Prospectively-validated deep learning model for segmenting swallowing and chewing structures in CT

Abstract

ObjectiveDelineating swallowing and chewing structures aids in radiotherapy (RT) treatment planning to limit dysphagia, trismus, and speech dysfunction. We aim to develop an accurate and efficient method to automate this process.ApproachCT scans of 242 head and neck (H&N) cancer patients acquired from 2004-2009 at our institution were used to develop auto-segmentation models for the masseters, medial pterygoids, larynx, and pharyngeal constrictor muscle using DeepLabV3+. A cascaded architecture was used, wherein models were trained sequentially to spatially constrain each structure group based on prior segmentations. Additionally, an ensemble of models, combining contextual information from axial, coronal, and sagittal views was used to improve segmentation accuracy. Prospective evaluation was conducted by measuring the amount of manual editing required in 91 H&N CT scans acquired February-May 2021.Main resultsMedians and inter-quartile ranges of Dice Similarity Coefficients (DSC) computed on the retrospective testing set (N=24) were 0.87 (0.85-0.89) for the masseters, 0.80 (0.79- 0.81) for the medial pterygoids, 0.81 (0.79-0.84) for the larynx, and 0.69 (0.67-0.71) for the constrictor. Auto-segmentations, when compared to inter-observer variability in 10 randomly selected scans, showed better agreement (DSC) with each observer as compared to inter-observer DSC. Prospective analysis showed most manual modifications needed for clinical use were minor, suggesting auto-contouring could increase clinical efficiency. Trained segmentation models are available for research use upon request via https://github.com/cerr/CERR/wiki/Auto-Segmentation-models.SignificanceWe developed deep learning-based auto-segmentation models for swallowing and chewing structures in CT and demonstrated its potential for use in treatment planning to limit complications post-RT. To the best of our knowledge, this is the only prospectively-validated deep learning-based model for segmenting chewing and swallowing structures in CT. Additionally, the segmentation models have been made open-source to facilitate reproducibility and multi-institutional research.

Dosimetric impact of adaptive proton therapy in head and neck cancer – A review

“…Besides, automation of the workflow can make the application of APT more efficient. Several studies are ongoing in the field of auto-segmentation of the targets [31] , [32] and OARs [33] , [34] using artificial intelligence. This can make the delineation faster and more consistent.…”

Advances in and applications of imaging and radiomics in head and neck cancer survivorship

“…Some steps to alleviate the burden of the latter have been achieved already (albeit still in infancy), with early work into deep learning models and convolutional neural networks for autosegmentation. Iyer et al [47 ▪▪ ] is the first known study to develop a prospectively-validated deep learning-based model for delineating a selection of swallowing muscles on CT. Other groups are currently exploring the adaptation of MRI-based deep learning models [48]. Such advances have significant potential to improve the clinical utility, improving processing times to measure muscle composition from hours to seconds without sacrificing accuracy or consistency.…”

“…Such advances have significant potential to improve the clinical utility, improving processing times to measure muscle composition from hours to seconds without sacrificing accuracy or consistency. The advantages of these models can be further realized when they are made available on open-source platforms, to facilitate reproducibility, continued refinement of model accuracy, and multi-institutional research [47 ▪▪ ].…”

Machine learning in the evaluation of voice and swallowing in the head and neck cancer patient

“…Mayo et al [44] combined big data analytics with ML to identify structure-dose-volume histogram metrics and clinically actionable dose thresholds most strongly associated with worsening dysphagia in HNC patients. Iyer et al [45] created and prospectively validated auto-segmentation models using DL for delineating swallowing structures in CT images, which is central in RT treatment planning to limit dysphagia, trismus, and speech dysfunction.…”