Statistical and Texture Descriptors of Symptomatic Plantar Fasciitis Using Ultrasound Shear Wave Elastography

Abstract

Plantar fasciitis (PFis) is a common cause of heel pain. This study aims to assess the plantar fascia (PF) quantitatively by using feature descriptors and seek valuable imaging biomarkers that can reliably diagnose PFis. A total of 63 participants underwent B-mode and longitudinal shear wave elastography (SWE) on unilateral plantar fasciae. To characterize the statistical and spatial texture features of the PF, ten statistical descriptors of the shear modulus in the standardized region of interest in PF and twenty texture descriptors in the SWE measurement window (in both horizontal and vertical directions) are proposed. Four statistical quantities (mode, avg, med, qG) and four texture descriptors (autoc, sosvh, savgh svarh) showed potential for diagnosing PFis, based on significant differences between the PFis and the healthy groups. Receiver operating characteristic (ROC) curve analysis revealed that the statistical descriptors have area under the curve (AUC) of approximately 0.9 (likelihood ratio >6.798) and the texture descriptors have AUC of approximately 0.85 (likelihood ratio >3.195). Combinations of statistical and texture descriptors can achieve higher AUCs0.968. In addition, these descriptors were related to the clinical indices (body mass index and visual analogue scale) with Spearman's correlation coefficient of r=-0.5-0.4(p < 0.05). The proposed statistical and texture descriptors showed valuable potential if applied to clinical shear wave elastography for the diagnosis of PFis. This work lays the foundation of using ultrasound shear wave image features for describing symptomatic PFis.