CT evaluation of extranodal extension of cervical lymph node metastases in patients with oral squamous cell carcinoma using deep learning classification
To clarify CT diagnostic performance in extranodal extension of cervical lymph node metastases using deep learning classification.
Seven-hundred and three CT images (178 with and 525 without extranodal extension) in 51 patients with cervical lymph node metastases from oral squamous cell carcinoma were enrolled in this study. CT images were cropped to an arbitrary size to include lymph nodes and surrounding tissues. All images were automatically divided into two datasets, assigning 80% as the training dataset and 20% as the testing dataset. The automated selection was repeated five times. Each training dataset was imported to a deep learning training system “DIGITS”. Five learning models were created after 300 epochs of the learning process using a neural network “AlexNet”. Each testing dataset was applied to each created learning model and resulting five performances were averaged as estimated diagnostic performances. A radiologist measured the minor axis and three radiologists evaluated central necrosis and irregular borders of each lymph node, and the diagnostic performances were obtained.
The deep learning accuracy of extranodal extension was 84.0%. The radiologists’ accuracies based on minor axis ≥ 11 mm, central necrosis, and irregular borders were 55.7%, 51.1% and 62.6%, respectively.
The deep learning diagnostic performance in extranodal extension was significantly higher than that of radiologists. This method is expected to improve diagnostic accuracy by further study with increasing the number of patients.
KeywordsDeep learning classification Extranodal extension Cervical lymph node metastasis Oral squamous cell carcinoma Computed tomography
Compliance with ethical standards
Conflict of interest
Yoshiko Ariji, Yoshihiko Sugita, Toru Nagao, Atsushi Nakayama, Motoki Fukuda, Yoshitaka Kise, Michihito Nozawa, Masako Nishiyama, Akitoshi Katumata and Eiichiro Ariji declare that they have no conflict of interest.
Human rights statements and informed consent
All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1964 and later versions.
Animal rights statement
This article does not contain any studies with animal subjects performed by any of the authors.
- 2.O'Sullivan B. Head and neck tumours. In: Brierley J, Gospodarowicz MK, Wittekind C, editors. UICC TNM classification of malignant tumours. 8th ed. Chichester: Wiley; 2017. p. 17–54.Google Scholar
- 13.Simonyan K, Zisserman A. Verd deep convolutional networks for large-scale image recognition. 2014. https://arxiv.org/abs/1409.1556
- 16.Ariji Y, Fukuda M, Kise Y, Nozawa M, Yanashita Y, Fujita H, et al. Contrast-enhanced computed tomography image assessment of cervical lymph node metastasis in patients with oral cancer by using a deep learning system of artificial intelligence. Oral Surg Oral Med Oral Pathol Oral Radiol. 2019 (in press). https://doi.org/10.1016/j.oooo.2018.10.002.Google Scholar
- 19.Bernier J, Cooper JS, Pajak TF, van Glabbeke M, Bourhis J, Forastiere A, et al. Defining risk levels in locally advanced head and neck cancers: a comparative analysis of concurrent postoperative radiation plus chemotherapy trials of the EORTC (#22931) and RTOG (# 9501). Head Neck. 2005;27:843–50.CrossRefGoogle Scholar
- 21.Prabhu RS, Magliocca KR, Hanasoge S, Aiken AH, Hudgins PA, Hall WA, et al. Accuracy of computed tomography for predicting pathologic nodal extracapsular extension in patients with head-and-neck cancer undergoing initial surgical resection. Int J Radiat Oncol Biol Phys. 2014;88:122–9.CrossRefGoogle Scholar