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

Abstract

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.

Computed tomography performance in predicting extranodal extension in HPV‐positive oropharynx cancer

“…As such, we cannot exclude the possibility that other as‐yet‐undefined imaging features may show stronger predictive value for ENE. In this regard, a machine learning approach has been reported to predict ENE with AUC of 0.91 (95% CI 0.85–0.97) . In contrast to other studies evaluating radiologic ENE, our data‐reporting sheet did not include a variable to indicate the radiologist's overall impression for the presence of ENE .…”

Predictive value of computed tomography in identifying extracapsular spread of cervical lymph node metastases in p16 positive oropharyngeal squamous cell carcinoma

“…Multiple studies have investigated the accuracy of CT in identifying ECS, (Table ) however, only a limited number of studies have focused on ECS in P16 + OPSCC which has implications for accuracy due to the differing clinical and radiological features, and for how this knowledge is applied to clinical practice as other cancers, such as cutaneous SCC, are readily amenable to resection. Geltzeiler et al .…”

Significance of Extranodal Extension in Surgically Treated HPV-Positive Oropharyngeal Carcinomas

“…Only recently Kann et al published their study on the diagnosis of lymph node metastases and ENE in HNSCC by means of pretreatment CT images and three-dimensional deep learning neural networks (22). In this study they trained the neural network using a data set of 2,875 CT-segmented lymph node specimens and achieved diagnostic results which exceeded those of human clinicians (22). The area under the receiver operating characteristics curve for diagnosing ENE and lymph node metastases was 0.91 (95% CI = 0.85-0.97)-ENE of lymph node metastases could be predicted with a sensitivity of 88% and specificity of 85% (22).…”