Pretreatment Identification of Head and Neck Cancer Nodal Metastasis and Extranodal Extension Using Deep Learning Neural Networks

• Published in: Scientific reports Vol. 8; no. 1; pp. 14036 - 11
• Main Authors: Kann, Benjamin H., Aneja, Sanjay, Loganadane, Gokoulakrichenane V., Kelly, Jacqueline R., Smith, Stephen M., Decker, Roy H., Yu, James B., Park, Henry S., Yarbrough, Wendell G., Malhotra, Ajay, Burtness, Barbara A., Husain, Zain A.
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 19.09.2018; Nature Publishing Group
• Subjects: 59; 631/114/1564; 692/4028/67/1536; 692/4028/67/2321; Cancer; Decision making; Head & neck cancer; Humanities and Social Sciences; Lymph nodes; Management; Metastases; Metastasis; multidisciplinary; Neural networks; Pathology; Science; Science (multidisciplinary); Deep Learning Neural Network (DLNN); Head And Neck Squamous Cell Carcinoma (HNSCC); DLNN Model; Extranodal Extension (ENE); Clinical Decision-making Tool
• ISSN: 2045-2322; 2045-2322
• DOI: 10.1038/s41598-018-32441-y

Identification of nodal metastasis and tumor extranodal extension (ENE) is crucial for head and neck cancer management, but currently only can be diagnosed via postoperative pathology. Pretreatment, radiographic identification of ENE, in particular, has proven extremely difficult for clinicians, but would be greatly influential in guiding patient management. Here, we show that a deep learning convolutional neural network can be trained to identify nodal metastasis and ENE with excellent performance that surpasses what human clinicians have historically achieved. We trained a 3-dimensional convolutional neural network using a dataset of 2,875 CT-segmented lymph node samples with correlating pathology labels, cross-validated and fine-tuned on 124 samples, and conducted testing on a blinded test set of 131 samples. On the blinded test set, the model predicted ENE and nodal metastasis each with area under the receiver operating characteristic curve (AUC) of 0.91 (95%CI: 0.85–0.97). The model has the potential for use as a clinical decision-making tool to help guide head and neck cancer patient management.