Deep Learning and Harmonization of Multi-Institutional Data for Automated Gross Tumor and Nodal Segmentation for Oropharyngeal Cancer

Abstract

considered (P < 0.01). When evaluating OARs individually, MC showed significantly higher ratings for brainstem, esophagus, larynx, eyes, optic nerves, while lips, whereas parotids, acoustics, and lenses were indistinguishable. Conclusion: Simple 3D architectures consistently outcompete more complex networks by quantitative measures. Qualitative assessment for clinical acceptability may not agree with quantitative performance, especially when the entire range of OARs is evaluated.

Screening for extranodal extension in HPV-associated oropharyngeal carcinoma: evaluation of a CT-based deep learning algorithm in patient data from a multicentre, randomised de-escalation trial

“…There is also the possibility that some node labels were misidentified and that pathological confirmation could be subject to interobserver variability, though the granular reporting and centralised pathological review of E3311 probably minimised these risks. Patient-level ENE prediction would promote translation to clinical use without reliance on manual segmentation, and work is ongoing to develop autosegmentation tools to facilitate this, 48 , 49 although there are inherent benefits of a node-based model. Above all, node-level prediction greatly denoises the imaging framework space and allows the algorithm to focus on the region where ENE is present.…”