No abstract

Parkinson's disease is a neurodegenerative disorder that results in progressive degeneration of nerve cells. It is generally associated with the deficiency of dopamine, a neurotransmitter involved in motor control of humans and thus affects the motor system. This results in abnormal vocal fold movements in majority of the Parkinson's patients. Analysis of vocal fold abnormalities may provide useful information to assess the progress of Parkinson's disease. This is accomplished by measuring the distance between the arytenoid cartilages during phonation. In order to automate this process of identifying arytenoid cartilages from CT images, in this work, a rule-based approach is proposed to detect the arytenoid cartilage feature points on either side of the airway. The proposed technique detects feature points by localizing the anterior commissure and analyzing airway boundary pixels to select the optimal feature point based on detected pixels. The proposed approach achieved 83.33% accuracy in estimating clinically-relevant feature points, making the approach suitable for automated feature point detection. To the best of our knowledge, this is the first such approach to detect arytenoid cartilage feature points using laryngeal 3D CT images.

No abstract

Contrastive Learning and Masked Image Modelling have demonstrated exceptional performance on self-supervised representation learning, where Momentum Contrast (i.e., MoCo) and Masked AutoEncoder (i.e., MAE) are the stateof-the-art, respectively. In this work, we propose MOMA to distill from pre-trained MOCo and MAE in a self-supervised manner to collaborate the knowledge from both paradigms. During the distillation, the teacher and the student are fed with original inputs and masked inputs, respectively. The learning is enabled by aligning the normalized representations from the teacher and the projected representations from the student. This simple design leads to efficient computation with extremely high mask ratio and dramatically reduced training epochs, and does not require extra considerations on the distillation target. The experiments show MOMA delivers compact student models with comparable performance to existing state-of-the-art methods, combining the power of both self-supervised learning paradigms. It presents competitive results against different benchmarks in computer vision. We hope our method provides an insight on transferring and adapting the knowledge from large-scale pre-trained models in a computationally efficient way.Recent studies (Chung et al., 2021) (Mishra et al., 2022 attempt to combine the power of contrastive learning and masked modelling, yielding promising results. They suggest that both paradigms are complementary with each other and can deliver stronger representations when they are combined into a unified framework. Furthermore, integrating two paradigms into one framework introduces higher computational cost, which requires extensive resources (e.g., hundreds of GPU hours, enormous memory capacity, and excessive storage requirements). It is also not energy-efficient to training different frameworks from the scratch as they

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.