Prediction model for cervical lymph node metastasis in human papillomavirus-related oropharyngeal squamous cell carcinomas

Assessment of Radiologic Extranodal Extension Using Combinatorial Analysis of Nodal Margin Breakdown and Metastatic Burden in Oropharyngeal Cancer

The importance of risk stratification in the management of oropharyngeal squamous cell carcinoma (OPSCC) is becoming increasingly obvious with the growing evidence of its variable prognosis. We identified and evaluated imaging characteristics predictive of extranodal extension (ENE) in OPSCC. Preoperative computed tomography and histopathologic results of 108 OPSCC patients who underwent neck dissection as primary treatment were analyzed. Imaging characteristics were reassessed for factors associated with nodal margin breakdown and metastatic burden. Moreover, the predictability of pathological ENE (pENE) was analyzed. Univariate and multivariate binomial logistic regression analyses were performed to examine the predictive power of ENE-related radiologic features. Imaging-based characteristics showed variable degrees of association with pENE. Factors associated with nodal margin breakdown (indistinct capsular contour, irregular margin, and perinodal fat stranding) and factors associated with nodal burden (nodal matting, lower neck metastasis, and presence of >4 lymph node metastases) were significantly predictive of ENE (odds ratio (OR) = 11.170 and 12.121, respectively). The combined utilization of the nodal margin and burden factors further increased the predictive ability (OR = 14.710). Factors associated with nodal margin breakdown and nodal burden were associated with pENE, demonstrating the use of combinatorial analysis for more accurate ENE prediction.

Auto-detection and segmentation of involved lymph nodes in HPV-associated oropharyngeal cancer using a convolutional deep learning neural network

•

Patients diagnosed with HPV-associated OPC are often found to have clinical evidence of lymph node involvement at the time of diagnosis.

•

Manual segmentation of radiographically involved lymph nodes is an integral part of treatment planning for those patients dispositioned to definitive radiotherapy.

•

We present a deep learning, convolutional neural network that can be used to automate the process of lymph node detection and segmentation and with potential to increase the efficiency to the radiotherapy workflow.

Purpose

Segmentation of involved lymph nodes on head and neck computed tomography (HN-CT) scans is necessary for the radiotherapy planning of early-stage human papilloma virus (HPV) associated oropharynx cancers (OPC). We aimed to train a deep learning convolutional neural network (DL-CNN) to segment involved lymph nodes on HN-CT scans.

Methods

Ground-truth segmentation of involved nodes was performed on pre-surgical HN-CT scans for 90 patients who underwent levels II-IV neck dissection for node-positive HPV-OPC (training/validation [n = 70] and testing [n = 20]). A 5-fold cross validation approach was used to train 5 DL-CNN sub-models based on a residual U-net architecture. Validation and testing segmentation masks were compared to ground-truth masks using predetermined metrics. A lymph auto-detection model to discriminate between “node-positive” and “node-negative” HN-CT scans was developed by thresholding segmentation model outputs and evaluated using the area under the receiver operating characteristic curve (AUC).

Results

In the DL-CNN validation phase, all sub-models yielded segmentation masks with median Dice ≥ 0.90 and median volume similarity score of ≥ 0.95. In the testing phase, the DL-CNN produced consensus segmentation masks with median Dice of 0.92 (IQR, 0.89–0.95), median volume similarity of 0.97 (IQR, 0.94–0.99), and median Hausdorff distance of 4.52 mm (IQR, 1.22–8.38). The detection model achieved an AUC of 0.98.

Conclusion

The results from this single-institution study demonstrate the successful automation of lymph node segmentation for patients with HPV-OPC using a DL-CNN. Future studies, including validation with an external dataset, are necessary to clarify its role in the larger radiation oncology treatment planning workflow.