Quantitative Ultrasound for Periodontal Soft Tissue Characterization

TL;DR

This study explores the use of Quantitative Ultrasound (QUS) to noninvasively characterize periodontal soft tissues, demonstrating significant differences between tissue types and high classification accuracy, thus offering a promising diagnostic tool.

Contribution

It is among the first investigations applying QUS to periodontal tissue characterization, providing initial in vivo evidence and demonstrating high segmentation accuracy.

Findings

Significant differences in QUS parameters between gingiva and alveolar mucosa.

QUS parameters correlated with histological tissue density.

Achieved over 90% accuracy in tissue classification.

Abstract

Periodontal diseases affect 45.9\% of adults aged 30 or older in the United States. Current diagnostic methods for clinical assessment of these diseases are visual examination and bleeding on probing that are subjective, qualitative, and/or invasive. Thus, there is a critical need for research on noninvasive modalities for periodontal tissue characterization. Quantitative Ultrasound (QUS) has shown promising results in noninvasive characterization of various soft tissues; however, it has not been used in periodontics. This study is among initial investigations into the application of QUS for periodontal tissue characterization in the literature. Here, QUS analysis of oral soft tissues (alveolar mucosa and gingiva) is performed in an in vivo animal study including 10 swine. US scanning was performed at the first molar of all four oral quadrants, resulting in a total of 40 scans. We investigated first order speckle statistics of oral tissues by using the two-parameter Burr (power-law b and scale factor l) and Nakagami models (shape factor m and scale factor $\alpha$). Parametric imaging of these parameters was created using a sliding kernel method sweeping regions of interest with a kernel size of 10 wavelengths from a phantom study. Results showed that the difference between gingiva and alveolar mucosa were statistically significant using Burr and Nakagami parameters ($p-value<0.0001$). The Burr b and Nakagami m were higher in gingiva while the Burr l and Nakagami {$\alpha$} were higher in alveolar mucosa. Findings from QUS analyses agreed with observation from histology that showed denser stains for gingiva. Linear classifications of these tissues using 2D parameter spaces of the Burr and Nakagami models resulted in a segmentation accuracy of 93.51\% and 90.91\%, respectively. We propose that QUS holds promising potentials as an augmented tool for disease diagnosis in periodontology.