Auto-Detection and Segmentation of Involved Lymph Nodes in HPV-Associated Oropharyngeal Cancer Using a Convolutional Deep Learning Neural Network

Abstract

Purpose: Segmentation of involved lymph nodes on head and neck computed tomography (HN-CT) scans is necessary for the radiotherapy treatment planning of human papilloma virus (HPV) associated oropharynx cancers (OPC). We aimed to train a deep learning convolutional neural network (DL-CNN) to identify and segment involved lymph nodes on contrast-enhanced HN-CT scans. Methods: 90 patients who underwent levels II-IV neck dissection for newly diagnosed, clinically node-positive, HPV-OPC were identified. Ground-truth segmentation of all radiographically and pathologically involved nodes was manually performed on pre-surgical HN-CT scans, which were randomly divided into training/validation dataset (n=70) and testing dataset (n=20). A 5-fold cross validation was used to train 5 separate DL-CNN sub-models based on a residual U-net architecture. Validation and testing segmentation masks were compared to ground-truth segmentation masks using overlap-based, volume-based, and distance-based metrics. A lymph auto-detection model was developed by thresholding segmentation model outputs, and 20 node-negative HN-CT scans were added to the test set to further evaluate auto-detection capabilities. Model discrimination of lymph node positive and negative HN-CT scans was 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 DSC > 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 external validation using a larger dataset, are necessary to clarify the role of the DL-CNN in the routine radiation oncology treatment planning workflow.

Application of simultaneous uncertainty quantification for image segmentation with probabilistic deep learning: Performance benchmarking of oropharyngeal cancer target delineation as a use-case

“…Deep learning (DL) has increasingly been used in the OPC RT space for automatically segmenting organs at risk 7,8 and target structures [9][10][11][12] . Impressively even for GTVp segmentation, several DL approaches have boasted exceptionally high performance in terms of volumetric and surface-level agreement with the ground-truth segmentations 13 .…”

Large-scale crowdsourced radiotherapy segmentations across a variety of cancer anatomic sites: Interobserver expert/non-expert and multi-observer composite tumor and normal tissue delineation annotations from a prospective educational challenge

“…For all file conversion processes, Python v. Consensus segmentation generation. In addition to ground-truth expert and non-expert segmentations for all ROIs, we also generated consensus segmentations using the simultaneous truth and performance level estimation (STAPLE) method, a commonly used probabilistic approach for combining multiple segmentations [25][26][27][28] . Briefly, the STAPLE method uses an iterative expectation-maximization algorithm to compute a probabilistic estimate of the "true" segmentation by deducing an optimal combination of the input segmentations and incorporating a prior model for the spatial distribution of segmentations as well as implementing spatial homogeneity constraints 29 .…”

Deep-Learning-Based Generation of Synthetic High-Resolution MRI from Low-Resolution MRI for Use in Head and Neck Cancer Adaptive Radiotherapy

“…Deep learning (DL) has found wide success in a variety of domains for RT-related medical imaging applications such as target and OAR segmentation (6)(7)(8)(9)(10)(11) and outcome prediction (12,13). One less routinely studied domain is synthetic image generation, i.e., mapping an input image to an output image.…”