 349–
360.
27. Chawla NV, Bowyer KW, Hall LO, Kegelmeyer WP. SMOTE: synthetic minority
over-sampling technique. J Artif Intell Res. 2002; 16: 321–357.
28. Bulsara VM, Worthington HV, Glenny AM, Clarkson JE, Conway DI, Macluskey M.
Interventions for the treatment of oral and oropharyngeal cancers: surgical treatment.
Cochrane Database Syst Rev. 2018; 12: CD006205.
29. Bur AM, Holcomb A, Goodwin S, et al. Machine learning to predict occult nodal
metastasis in early oral squamous cell carcinoma. Oral Oncol. 2019; 92: 20–25.
30. Shan J, Jiang R, Chen X, et al. Machine Learning Predicts Lymph Node Metastasis in
Early-Stage Oral Tongue Squamous Cell Carcinoma. J Oral Maxillofac Surg. 2020; 78:
2208–2218.
31. Kwak MS, Eun YG, Lee JW, et al. Development of a machine learning model for the
prediction of nodal metastasis in early T classification oral squamous cell carcinoma:
SEER-based population study. Head Neck. 2021; 43: 2316–2324.
32. Matos LL, Manfro G, Santos RV, et al. Tumor thickness as a predictive factor of lymph
node metastasis and disease recurrence in T1N0 and T2N0 squamous cell carcinoma of
the oral tongue. Oral Surg Oral Med Oral Pathol Oral Radiol. 2014; 118: 209–217.
33. Haraguchi K, Yoshiga D, Oda M, et al. Depth of invasion determined by magnetic
resonance imaging in tongue cancer can be a predictor of cervical lymph node metastasis.
Oral Surg Oral Med Oral Pathol Oral Radiol. 2021; 131: 231–240.
34. Baba A, Masuda K, Hashimoto K, et al. Correlation between the magnetic resonance
imaging features of squamous cell carcinoma of the buccal mucosa and pathologic depth
of invasion. Oral Surg Oral Med Oral Pathol Oral Radiol. 2021; 131: 582–590.
35. Sun T, Zhang R, Wang J, Li X, Guo X. Computer-aided diagnosis for early-stage lung
cancer based on longitudinal and balanced data. PLoS One. 2013; 8: e63559.
36. Yan S, Qian W, Guan Y, Zheng B. Improving lung cancer prognosis assessment by
incorporating synthetic minority oversampling technique and score fusion method. Med
Phys. 2016; 43: 2694–2703.
Figure legends
Figure 1. Contouring of each neck node level.
Blue, green, red, yellow, and purple lines show levels Ia, Ib, II, III, and IVa, respectively, in
the ipsilateral neck side. The total neck node level consisted of these levels.
Figure 2. Data labeling.
a) Analysis in total neck node level. The total neck node level for each patient was labeled
“positive” or “negative” for occult cervical lymph node metastasis (OCLNM) based on the
presence of at least one metastasis in any level.
b) Analysis in each neck node level. Each neck node level was labeled individually as
“positive” or “negative” for OCLNM based on the presence of at least one metastasis in that
level.
Figure 3. Data analysis workflow.
CT, computed tomography; LASSO, least absolute shrinkage and selection operator; kNN,
k-nearest neighbor; SVM, support vector machine; CART, classification and regression trees;
RF, random forest; Ada, Ada Boost; SMOTE, synthetic minority oversampling technique.
Figure 4. Analysis of the total neck node level. Receiver operator characteristic (ROC) curves
with accuracy and an area under the curve (AUC) scores for each model with or without the
synthetic minority oversampling technique (SMOTE).
a) k-nearest neighbor (kNN)
b) Support vector machine (SVM)
c) Classification and regression trees (CART)
d) Random forest (RF)
e) Ada boost (Ada)
Figure 5. Analysis of each neck node level. Receiver operator characteristic (ROC) curves
with accuracy and an area under the curve (AUC) scores for each model with or without the
synthetic minority oversampling technique (SMOTE).
a) k-nearest neighbor (kNN)
b) Support vector machine (SVM)
c) Classification and regression trees (CART)
d) Random forest (RF)
e) Ada boost (Ada)
Table 1. Patient and tumor characteristics.
N = 161
100%
65 (22–91)
Male
80
49.7
Female
81
50.3
Right
83
51.6
Left
78
48.4
Tis
28
17.4
T1
45
28.0
T2
81
50.3
T3
4.3
Yes
38
23.6
No
122
75.8
0.6
Yes
75
46.6
No
73
45.3
Not available
13
8.1
Surgery
102
63.4
Brachytherapy
59
36.6
Yes
46
28.6
No
115
71.4
Age, years, median (range)
Sex
Location
Clinical T category
Ulcer
Not available
Smoking history
Treatment
Occult neck metastasis
Level of occult neck metastasis
N = 46
Ia
6.5
Ib
24
52.2
II
25
54.3
III
17.4
6.5
IVa
Texture analysis feature
Wavelet-HLL_glcm_Imc1
Wavelet-HLL_glcm_MaximumProbability
Wavelet-HHL_glcm_Correlation
Wavelet-HHH_glcm_ClusterShade
Wavelet-LLL_glcm_Imc1
Wavelet-HLH_glszm_SmallAreaLowGrayLevelEmphasis
Wavelet-HLH_glszm_GrayLevelNonUniformityNormalized
Wavelet-HLH_glszm_GrayLevelNonUniformity
Wavelet-LLH_glrlm_RunLengthNonUniformity
Original_shape_Maximum2DDiameterRow
matrix; glszm, gray level size zone matrix.
