Automated measurement of inter-arytenoid distance on 4D laryngeal CT: A validation study

Abstract

Changes to the voice are prevalent and occur early in Parkinson's disease. Correlates of these voice changes on four-dimensional laryngeal computed-tomography imaging, such as the inter-arytenoid distance, are promising biomarkers of the disease's presence and severity. However, manual measurement of the inter-arytenoid distance is a laborious process, limiting its feasibility in large-scale research and clinical settings. Automated methods of measurement provide a solution. Here, we present a machine-learning module which determines the inter-arytenoid distance in an automated manner. We obtained automated inter-arytenoid distance readings on imaging from participants with Parkinson's disease as well as healthy controls, and then validated these against manually derived estimates. On a modified Bland-Altman analysis, we found a mean bias of 1.52 mm (95% limits of agreement -1.7 to 4.7 mm) between the automated and manual techniques, which improves to a mean bias of 0.52 mm (95% limits of agreement -1.9 to 2.9 mm) when variability due to differences in slice selection between the automated and manual methods are removed. Our results demonstrate that estimates of the inter-arytenoid distance with our automated machine-learning module are accurate, and represents a promising tool to be utilized in future work studying the laryngeal changes in Parkinson's disease.

Fig 1. Illustrative diagrams comparing the image annotation techniques.

1A, axial slice through the larynx during voice rest, with the left and right arytenoid cartilages (Ar) marked in blue. 1B, fiducial markers manually placed at the anteromedial aspect of each arytenoid cartilage, with the corresponding IAD_M marked. 1C, bounding boxes marked in red around each arytenoid cartilage by the automated machine learning algorithm, with the corresponding IAD_A marked.

Fig 2. Scatter plots of the paired estimates.

2A compares the paired estimates of the IAD_M with IAD_A, while 2B compares the paired estimates of IAD_M with IAD_S. Regression lines are shown as solid black lines with their respective equations provided. The grey dashed lines represent the lines of equivalence.

Fig 3. Bland-Altman plots.

3A compares the IAD_M against the IAD_A, while 3B compares IAD_M to the IAD_S. The mean bias (solid line), 95% upper and lower limits of agreement (dotted lines) are shown.