LASSO, least absolute shrinkage and selection operator; glcm, gray level co-occurrence matrix; Imc1, informational measure of correlation 1; glrlm, gray level run length
Texture analysis feature
Shape-based feature
Wavelet-LHH_firstorder_Kurtosis
Original_shape_Maximum2DDiameterSlice
First-order statistical feature Wavelet-HHL_firstorder_Median
Original_firstorder_Median
First-order statistical feature
Each neck node level
Total neck node level
Extracted Features after LASSO logistic regression analysis
Table 2. Extracted features after the least absolute shrinkage and selection of operator logistic regression analysis.
Precision
Accuracy
0.63
0.71
0.59
0.67
0.64
0.71
0.80
0.74
0.85
0.81
Model
kNN
SVM
CART
RF
Ada
kNN + SMOTE
SVM + SMOTE
CART + SMOTE
RF + SMOTE
Ada + SMOTE
0.77
0.82
0.70
0.76
0.68
0.60
0.63
0.56
0.69
0.59
Recall
0.90
0.92
0.77
0.84
0.78
0.67
0.75
0.61
0.72
0.60
AUC
Ada + SMOTE
RF + SMOTE
CART + SMOTE
SVM + SMOTE
kNN + SMOTE
Ada
RF
CART
SVM
kNN
Model
Each neck node level
0.86
0.92
0.75
0.96
0.75
0.65
0.70
0.58
0.55
0.42
Accuracy
0.91
0.95
0.77
0.96
0.79
0.85
0.84
0.85
0.85
0.83
Precision
0.83
0.89
0.73
0.95
0.73
0.60
0.66
0.55
0.53
0.45
Recall
random forest; Ada, Ada Boost; SMOTE, synthetic minority oversampling technique
AUC, area under the curve; kNN, k-nearest neighbor; SVM, support vector machine; CART, classification and regression trees; RF,
0.88
0.88
0.83
0.86
0.78
0.79
0.78
0.77
0.77
0.79
Total neck node level
Table 3. Prediction performance of all models with or without the synthetic minority oversampling technique.
0.93
0.97
0.82
0.98
0.80
0.65
0.65
0.58
0.54
0.61
AUC
782
145
16878
64
116
1402
161
Bur AM et al29
Shan J et al30
Kwak MS et al31
Forghani R et al14
Yuan Y et al15
Zhong YW et al16
This study
CT (neck node level)
CT (primary tumor)
MRI (primary tumor)
CT (primary tumor)
Clinicopathologic data
Clinicopathologic data
Clinicopathologic data
Data
SVM + SMOTE
cNRAD99
Naïve Bayes
Random forest
XGBoost
SVM
Decision forest
Model
0.96
0.88
0.74
0.88
0.95
NA
NA
Accuracy
0.98
0.94
0.80
NA
0.96
0.96
0.84
AUC
0.96
0.85
0.63
0.91
0.92
Sensitivity
imaging; SMOTE, synthetic minority oversampling technique
0.95
0.91
0.82
0.67
0.86
0.88
0.58
Specificity
AUC, area under the curve; NA, not available; SVM, support vector machine; CT, computed tomography; MRI, magnetic resonance
Table 4. Best prediction performance in other studies
IVa
III
II
Ib
Ia
Caudal edge of cricoid cartilage
hyoid bone
Caudal edge of the body of the
process of C1
Caudal edge of the lateral
m.
gland; anteriorly, mylo-hyoid
Cranial edge of submandibular
Mylo-hyoid m.
Cranial
Caudal
manubrium
2 cm cranial to sternal
Caudal edge of cricoid cartilage
hyoid bone
Caudal edge of the body of the
caudal)/platysma m.
(whichever is more
edge of submandibular gland
mandible; alternatively caudal
hyoid bone and caudal edge of
Plane through caudal edge of
anterior belly of digastric mm)
Platysma m. (caudal edge of the
Table 1. Node levels Ia, Ib, II, III, and IVa.
(cranially)/body of
sternocleidomastoid m.
Anterior edge of
thyro-hyoid m.
m./posterior third of
sternocleidomastoid
Anterior edge of
digastric m.
edge of posterior belly of
submandibular gland/posterior
Posterior edge of the
Symphysis menti
Symphysis menti
Anterior
(cranially)/scalenius mm.
sternocleidomastoid m.
Posterior edge of
sternocleidomastoid m.
Posterior edge of
sternocleidomastoid m.
Posterior edge of
digastric m. (cranially)
(caudally)/posterior belly of
submandibular gland
Posterior edge of
m.
Body of hyoid bone/mylo-hyoid
Posterior
(cranially)/lateral edge of
sternocleidomastoid m.
Deep (medial) surface of
sternocleidomastoid m.
Deep (medial) surface of
digastric m.
gland/posterior belly of
m./platysma m./parotid
sternocleidomastoid
Deep (medial) surface of
(posteriorly)
(caudal)/medial pterygoid m.
edge/platysma m.
mandible down to caudal
Medial aspect (innerside) of
digastric m.
Medial edge of ant. belly of
Lateral
gland/scalenius mm.
artery/lateral edge of thyroid
Medial edge of common carotid
artery/scalenius mm.
Medial edge of common carotid
artery/scalenius m.
Medial edge of internal carotid
belly of digastric m. (cranially)
digastric m. (caudally)/posterior
Lateral edge of ant. belly of
NA
Medial
NA, not available.
(caudally)
sternocleidomastoid m.
(caudally)
(caudally)
sternocleidomastoid m.
(caudally)
sternocleidomastoid m.
(cranially)/medial edge of
...
論文の公開元へ
